LB 
 
 UC-NRLF 
 
 $B 13 Mil 
 
 sychological Examinations of 
 College Students 
 
 BY 
 
 F. EDITH CAROTH1 
 
 Submitted In partial fulfillment of the requirements 
 
 for the Degree of Doctor of Philosophy^ in the Faculty of 
 
 Philosophy > Columbia University 
 
 NEW YORK 
 1922 
 
EXCHANGE 
 
Psychological Examinations of 
 College Students 
 
 I 
 
 BY 
 
 F. EDITH [CAROTHERS 
 
 Submitted in partial fulfillment of the requirements 
 
 for the Degree of Doctor of Philosophy, in the Faculty of 
 
 Philosophy, Columbia University 
 
 NEW YORK 
 1922 
 
ACKNOWLEDGMENT 
 
 The writer is indebted to Professor H. L. Hollingworth for the 
 original plan of this investigation and the opportunity to conduct 
 the tests in his laboratory; for his constant interest and his valuable 
 suggestions in the preparation of the results. For helpful criticism 
 regarding the arrangement and statistical treatment of the data 
 she is also indebted to Professor R. S. Woodworth, Professor A. T. 
 Poffenberger, and Professor J. McKeen Cattell. 
 
\ 
 
SECTION I 
 HISTORY OF FRESHMAN TESTS 
 
 The recommendation in 1882 by Galton 1 of the establishment 
 of anthropometric and medico-metric laboratories for the examina- 
 tion of individuals represents the first definite recognition of the 
 need of examining individuals in order to give them vocational 
 guidance. Galton saw the importance both to science and to 
 individuals of collecting complete life-histories of people which 
 should include photographs, anthropometric measurements, and 
 medical facts. To meet this need he established his now famous 
 laboratory in the South Kensington Museum, London. There, by 
 payment of a small fee, individuals could go and have certain 
 physical measurements made and undergo tests for keenness of 
 vision and hearing, dynamometer pressure, reaction time, etc. 
 
 Several years later, at the World's Columbia Exposition in i893, 2 
 Professor Joseph Jastrow arranged a laboratory devoted to tests 
 of a strictly psychological nature. Prior to Jastrow's work, however, 
 Cattell proposed 3 and tried out a series of ten mental tests and 
 measurements on students in the psychological laboratory of the 
 University of Pennsylvania. In devising his series of tests Cattell 
 followed Galton in combining physical measurements with psy- 
 chophysical and strictly mental tests. He went a step farther, 
 however, by emphasizing the necessity of standardizing methods of 
 procedure in administering tests so that results secured by different 
 experimenters might be comparable. In addition to the Pennsyl- 
 vania students, tests were also given to the students of Cambridge 
 University and Bryn Mawr College. 
 
 Gal ton's work stimulated other investigators to devise tests for 
 measuring the capacities of individuals. Of particular interest is 
 the list of ten fundamental traits or properties proposed by Kraepe- 
 lin 4 as the basic factors to be considered in examining both normal 
 individuals and the "mentally sick." These so-called fundamental 
 dispositions include: the mental capacity to do work, the ability to 
 
 1 Fortnightly Review, 1882, p. 332. 
 
 8 Cattell and Farrand, L. Physical and Mental Measurements of the Students of Columbia 
 University. 
 
 "Mental Tests and Measurements," J. McK. Cattell with appendix by Francis Galton, Mind, 
 1890. 
 
 4 Der Psychologische Versuch in der Psychiatric; Emil Kraepelin, Psychologische Arbeiten, 
 1895. 
 
2 Psychological Examinations of College Students 
 
 be influenced by practice, strength of practice or general memory, 
 special memory ability, susceptibility, fatigability, the ability to 
 recuperate, the depth of sleep, the intensity of distraction and 
 adaptability. To each one of these fundamental traits Kraepelin 
 arbitrarily assigned a certain test, assuming that excellence of 
 performance in the assigned test, say adding, would indicate excel- 
 lence in the corresponding quality, say the capacity to do work. 
 Although his assumption, without statistical proof, that certain 
 tests would measure certain functions rendered his results inac- 
 curate, from the modern standpoint, his work is interesting in that 
 it is representative of a distinct stage in the use of tests for diag- 
 nostic purposes. 
 
 With the accumulation of data and the gradually increasing 
 clearness of conception of the meaning of tests, methods of admin- 
 istering them were revised. In i896 5 appeared the first report of 
 the results of mental and physical tests made on freshmen only. 
 It concerned the work done by Professor Cattell and Dr. Farrand 
 on one hundred Columbia University students in 1894-5 and 1895-6. 
 At this time there was conceived the plan of testing Columbia 
 students during their freshman and senior years. Their tests 
 comprised ten records and twenty-six measurements. Such physical 
 measurements were taken as the color of hair and eyes, height and 
 weight, breathing capacity, sensation areas, and strength of right 
 and left hands. Other measures were of a sensory character, while 
 certain simple tests of a mental character were taken, such as the 
 rate of perception and the perception of space and time. In addi- 
 tion, a personal record-blank was filled out by the student and a 
 record of the impressions made upon him by the subject was filled 
 in by the experimenter both before and after testing. The tests 
 were given individually, the investigators and several assistants 
 acting as experimenters, and required from forty minutes to one 
 hour for their completion. The underlying purpose in giving these 
 tests is clearly stated in this statement by Cattell and Farrand: 6 
 
 "When used with freshmen on entering college the record is of interest to the 
 man and may be of real value to him. It is well for him to know how his physical 
 development, his senses, his movements, and his mental processes compare with 
 those of his fellows. He may be able to correct defects and develop aptitudes. 
 Then when the tests are repeated later in the college course and in subsequent 
 life the record of progress or regression may prove of substantial importance to 
 the individual." 
 
 5 Cattell, J. McK., and Farrand, L. Physical and Mental Measurements of the Students of 
 Columbia University, Psychological Review, 1806, III, 618-647. 
 
 6 Above reference. 
 
History of Freshman Tests 3 
 
 These Columbia freshman tests continued to be given each year 
 under Professor Cattell's direction. In 1901 7 an account and 
 discussion of the results was published by Wissler. He discusses the 
 changes and additions made in the tests and considers the records 
 of 250 freshmen, a small number of seniors, and some Barnard girls. 
 The tests employed were: length and breadth of head, strength of 
 hands, fatigue, eyesight, color-vision, hearing, perception of pitch, 
 perception of weight, sensation areas, sensitiveness to pain, per- 
 ception of size, color preference, reaction time, rate of percep- 
 tion, naming colors, rate of movement, accuracy of movement, 
 perception of time, association, imagery, memory, (auditory, visual, 
 logical, and retrospective). Records of stature, weight, etc., to- 
 gether with data concerning parentage, personal habits, and health, 
 the physical measurements taken in the gymnasium, and academic 
 marks were also secured. From the similarity of the results of 
 freshmen tested each year, Wissler concluded that freshmen enter- 
 ing Columbia from year to year are a homogeneous group and 
 represent a type. His general conclusions are: 
 
 1. That the laboratory mental tests show little intercorrelation 
 in the case of college students. Correlations range from .28 
 (accuracy and speed in marking out A's), to +.39 (auditory and 
 visual memory correctly placed). 
 
 2. That the physical tests show a general tendency to correlate 
 among themselves, but only to a very slight degree with the mental 
 tests. 
 
 3. That the markings of students in college classes correlate with 
 themselves to a considerable degree. Correlations run from +.11, 
 (mathematics and logical memory) to +-75 (Latin and Greek). 
 
 These early Columbia tests and measurements were principally 
 motor and sensory in character, and the few tests that might be 
 considered to have an intellectual quality were so simple that they 
 proved of little value for determining the mental status of the college 
 freshman. They are, however, significant in that they represent the 
 first definite attempt to establish standards of performance for 
 freshmen and to show students how their standing in various tests 
 compared with the average standing of their class. 
 
 Subsequent to the establishment of the practice of testing the 
 Columbia students in their freshman and senior years, committees 
 were appointed by the American Psychological Association in 1896 
 
 7 Wissler, Clark; The Correlation of Mental and Physical Tests; Psychological Review. Mono- 
 graph Suppl., Vol. Ill, No. 1901, p. 62. 
 
4 Psychological Examinations of College Students 
 
 and 1907, respectively, to consider the possibility of accumulating 
 mental and physical statistics through cooperation on the part of 
 various psychological laboratories and to devise a standard series 
 of group and individual tests. In 1896 the committee drew up a 
 series of physical and mental tests appropriate for college students 
 tested in a psychological laboratory. 
 
 Various other proposals were made for the scientific study of the 
 college student. In 1899 President Harper of Chicago recommended 
 that special study be made of the college student's character, intel- 
 lectual capacity, and tastes, by the questionnaire method. In 1906 
 Thorndike 8 called attention to the fact that the entrance examina- 
 tions given by the College Entrance Board of the Association of 
 Colleges and Preparatory Schools of the Middle States and Mary- 
 land did not measure at all accurately the candidate's capacity 
 and emphasized the need of the scientific study of this matter. 
 Williams 9 also stressed the importance of studying the college 
 student. Like President Harper, he recommended the questionnaire 
 method for ascertaining facts concerning the student's personality, 
 and suggested the use of Whipple's information test for obtaining 
 a knowledge of the student's range of information. He also pointed 
 out the need of vocational advisors for freshmen. 
 
 Calfee 10 in 1913 reported the results of four general intelligence 
 tests on 103 freshmen (51 boys and 52 girls) of the University of 
 Texas. The tests used were card-dealing, card-sorting, alphabet- 
 sorting, the mirror test, and the spirometer test for vital capacity. 
 She finds inter-test correlations for the boys and girls combined 
 ranging all the way from +.50 to .00. The correlations between 
 the tests and college grades range from +.32 (card sorting and 
 grades) to +.16 (mirror test and grades). The correlation between 
 the lung test and grades is .11. Considering the girls' records 
 alone, the inter-test correlations range from +.45 to +.19, and 
 the correlations with college grades from +.28 to +.13, and with 
 the lung test the correlation is .00. 
 
 No further attempt to measure the performance of college fresh- 
 men in tests is reported until December, 1915, when Dr. Karl T. 
 Waugh presented a paper on "A New Mental Diagnosis of the 
 College Student" before the American Psychological Associa- 
 
 Thorndike, E. L. An Empirical Study of College Entrance Examinations. Science, N.S., 
 1906, 23, 839-845- 
 
 Williams, C. W. Scientific Study of the College Student. 
 
 Calfee, M. College Freshmen and Four General Intelligence Tests, Journ. of Educ. Psychol., 
 1913, 4, 223-231. 
 
History of Freshman Tests 5 
 
 tion. 11 In 1912 he applied seven tests 12 individually to freshmen in 
 Beloit College, and three years later, in 1915, he gave the same 
 tests to thirty-nine of the same subjects. Waugh's inter-test cor- 
 relations range from .43 to +.54, and he finds some improve- 
 ment in the tests from freshman to senior year. 
 
 During the year 1913-14 Bingham 18 gave nine tests to 200 Dart- 
 mouth freshmen, seven of them being given individually. As a 
 number of psychology students, unpracticed experimenters, assisted 
 Professor Bingham in his testing, the results of his investigation are 
 somewhat inaccurate. He gives norms for the nine tests, (median, 
 standard deviation and coefficient of variability) and the range 
 from the poorest to the best. As no correlations are reported we 
 have no information as to the relationships between the tests. 
 Bingham's chief contribution consists in his use of the method of 
 ogive percentile graphs. The data in seven of his tests are presented 
 in this form, thus serving as a scale. Given the score made by any 
 individual, the experimenter by reference to the chart can readily 
 assign him a rank among his classmates. The speed with which a 
 student may be thus assigned his relative position in any given 
 trait makes this method a most convenient one for the instructor. 14 
 
 At the University of Texas the same year Bell 15 gave nine tests w 
 to about seven hundred and fifty freshmen. Bell definitely states 
 that his aim was to devise a series of tests that would "be of assis- 
 tance to college authorities in aiding freshmen to adjust themselves 
 to their environment." The time required for testing was from 
 forty to forty-five minutes. The tests were given not individually, 
 but in groups averaging a little less than twenty each. The time- 
 limit method was used. This, together with his arbitrary method 
 of scoring the tests may account in some measure for the unsatis- 
 factory nature of his results. He weighted each test so that a perfect 
 
 11 Waugh, Dr. Karl T. A new Mental Diagnosis of the College Student. New York Times 
 Magazine, January 2, 1916. 
 
 12 Waugh's tests were: I. Concentration of attention (cancellation of A's); 2. Range of infor- 
 mation; 3. Speed of learning (substitution); 4. Quickness of association (opposites); 5. Ingenuity 
 (puzzle-box); 6. Steadiness; 7- Memory for a passage (immediately after hearing it read and after 
 an interval of two weeks). 
 
 "Bingham, W. V. Some norms of Dartmouth Freshmen; Journ. of Educ. Psychol., March, 
 1916, Vol. 7, PP. 129-142. 
 
 14 Bingham's tests were: i. Endurance of grip; 2. Tapping; 3. Memory span for auditory 
 digits; 4. Logical memory; 5. Cancellation; 6. Color Naming; 7. Logical relations; 8. Mixed 
 relations; 9. Perception of form. 
 
 Bell, J. Carleton. Mental Tests and College Freshmen; Journ. of Educ. Psychol., Sept., 1916. 
 Vol. 7, pp. 381-399. 
 
 16 Bell's Tests include: i. Cancellation of triangles; 2. Addition; 3. Association or learning 
 pairs; 4. Recognizing forms; s- Marking right statements; 6. Easy directions; 7. Hard Directions; 
 8. Alternatives; 9. Completion (using "The Strength of the Eagle" as material). 
 
6 Psychological Examinations of College Students 
 
 mark or the highest mark would approximate 100, and the other 
 marks range downward from this to zero. For example, in the 
 Triangles test there were fifty triangles to be crossed out. Each 
 one correctly crossed out counted two points and five points were 
 deducted for each error, positive or negative. For example, if a 
 student crossed out 35 triangles, omitted 3, and crossed out one 
 circle, his score was 70 minus 20 = 50. The other tests were scored 
 in similar manner. 
 
 Bell also obtained the correlations of freshmen university grades 
 with each other and of the university grades with the mental tests. 
 His conclusions are : 
 
 1. The correlations between freshmen university grades vary 
 from + -34 (mathematics history) to + .59 (English history, 
 science history) . 
 
 2. The highest correlation between class marks and test scores 
 is + .31 (English Completion). 
 
 3. Among the tests themselves the highest correlations are found 
 between the Association and Recognition tests, and between the 
 Directions, Alternatives and Completion tests. 
 
 4. There is a considerable difference in the results of the tests 
 with the best and the poorest students, but the scores are so variable 
 as to be of little value for individual diagnosis. 
 
 The investigations of Calfee, Waugh, Bingham, and Bell illustrate 
 the striking change that has taken place in the character of mental 
 tests since the early Columbia tests were first instituted. In place 
 of sensory and motor tests we now employ tests which will measure 
 diverse mental functions. Motivated by this same desire to secure 
 a group of tests for college students indicative of mental ability, 
 and correlative with college grades, Rowland and Lowden 17 began 
 to try out groupings of psychological tests in 1912-13 and carried 
 out their investigations over a period of three years. The tests 
 were conducted individually on all the students in Reed College, 
 twelve students of experimental psychology assisting in conducting 
 the tests. The first grouping of tests was tried out on 54 students 
 during 1912-13, after which the grouping was revised and given to 
 195 more subjects. No inter-test correlations are reported. The 
 highest correlation between university grades and the groupings 
 was between the grades and the letter-group g-r-s-t., cancellation, 
 opposites, logical memory, judgment (syllogism), rote memory, 
 
 17 Rowland, E. and Lowden, G. Report of Psychological Tests at Reed College. Journ. of 
 Exper. Psychol., 1916, I, 211-217. 
 
History of Freshman Tests 7 
 
 cancellation of words with a and /, (a correlation of + .37 with a 
 P.E. of .06). 
 
 Psychological tests have also been conducted for several years at 
 Vassar College. Results of tests made upon Vassar freshmen during 
 the years I9I4, 18 1915, and 1916 19 show data collected from four 
 sources, namely: I. Answers to a questionnaire calling for infor- 
 mation regarding the student's imagery, interests, language facility, 
 and habits; 2. Results of the tests; 20 3. Freshmen academic grades ; 
 4. Reports of promising students by their instructors. To deter- 
 mine roughly the correlation between academic marks and test 
 scores, the difference between the average class standing of students 
 having test scores in the first or highest quarter and the average 
 class standing of students with test scores in the last quarter was 
 found. If there was a marked difference the experimenters con- 
 cluded that a positive correlation existed. According to this rough 
 method they found a positive correlation between academic marks 
 and the tests except Hard Directions. On the whole, the results of 
 the Vassar tests appeared to indicate that ability in the tests 
 correlates well with ability in freshman studies, while inability to 
 do well in the tests is correlated with a similar inability to do well 
 in freshman studies. Moreover, students designated as "promising" 
 by their instructors tend to manifest a high grade of performance 
 in the tests. (14.5% of 317 freshmen tested in 1917 who passed 
 all the tests in the Terman Superior Adult Tests were rated by their 
 instructors as being of only average ability.) The experimenters 
 also found that the relation between success in freshman tests and 
 academic success in three years' work is less than that between 
 success in freshman tests and academic success in the freshman 
 year. Inasmuch as there were thirty different testers, each one 
 being assigned a small group of freshmen, little confidence may be 
 placed in the accuracy of the data. The tests as conducted at 
 Vassar are of value more for the opportunity they afford students 
 of psychology to acquire training in experimental methods of pro- 
 cedure than for any contribution they make to our knowledge of 
 freshman standards of performance in various tests. 
 
 "White, Sophie D.; May, Sybil; and Washburn, M. F. A study of Freshmen. Minor Studies 
 from the Psychological Laboratory of Vassar College, No. 31, Amer. Jour, of Psychol., 1917, Vol. 
 28, pp. 151-154. 
 
 "Montagne, M.; Reynolds, M. M.; and Washburn, M. F. A Further Study of Freshmen . 
 Amer. Jour, of Psychol., 1918, 29, 327-330. 
 
 20 The tests described include: Verbal memory and memory for ideas; Reading Backwards; 
 Hard Directions; Analogies; Sentence Building; Suggestibility; Free Association; Thurstone 
 Reasoning. 
 
8 Psychological Examinations of College Students 
 
 An interesting contribution in connection with the application of 
 psychological tests to college freshmen is that of Kitson 21 at the 
 University of Chicago. With the general purpose of devising a 
 "system for measuring the mental capacity of college students in 
 order to guide their college work," Kitson selected sixteen tests. 22 
 About half the tests were given by the group method. The time 
 required for testing was two and one half hours. From forty com- 
 plete records Kitson computed norms of performance in the various 
 tests. In addition, a graphic chart was arranged for each student 
 to show his standing in each test and to furnish a net score com- 
 bining his standing in all the tests. In the particular tests used, 
 Kitson found a significant positive correlation only between: 
 
 1. Memory for meaningful material seen and heard (+.54); 
 
 2. Between the first and second reproductions of this material 
 (+ .49); 3. Between the Opposites and Constant Increment tests 
 (-f .40). When correlations were computed of standings in each 
 test with standings in the net score, they were found to be some- 
 what higher. The correlation between college marks and psychologi- 
 cal tests was found to be + .44 (P.E. .09) but from forty records 
 secured from a second group of freshmen tested the correlation was 
 found to be only + .20 (P.E. .11). Kitson explains this low correla- 
 tion on the ground that many other factors besides intelligence enter 
 in to determine standing in school studies, such as the personal 
 factor of the instructor, the student's will power, social surroundings, 
 economic conditions, and physical condition. The correlation 
 between the psychological tests and intelligence as estimated by 
 the dean was -f- .57 (P.E. .05). Twenty-one of the 1915 freshmen 
 were retested in seven of the tests in their Sophomore year and 
 improvement was shown in every test except one. (Numbers heard.) 
 Comparison between the net score for freshman and sophomore 
 year shows a correlation of + .88 (P.E. .03). 
 
 Although his norms of performance in the tests and his inter- 
 test correlations are not very reliable, based as they are upon only 
 forty records, there is much to be said in favor of Kitson's general 
 method of procedure. His emphasis upon the importance of study- 
 ing the individual student in his relation to the college and his 
 
 a Kitson, H. D. The Scientific Study of the College Student. Psychol. Monog., 1917, 23 (No. 98), 
 p. 8x. 
 
 The tests employed were: Number-checking; Memory for numbers heard; Memory for objects 
 seen; Memory for logical material heard; Secondary memory for same; Immediate memory for 
 logical material, seen; Secondary memory for same; Loss in logical material, heard; Loss in logical 
 material, seen; Opposites; Constant increment; Hard directions, printed and oral; Word build- 
 ing; Sentence-building; and Business ingenuity. 
 
History of Freshman Tests 9 
 
 realization of the fact that psychological measurements, however 
 large the role they may play in determining a student's abilities 
 and aptitudes, must not be considered the sole factor in such a 
 determination, but rather should be so coordinated with measures 
 of the student from various other aspects as to lead to our fuller 
 understanding of the nature of the individual student and his 
 potentialities, signify a decided advance in the method of treating 
 the problem. The splendid cooperation of all the students and 
 his success in dealing with delinquent cases speak much for Kitson's 
 general method. 
 
 Other minor investigations have been made on freshmen with the 
 same purpose. Sunne, 23 working at Newcomb College, found a 
 low correlation between college grades and an information test 
 tried on twenty- five freshmen, and with ninety-nine freshmen who 
 were given a series of tests found correlations of tests with grades 
 ranging from o to + .25. Haggerty 24 found a correlation of a 
 quality of reading test and omnibus test with medical marks of 
 + .62 and + .60, respectively, and of the two combined of -f- .65, 
 in the case of sixty-nine candidates for medical school who had 
 already completed two years of college. 
 
 At the University of Iowa King, 28 working with a little group of 
 nineteen freshmen, found a tendency for the students with high 
 academic marks to make higher scores in the completion, logical 
 memory, and lanes test than the students with low academic marks. 
 He gives no statistical evidence in support of this statement. Later, 
 using a series of five tests with 56 freshman engineers, he obtained 
 a correlation between students' ranks in all the tests combined and 
 their academic grades of -f .27. The tests employed by King were: 
 
 1. Courtis Arithmetic, Series B, (graded for speed and accuracy); 
 
 2. Hard Opposites; 3. Recognition of Forms; 4. The Kansas Silent 
 Reading Test, (H.S. Series); and 5. "Hall Cube Test," a test of 
 visual imagination. 
 
 A little later Irving King and James M'Crory 26 followed Kitson's 
 method more definitely. In the fall of 1916 they tested 276 women 
 and 268 men freshmen in seven different tests: the Courtis Standard 
 
 Sunne, D. The Relation of Class Standing to College Tests, Journ. of Educ. Psychol., 1917, 
 8, 193-211. 
 
 * 4 Haggerty, M. E. Tests of Applicants for Admission to University of Minnesota Medical 
 School. Journ. of Educ. Psychol., 1918, 9, 278-286. 
 
 King, I. The relationship of abilities in certain mental tests to ability as estimated by teachers, 
 School & Society, 1917, 5, 204-209. 
 
 King, I. and M'Crory, J. Freshman Tests at the University of Iowa, Journ. of Educ. Psychol., 
 1918, 9, 32-46. 
 
io Psychological Examinations of College Students 
 
 Arithmetic Test, Series B; mixed relations; two tests of "opposites;" 
 a completion test used by Simpson; visualization; Whipple's infor- 
 mation test, and a logical memory test. The group method of test- 
 ing was used, the tests being given in groups of from ten to twenty- 
 five. Their rather low inter-test correlations indicate, they state, 
 that they are measuring a variety of mental functions. They find, 
 moreover, fairly good correlations between the tests and academic 
 grades (+ .14 to -f .45 in the case of the girls, and -f .21 to + .84 
 in the case of the boys). In their attempt to make practical applica- 
 tion of the tests for the diagnosis of their students in general and 
 cases of special ability and disability, as Kitson does, they have 
 been fairly successful. 
 
 At Northwestern University Uhl 27 obtained inter-test correla- 
 tions ranging from -f .18 (Trabue Completion K and Information), 
 to -|~ .42 (Trabue Completion M and Information), for a group of 
 one hundred freshmen tested in the fall of 1916. His series contained 
 only four tests: Trabue Completion K and M, a hard opposites list 
 of iwenty words, and an information test which consisted of the 
 seventy most familiar words in Whipple's list plus thirty new words. 
 Test correlations with the first semester English and Mathematics 
 grades were determined and found to range from -f- .48 (English and 
 Mathematics), to + .16 (Completion K and Mathematics). When 
 he had three mathematics instructors rate these one hundred stu- 
 dents for ability, Uhl found a correlation of -f .93 between their 
 ratings and the Mathematics grades of the students. This high 
 correlation was no doubt due to the tendency on the part of the 
 teachers to make their judgments of the students practically equiva- 
 lent to the students' course grades. The correlation between the 
 instructor's judgments and the ranks of these same students in their 
 last year of high school was + .59, and with all the tests combined 
 was + .36. Uhl thinks his tests fail to measure accurately, the 
 information test being the most unsatisfactory, and attributes his 
 low correlations to the homogeneity of his group, the relative sim- 
 plicity of the tests, and the unreliability of school marks. 
 
 Thurstone's 28 work represents a further development in the use 
 of psychological tests. At the Carnegie Institute of Technology the 
 attempt is made to use psychological tests as a criterion for admis- 
 sion. A series of six mental tests was given to 114 freshmen of the 
 Margaret Morrison Carnegie School in October, 1917. The problem 
 
 Uhl, W. L. Mentality Tests for College Freshmen, Journ. of Educ. Psychol M I9i9, io, 13-28 
 28 Thurstone, L. L. Journ of Educ. Psychol., March, 1919. 
 
History of Freshman Tests 1 1 
 
 was to determine whether they could reduce the number of students 
 who were dropped for poor scholarship or placed on probation for 
 poor scholarship by the use of the mental tests, and to determine 
 whether the mental test ratings correlated with faculty estimates 
 concerning the general ability of the students. The tests which 
 agreed well with the judgment of the faculty were retained. In 
 working up his results Thurstone used the method of critical scores. 
 After plotting scatter diagrams for each test, upper and lower critical 
 scores were determined such that every student above the upper 
 critical score is above the average in the opinion of the faculty, 
 and every student below the lower critical score is below the average 
 in the opinion of the faculty. The mental test rating was designated 
 as the medium percentile rank in all six tests plus 5 points for each 
 test in which the student is above the upper critical score, and 
 minus 5 points for each test in which he is below the lower critical 
 score. Students with a mental test rating of 10 or below were 
 reported as doubtful. 
 
 Thurstone found a correlation between instructors' estimates of 
 students' ability and the combined mental test rating of + .60. 
 From his results he concluded that: I. The mental test rating 
 would have eliminated seven of the eleven total failures at the 
 beginning of the year. 2. No average or good student would have 
 been eliminated by the mental test rating. All students who scored 
 below the lower critical mental test rating were, without exception, 
 poor students. 
 
 Moreover, all the freshmen who were rated high by the faculty 
 were above the average in the mental test rating. From all indica- 
 tions, this method is working out well at Carnegie. 
 
 The past three years have brought a further development in the 
 use of psychological tests for measuring the intelligence of college 
 freshmen. Since 1918 the Army Alpha test has been administered 
 to freshmen in several colleges with varying degrees of success. 
 Professor Stone 29 reports that its use at Dartmouth justifies the 
 recent proposal to admit students scholastically in the upper quarter 
 of their class in approved schools. Strictly speaking, the work at 
 Dartmouth should not be included in this history, since it deals 
 with the results obtained in testing all the college classes rather 
 than freshmen only. We mention it here, however, because the 
 college authorities are now devoting particular attention to admin- 
 
 Stone, Charles Leonard. "Intelligence and Scholarship;" The Dartmouth Alumni Magazine 
 March, 1920. 
 
12 Psychological Examinations of College Students 
 
 istering the test to the freshman class. During the fall of 1918 the 
 Army Alpha test was given to all the students in the Students Army 
 Training Corps which included practically the entire student body. 
 The average score in Alpha for the 677 S. A. T. C. men tested was 
 147.5. The average academic grade for the same men was 2.12, 
 using the scale D = i, C = 2, 6=3, A = 4. The correlation 
 between the academic marks and Alpha scores was + .44. There is 
 also a significant correspondence between a student's score in the 
 Alpha test and the scholarship quintile his academic record places 
 him in. Although less exact than Thurstone's method of assigning 
 individuals their relative position in a group, this method serves to 
 give a rough and quick estimate of a student's status. 
 
 Similar to this Dartmouth study is that of Walcott's 30 at Hamline 
 University. Here, too, not freshmen alone, but all students were 
 given the Alpha test in the fall of 1918. Walcott's results are based 
 on data secured from 61 men and 145 women. As in the Dartmouth 
 investigation, a far greater proportion of men and women students 
 secure a score in Alpha in the high grade intelligence group than 
 was found in any of the army camps. The median score is 129 for 
 the Hamline men and 133 for the women, with the same sharp 
 differentiation between the poor and the good groups as Stone found 
 at Dartmouth. The correlation between the results of the Alpha 
 test for the women and their first term academic grades was -f- 47, 
 slightly higher than the Dartmouth result. Although Walcott does 
 not consider the army test the best device for determining the fitness 
 of students for college work, he sees in the significant difference in 
 score between the upper and lower half of the students tested, the 
 practical use to be made of this fact in the placing of students. 
 
 Similar investigations have also been conducted by Hill, Filler, 31 
 and Hunter at the University of Illinois, Dickinson College, and 
 Southern Methodist University, respectively. At the University 
 of Illinois 3,500 students were tested in twenty-four groups in 
 March, I9I9, 32 members of the faculty acting as experimenters. 
 As at Dartmouth and Hamline, the scores of the students at each 
 of these colleges show them to be a very select group compared to 
 the army men. The median score of the freshmen in the school of 
 liberal arts and sciences at the University of Illinois is 147. At 
 
 o Walcott, G. D. "Mental Testing at Hamline University." School and Society, 1919, ro, S7-6o. 
 M Filler, M. G. A Psychological Test. School & Society, 1919, 10, 208-209. 
 Hill, D. S. Results of Intelligence Tests at the University of Illinois; School & Society, 1919, 
 9, 542-545- 
 
History of Freshman Tests 13 
 
 Southern Methodist University 88 the effort was made to secure 
 select groups of students in order to compare their scores with the 
 average score for the school. Each student was asked to name men 
 and women students whom they thought would make high scores 
 in the Alpha test. For 16 men and 8 women named by from five to 
 forty students as being able to make the highest scores, the average 
 score for the men was 154, and for the women 156, justifying the 
 judgment of the students. With a similar group of students named 
 by the faculty as being able to make the highest scores even better 
 results were obtained, the average for the men being 161 and for 
 the women 167. In selecting a group of men and women whom they 
 judged would make low scores the faculty were equally successful. 
 Both faculty and students thus showed themselves fairly good in 
 their ability to select students on the basis of intelligence, though 
 this method of selection is inferior to selection on the basis of actual 
 scores. The correlation between the Alpha scores of the women 
 students and their college grades for the fall term was + -52. No 
 correlations are given in the Illinois and Dickinson reports, which 
 are only preliminary. 
 
 The following is a comparative table showing scores obtained 
 at the University of Illinois, Dickinson College, and Southern 
 Methodist University: 
 
 Southern 
 
 University Dickinson Methodist 
 
 of Illinois College University 
 
 Total number tested 3,254 213 321 
 
 Number of freshmen 489 72 128 
 
 Lowest freshman score ..... 52 75 60 
 
 Highest freshman score 188 195 188 
 
 Median freshman score 147 141 127 
 
 Hunter explains the lower median score at Southern Methodist 
 University as due to a difference in the method of conducting the 
 test. 
 
 More fully developed than these three preliminary investigations 
 is the work being done at Brown University. 34 Colvin reports the 
 results obtained from 103 freshmen with the Alpha test and two 
 series of psychological tests, known as Brown University Series I 
 and II, which were separated by an interval of several days. Each 
 series consisted of four tests: mutilated sentences, vocabulary, 
 analogies or mixed relations, and a reasoning test. The distribution 
 
 "Hunter, H. T. Intelligence Testa at Southern Methodist University; School & Society, 
 1919. io, 437-440. 
 
 u Colvin, S. S. Psychological Tests at Brown University; School & Society, 1919, io, 27-30. 
 
14 Psychological Examinations of College Students 
 
 of scores for both Series I and Series II separately and for the 
 combined scores of Series I and II conformed closely to a normal 
 probability curve. The correlation between Brown University 
 Series I and II is + .75, and between the average of these two series 
 and the Alpha test is + .79. The correlation between the Brown 
 University tests and the average academic marks of the first and 
 second terms is + .59, and between the army test and the average 
 of the marks of the first and second terms is + -45- Practical appli- 
 cation was made of the tests to foretell a student's probable aca- 
 demic success and to aid in diagnosing cases of failure in school 
 work. Colvin found that two-thirds of 80 students reported as 
 doing unsatisfactory work in the first term had made low scores in 
 their psychological tests, while only one-sixth of the men had a 
 satisfactory grade. Most of the cases of students doing poor college 
 work who had obtained high scores in the tests were due not to 
 lack of ability, but to other reasons. So satisfactory have the tests 
 been in determining the students' mental status and helping them 
 that they are still being employed. 
 
 In a recent article in the Educational Review 35 Professor Colvin 
 compares in greater detail the scores and correlations obtained in 
 the Brown University tests and the Alpha test, and reports results 
 secured in giving the Brown tests and the Thorndike tests to 300 
 freshmen. The Brown tests require about fifty-five minutes of 
 actual working time as contrasted with about three hours required 
 by the Thorndike tests. The median score for the Brown tests is 
 62.4 with a standard deviation of 10.59, compared to the median 
 score for the Thorndike tests of 76.5 with a standard deviation of 
 14.89, the difference being due to the fact that the Brown tests 
 have a maximum score of 100, while the Thorndike tests have a 
 maximum score of about 150. The correlation between the scores 
 obtained by students in the two tests is + .816 with a P. E. of .0138, 
 but the Thorndike tests show a higher correlation with academic 
 marks (+ .53) than the Brown tests (+ 46). While the Thorndike 
 tests show a slight superiority in prognostic value, nevertheless 
 results show that men receiving scores in the lowest fifteen per- 
 centile of either the Brown or the Thorndike tests have a relatively 
 small chance of graduating from college. Colvin warns against the 
 danger of refusing men admission to college solely because of a low 
 psychological record. He advocates the conservative position of 
 
 Colvin, S. S. The Validity of Psychological Tests for College Entrance. Educational Review. 
 June, 1920. 
 
History of Freshman Tests 15 
 
 regarding the psychological record as one among many factors to 
 be considered in diagnosing cases of individual students. 
 
 At Ohio State University the Alpha test was successfully given 
 to 5,950 students October 10, 1919, in groups of one hundred to 
 two hundred and fifty. The distribution of scores for the entire 
 group conformed to the normal probability curve, the students 
 being grouped into five classes as follows: 
 
 Approximate 
 Class Score Percentage in Each Class 
 
 I. 178-212 Very superior intelligence 5 
 
 II. 155-177 Superior 20 
 
 III. 115-154 Average 50 
 
 IV. 85-114 Fair 20 
 
 V. o- 84 Poor 5 
 
 The percentage of students falling into each of these five classes 
 was then determined for the various university units separately, 
 such as the Graduate School, Commerce and Journalism, Law, 
 Medicine, Engineering, Arts Education, Agriculture, Pharmacy, 
 etc. The median, highest, and lowest scores, and the number 
 examined for each class (college year), in each college and in the 
 whole university, are reported. The highest median score, 157, 
 was obtained by the Graduate School; Arts received second place 
 with a median score of 147; Commerce and Journalism third, with 
 a median score of 146; and so on down to a median of 112, (Veterin- 
 ary Medicine group). The report gives an interesting comparison 
 of the various college groups. 
 
 The Thorndike tests, previously mentioned, are rapidly becoming 
 more widely employed for freshmen testing than the Army Alpha. 
 Jones, 86 writing in the Educational Review, clearly describes the 
 general nature of these tests. Although conceding their practical 
 value, he urges that they should be employed "not to the exclusion 
 of other measures for determining fitness, but along with them." 
 Evidence of a student's fitness to undertake college work should, in 
 Professor Jones' opinion, include his preparation for college work, 
 his character and promise, his health, and his intelligence denoted 
 by his score in the Thorndike test. In a brief report before the 
 New York Branch of the American Psychological Association this 
 year Mr. Wood stated that the purpose of the Thorndike tests 
 was fourfold: I. To select those fit for a college course; 2. To aid 
 college committees; 3. To assist the progress of schools; 4. To 
 
 M Jones, A. L. Psychological Tests for College Admission; Educational Review, 1919, $8. 
 271-278. 
 
1 6 Psychological Examinations of College Students 
 
 assist the Dean in the administration of the college. Results from 
 a large number of freshmen showed a correlation between the total 
 Thorndike score and the average college grade of + .52, and the 
 median college grade of + .54. Although no published reports of 
 results secured with the Thorndike tests have appeared, investi- 
 gators who are employing the tests find them highly satisfactory. 
 
SECTION II 
 
 STATEMENT OF THE PROBLEM WITH A LIST 
 OF THE TESTS EMPLOYED 
 
 The present investigation, begun at Barnard College in the fall 
 of 1915, about two years before the Army Alpha and the Thorndike 
 Tests were originated, was carried on during the years 1915-16, 
 1916-17, the fall of 1917, and the spring of 1919. The general 
 purpose underlying the investigation was similar to that underlying 
 the investigations of other experimenters during this period a 
 purpose which continues to motivate present studies. The aim was 
 first, to establish norms and standards of performance in mental 
 tests for Barnard freshmen, and second, to give students a clear 
 conception of their abilities and aptitudes along various lines. 
 More specifically, this investigation concerns the trial of a group 
 of tests with the object first, of determining their reliability as 
 measures; second, their correlation with freshman university grades; 
 and third, with physical records taken in the gymnasium. 
 
 In selecting the particular group of tests to be used several factors 
 contributed. Paramount in importance was the desire to select 
 a series of tests of such nature as to call into play various mental 
 functions. In addition, it was desired to secure tests which previous 
 investigators had found to have a positive correlation with such 
 factors as age, ability along some vocational line, or general intelli- 
 gence. Equally important in determining the final selection was the 
 time-limitation factor. Owing to unwillingness on the part of stu- 
 dents to act as subjects for a longer period, and to the factor of 
 fatigue which would probably influence the results of tests com- 
 pleted after that time, it was found necessary to have a series of 
 tests such as could be completed in one hour. Consideration of all 
 these factors finally lead to this selection of tests : 
 
 1. Coordination 8. Verb-object 14. Word Memory 
 
 2. Tapping 9. Mixed Relations 15. Logical Memory 
 
 3. Cancellation 10. Word Building 16. Substitution 
 
 4. Checking II. Word Naming 17. Completion 
 
 5. Color Naming 12. Knox Cube 18. Information 
 
 6. Directions 13. Digit Span 19. Vocabulary 
 
 7. Opposites 
 
SECTION III 
 METHOD AND TECHNIQUE OF THE INVESTIGATION 
 
 Shortly after the beginning of the academic year, in the fall of 
 1915, the series of tests selected according to the manner described 
 in the preceding section was submitted to a preliminary trial in 
 order to determine the best method of conducting the tests, and to 
 afford the writer practice in their administration. After determining 
 the general method of procedure, a notice was posted in the Fresh- 
 man Study of Barnard, stating that a series of psychological exam- 
 inations had been instituted for Barnard freshmen, and giving a 
 description of the nature and purpose of the tests. It was stated 
 that the time required for the examination was one hour, and an 
 accompanying schedule indicated the hours at which the test 
 might be taken. The place where the examinations were to be 
 held was also indicated, and all freshmen interested were requested 
 to sign their names on the schedule opposite the hour at which 
 they could take the test. This method of permitting the student 
 to take the test at the hour most convenient for her, rather than at 
 a time prescribed by the experimenter, seems advisable in that it 
 establishes a certain uniformity in conditions, the student usually 
 being in her best physical condition at the time of testing. In addi- 
 tion, letters were sent to individual students in the class, reminding 
 them of the examination, and an account, written by Professor 
 Hollingworth, of the widespread use of similar tests by reliable 
 business firms and their value in selecting candidates for positions 
 along various lines, appeared in the college weekly. A similar notice 
 of the tests was posted in Freshman Study in the fall of 1916, and in 
 the fall of 1917. Letters were also sent to individual students at 
 these times. 
 
 The subjects, as indicated, were Barnard students in their fresh- 
 man year. The fact that they had had no training in experimental 
 psychology, and were unfamiliar with the tests employed, made 
 them a suitable group for testing. Out of a class of about one 
 hundred and forty freshmen during 1915-16, one hundred were 
 tested. This constitutes our first group of subjects whom we will 
 designate as Group I. During the year 1916-17 (class of 1920), 
 
Method and Technique of the Investigation 19 
 
 eighty- five freshmen were tested, and in the fall of 1917 fifteen more 
 (class of 1921) were given the tests. These last two groups together 
 constitute our second group of one hundred freshmen whom we will 
 designate Group II. In addition, in order to determine the reliability 
 of the tests, the series was divided into two equivalent parts in a 
 manner to be described later. In the spring of 1919, during the 
 period extending from March 14 to May 15, forty-five freshmen 
 from the class of 1922 were tested twice on the same day, each test 
 requiring forty-five minutes of the student's time. 
 
 All the tests were given individually. This enabled the experi- 
 menter to supervise personally the performance of each subject and 
 to stop her at any indication that she did not fully understand the 
 directions given. It was likewise an important factor in contributing 
 to the standardization of the conditions of the experiment. The 
 subject was by this means freed from any feelings of irritation or 
 discouragement that might have arisen if she had taken the test 
 with a group of students whom she knew to be more rapid workers 
 than herself. In such a case the knowledge that others were accom- 
 plishing their work in a shorter period of time would operate to 
 arouse in some subjects such feelings of the futility of competing with 
 their companions that their resulting performance would have been 
 much slower than would have been the case where the tests were 
 taken under more favorable conditions. Each freshman, then, was 
 examined individually, and every effort was exercised to make the 
 conditions of the experiment as uniform as possible. The room 
 employed for the testing was one regularly used by the Department 
 of Psychology for advanced experimental work, and from the point 
 of view of light and ventilation it is well adapted for research. 
 Except during the tapping and coordination tests, the subject sat 
 at a small laboratory table, opposite the experimenter. As the room 
 was so situated as to be almost unaffected by sounds from neighbor- 
 ing rooms, and was itself kept in a quiet condition, there was nothing 
 to distract the subject's attention from her work. 
 
 As previously indicated, attempt to secure uniformity in admin- 
 istering the tests was also made. Besides giving the tests individ- 
 ually, the order in which the tests are listed was followed. In a few 
 cases circumstances rendered it necessary to deviate slightly from 
 this order, but in general it was followed rigidly. The result of the 
 preliminary trial had been to indicate the most satisfactory manner 
 in which the tests should be administered. The aim was to make 
 the directions as clear, simple, and direct as possible. As a detailed 
 
2O Psychological Examinations of College Students 
 
 account of the instructions given for each test will be considered in 
 the next section, it is only necessary to mention here that the method 
 of procedure agreed upon was carefully followed with one or two 
 exceptions where misinterpretation of the directions resulted in the 
 experimenter's repeating the instructions in a slightly different form. 
 
SECTION IV 
 
 DISCUSSION OF THE TESTS, INCLUDING MATERIALS 
 USED, METHODS OF PROCEDURE, AND RESULTS 
 
 Test No. I. Coordination 
 
 This test, popularly termed the "three-hole test" calls for both 
 speed and accuracy of movement and gives an indication of the 
 subject's motor ability and coordination. 
 
 Apparatus: An oak plate tilted at an angle of 45 degrees to the 
 base board, containing three brass-line holes arranged in the form 
 of an equilateral triangle, about 8 cm. apart. Contact of the metal 
 rod with the bottom of the hole makes an electrical connection 
 recorded by the automatic counter. Stop watch. 
 
 Instructions: "I want you to hold this (stylus) in your right hand 
 and to touch the bottom of each one of these targets as quickly as 
 possible, going around in a circle without skipping any of the holes. 
 You see every time you do so, the contact is registered on the 
 electric counter. I want to see how many contacts you can make 
 in one minute. You start then when I say, 'Go' and stop when I 
 say, 'Stop.'" 
 
 Method of scoring : The score represents the number of contacts 
 made in one minute. 
 
 Results: The average, standard deviation, and range for groups 
 I and II (200 freshmen in all), is indicated in Table I below: 
 
 TABLE I 
 
 Test No. i 
 Coordination 
 
 Group I 
 
 Group II 
 
 Test No. 2. Tapping 
 
 This test has been widely used as a test of motor speed and endur- 
 ance and has been considered by some experimenters to afford the 
 best index of motor capacity. 
 
 Apparatus: Tapping board with metal plate and electric counter. 
 
 Range 
 
 
 
 Poorest 
 
 Best 
 
 
 
 (Av. of 
 
 (Av. of 
 
 Average 
 
 S. D. 
 
 lowest 5) 
 
 best s) 
 
 82.7 
 
 10.77 
 
 63-8 
 
 109.0 
 
 8 4 .I 
 
 11.92 
 
 60.8 
 
 IIO.4 
 
22 Psychological Examinations of College Students 
 
 Tapping stylus with flexible connecting wire attached. Two dry 
 cells. Stop watch. 
 
 Instructions: "I want you to hold this (stylus) in your right hand 
 and tap on here (indicating the brass plate) as quickly as possible. 
 I want to see how many times you can tap in a minute. Start when 
 I say 'Go' and stop when I say 'Stop.' " These instructions were 
 accompanied by an illustration of tapping by the experimenter. 
 For this test the subject sat directly in front of the tapping board, 
 resting her arm on the table, and assumed the position most con- 
 venient for her. 
 
 Method of scoring: The score represents the number of taps 
 made in one minute. 
 
 Results: Table II shows the results obtained in this test: 
 
 TABLE II 
 
 Test No. 2 
 Tapping 
 
 Group I 
 Group II 
 
 Range 
 
 
 
 Poorest 
 
 Best 
 
 
 
 (Av. of 
 
 (Av. of 
 
 Average 
 
 S. D. 
 
 lowest 5) 
 
 bests) 
 
 376.26 
 
 51.69 
 
 263.2 
 
 499.0 
 
 368.54 
 
 39-32 
 
 283.0 
 
 451.4 
 
 Test No. j. Cancellation 
 
 This test is well adapted for measuring concentration and alert- 
 ness of attention, maximum effort being required to accomplish 
 the task quickly and accurately. In addition to involving such 
 factors as "speed of perception" and "discrimination" it is partly 
 dependent upon the subject's muscular reaction to stimuli presented. 
 Owing to the fact, previously mentioned, that it was necessary to 
 complete all the tests in one hour, it was found advisable to limit 
 some of the tests. Inasmuch as we desired to include the Checking 
 Test which involves functions similar to those involved in Cancella- 
 tion and as it was believed that these two tests together would 
 exert an unfavorable influence upon the results of following tests 
 due to the eye-strain they would cause, it was deemed advisable to 
 use only one half of the Cancellation blank and one half of the 
 Checking blank. The halves of these blanks have been found by 
 Woodworth and Wells to be equal in difficulty and they suggest 
 that one half of the blank in the case of both these tests is a suf- 
 ficient test. Thus we were able to avoid undue eye-strain and were 
 further able to spend the extra time, saved from halving these two 
 tests, in lengthening three of the Association tests. 
 
Discussion of Tests 23 
 
 Materials: Woodworth-Wells number blank, Form A. 87 Stop 
 watch. A pencil was used for checking. 
 
 Instructions: After placing the blank on the table before the 
 subject, face downwards, the following instructions were given: 
 "When I say 'Go' I want you to turn over this sheet of paper, and 
 cross out all the 3*3, as quickly as possible, going across the paper 
 like this (illustrating). There are five 3's on every cross line so you 
 want to be sure to cross out all those on the first line before passing 
 to the second line. Start when I say 'Go.' " 
 
 Method of scoring: The time taken to complete the cancellation 
 was the score. Errors were very rare and were therefore entirely 
 disregarded. 
 
 Results: Table III indicates the performance in this test. 
 
 TABLE III 
 
 Range 
 
 Poorest Best 
 
 Test No. 3 (Av. of (Av. of 
 
 Cancellation Average S. D. lowest 5) best 5) 
 
 Group I 76.51 17.51 128.28 52.12 
 
 sec. 
 Group II 76.77 13.82 105.60 50.76 
 
 sec. 
 
 Test No. 4. Checking 
 
 This test measures functions similar to those employed in the 
 Cancellation test, although here the functions involved are more 
 complex. To quote Woodworth and Wells, "The detection of a 
 pair of digits in a group is a specialized performance, not reducible 
 to the acts of detecting the single digits. The difficulty of this test 
 is mainly perceptual and the overlapping which is effective in find- 
 ing pairs of digits must occur in the perceptive process." 38 Inas- 
 much as Professor Woodworth found the first half of his number 
 blank, Form B, to be equal in difficulty to the second half, for the 
 reason mentioned under "Cancellation" only one half of this blank 
 was employed. 
 
 Materials: Woodworth-Wells' number blank, Form B. Stop 
 watch. Pencil. 
 
 Method of procedure: As in the Cancellation Test, the blank 
 was placed before the subject, face downwards, and the following 
 instructions were given: "When I say 'Go' I want you to turn this 
 
 " Woodworth, R. S., and Wells, F. L. Association Tests. Psychological Monograph, No. 57, 
 1911, P. 24. 
 
 Woodworth, R. S., and Wells, F. L., Op. cit. 
 
Psychological Examinations of College Students 
 
 paper over and check any way at all, as quickly as possible, all the 
 numbers that contain both a 9 and a 6. Start when I say 'Go.' " 
 
 Method of scoring: The total number of checks to be made was 
 35. Therefore the score was obtained by dividing the time taken 
 by the subject by the number of correct checks made and then 
 multiplying by 35. No account was taken of wrong checks made as 
 it was believed that the time spent in making them sufficiently 
 penalized the subject. 
 
 Results : Table IV shows the performance attained in this test. 
 
 Teat No. 4 
 Checking 
 
 Group I 
 Group II 
 
 TABLE IV 
 
 Average 
 
 102.93 
 
 sec. 
 
 105.98 
 
 S. D. 
 19.64 
 
 20.45 
 
 Range 
 
 Poorest Best 
 
 (Av. of (Av. of 
 
 lowest 5) best 5) 
 
 152.28 72.6 
 
 161.0 76.86 
 
 Test No. 5. Color Naming 
 
 "This is a test of discrimination-reaction, involving prompt 
 decision and correct reaction to a situation." 
 
 Materials: Wood worth-Wells' Color Naming blank. 39 Stop 
 watch. 
 
 Method of procedure: Preliminary to the actual test the blank 
 was placed before the subject with only the sample line of five 
 colors showing. The subject was then asked to give the names of 
 each color. Then the following directions were given : "I want you 
 to name all these colors for me, as quickly as possible, going across 
 the paper, from left to right, as in reading. Start when I say 'Go.' w 
 
 Method of scoring: The score was the time taken by the subject 
 to complete the entire series of 100 reactions. 
 
 Results : The results are shown in Table V. 
 
 Test No. 5 
 Color Naming 
 
 Group I 
 Group II . 
 
 TABLE V 
 
 Average 
 56.01 
 
 sec. 
 
 58.55 
 sec. 
 
 s. D. 
 
 8-75 
 
 9-3 6 
 
 Range 
 
 Poorest Best 
 
 (Av. of (Av. of 
 
 lowest 5) best 5) 
 
 78.84 
 sec. 
 
 81.32 
 sec. 
 
 41.16 
 sec. 
 
 39-0 
 sec. 
 
 Op. cit. 
 
**"\. 
 
 Discussion of Tests (25 j 
 
 Test No. 6. Directions 
 
 This test measures the subject's speed in apprehension and her 
 general intelligence. 
 
 Materials: Woodworth-Wells' Hard Directions blank. Stop 
 watch. 
 
 Instructions: "When I say 'Go' I want you to turn this blank 
 over and follow directions do just what the directions say, as 
 quickly as possible." 
 
 Method of scoring: The score is the time in seconds required to 
 complete the test. Errors were counted separately. 
 
 Results : Table VI indicates the performance in this test. 
 
 TABLE VI 
 
 Range 
 
 Poorest Best 
 
 Test No. 6 (Av. of (Av. of 
 
 Directions Average S. D. lowest 5) best 5) 
 
 Group I 126.15 52.00 296.6 64.08 
 
 sec. sec. sec. 
 
 Group II 119.76 41.65 243.2 61.6 
 
 Test No. 7. Opposites 
 
 For a test which would indicate a general tendency or "adjustment ' 
 to react according to instructions" and also measure the quickness 
 and accuracy of association of ideas, the two equal lists of opposites 
 proposed by Woodworth and Wells were combined into one list. 
 Our reason for combining the lists was in order to get a real measure 
 of the individual's ability to name opposites. If we had taken only 
 the short list we would have obtained an adequate measure of the 
 subject's alertness of attention and ability to adapt herself to a 
 situation, but we desired to go further than this and find out whether 
 the individual really had any special ability for naming opposites. 
 This test also indicates facility in handling words and is generally 
 considered to have a high correlation with general intelligence. 
 
 Materials: Woodworth-Wells' Lists of Opposites printed on 
 cardboard. Stop watch. 
 
 Method of procedure: These instructions were given: "I want 
 you to name the opposite for each one of these words (showing 
 card with lists, at a distance) as quickly as possible, not repeating 
 the words themselves but just naming the opposite. For instance, 
 if the word were 'tall,' you would say 'short.' Be sure you give the 
 
26 Psychological Examinations of College Students 
 
 exact opposite of each word before proceeding to the next. Do you 
 understand?" 
 
 The subject was stopped if a wrong opposite was given and not 
 permitted to proceed with the other words until the right opposite 
 was given. 
 
 Method of scoring: As no errors were permitted to be made in the 
 test, the score represents the time taken for completing the task. 
 
 Results: Table VII indicates the results obtained in this test. 
 
 TABLE VII 
 
 Range 
 
 Poorest Best 
 
 Test No. 7 (Av. of (Av. of 
 
 Opposites Average S. D. lowest 5) best 5) 
 
 Group I 51.08 10.33 79-oo 34- 8 4 
 
 sec. sec. 
 
 Group II 50.88 8.55 71.52 35.92 
 
 Test No. 8. Verb-object 
 
 This is also one of the association tests and measures ability to 
 handle verbal relations. As in the Opposites Test we combined the 
 two equivalent lists of verbs proposed by Woodworth and Wells 
 into one test. Desire to obtain a real measure of the subject's 
 innate ability to name objects was the reason for lengthening this 
 test. 
 
 Materials: Two equal lists of verbs combined into one list and 
 printed on cardboard. Stop watch. 
 
 Method of procedure: These instructions were given: "In this 
 case I want you to name an object for each one of these verbs, as 
 quickly as possible, not repeating the verbs themselves but simply 
 naming the objects. For instance, if the verb were 'bake,' you 
 would say 'bread' or 'cookie.' Do you understand?" 
 
 Method of scoring: As no errors were permitted to be made, the 
 score presents the time required to complete the test. 
 
 Results: The results are indicated in Table VIII. 
 
 TABLE VIII 
 
 Range 
 
 Poorest Best 
 
 Test No. 8 (Av. of (Av. of 
 
 Verb-object Average S. D. lowest s) best 5) 
 
 Group I 65.55 12.32 99.56 45.48 
 
 sec. 
 Group II 67.35 I2 -9i 99- 8 47-24 
 
Discussion of Tests 27 
 
 . Test No. 9. Mixed Relations or Analogies 
 
 This test measures facility in handling associations, and ability 
 to perceive relationships among logical material. As in the two 
 preceding Association Tests the two equal lists proposed by Wood- 
 worth and Wells ("Eye: see = Ear- ; Oyster: shell = 
 
 Banana: " and "Good: bad = Long: ; Man: woman 
 
 = Boy: ") were combined into one long list for a reason 
 
 similar to that which led us to lengthen the Verb-object and Oppo- 
 sites tests. 
 
 Materials: Combination of Woodworth Wells' two equal lists 
 for Mixed Relations test, printed on cardboard. Stop watch. 
 
 Method of procedure: The subject was shown sample analogies 
 and the following instructions given: "In this case there are three 
 words given and you are to supply a fourth word that has the same 
 relation to the third word as the second word has to the first. For 
 
 example, in this case, 'Box: square = Orange: ,' square 
 
 gives the shape of the box. Then the shape of an orange is round, 
 so you would supply 'round* as the fourth term. (Two other illus- 
 trations were then given.) The relations involved won't always be 
 the same; it may be the case of shape, or opposites, etc. But you 
 look at the first pair of terms in every case and then make the 
 second pair express the same relationship as the first pair. Do you 
 understand ?" 
 
 Method of scoring: As no mistakes were allowed, the score is 
 the time required to complete the test. 
 
 Results: The results are shown in Table IX. 
 
 TABLE IX 
 
 Range 
 
 Poorest Best 
 
 Test No. 9 (Av. of (Av. of 
 
 Mixed Relations Average S. D. lowest 5) best 5) 
 
 Group I 139.64 42.97 266.6 82.88 
 
 sec. sec. sec. 
 
 Group II 131.66 32.97 227.2 79.56 
 
 sec. sec. sec. 
 
 Test No. jo. Word Building 
 
 For a test that would indicate ingenuity and skill in the manipu- 
 lation of letters and give a measure of the subject's command of 
 vocabulary, the word building test was used. The number of words 
 written in a given time depends in part on whether the subject 
 
28 Psychological Examinations of College Students 
 
 proceeds with a definite plan, combining, for example, "a" with all 
 the other letters, then "e" with all the other letters, -etc., or goes 
 about the task in a vague or random fashion. 
 
 Materials: Sheet of paper at the top of which were written the 
 letters a e i I p r. 
 
 Method of procedure: The procedure as given by Whipple 40 was 
 followed with the exception that the time-limit was three minutes 
 instead of five. 
 
 Method of scoring: The score represents the number of words 
 written. A word was considered correct if it is included in Whipple's 
 list of admitted words. 
 
 Results: Table X shows the results secured in this test. 
 
 TABLE X 
 
 Range 
 
 Poorest Best 
 
 Test No. 10 (Av. of (Av. of 
 
 Word Building Average S. D. . lowest s) best 5) 
 
 Group I . 16.33 4-93 6.0 27.2 
 
 words 
 Group II 16.23 4-5 2 6.4 24.6 
 
 Test No. ii. Word Naming 
 
 This uncontrolled association test appears to be a good test for 
 determining individual differences, the subjects tending to write 
 words belonging to various categories. Such differences as the 
 tendency to write series of rhymed words, to write a series of words 
 that are grouped about one central idea, then to write another 
 series of words grouped about a second central idea, suggested 
 perhaps by the last word in the first series, etc., are revealed in this 
 test. It also depends in part on the subject's speed of writing. 
 
 Materials: Stop watch. Sheet of paper and pencil. 
 
 Instructions as follows were given: "I am going to give you three 
 minutes in which to write all the words you can. It makes no dif- 
 ference what sort of words they are they can be anything you 
 want to write." 
 
 Method of scoring: The score equals the number of words written. 
 
 Results: Table XI shows the results for this test. 
 
 Test 12. Knox Cube 
 
 This test gives an indication of the subject's power of observa- 
 tion, memory, and ability to concentrate her attention. It involves 
 
 Whipple, G. M. Manual of Mental and Physical Tests. Part II, p. 275. 
 
Discussion of Tests 29 
 
 the ability to handle concrete objects and to imitate another's 
 performance with accuracy. 
 
 TABLE XI 
 
 Range 
 
 Poorest Best 
 
 Test No. ii (Av. of (Av. of 
 
 Word Naming Average S. D. lowest 5) best 5) 
 
 Group I ....... 67.14 12.78 40.8 94.2 
 
 words words words 
 
 Group II 67.87 11.86 45.0 93.0 
 
 Materials: Five one-inch cubes. 
 
 Method of procedure: Pintner's standardization of the Knox 
 test was followed. Care was exercised to execute all movements 
 slowly and deliberately and at a uniform rate. 
 
 Method of scoring: The score represents the number of lines 
 correctly imitated. 
 
 Results: Results are indicated in Table XII. 
 
 TABLE XII 
 
 Range 
 
 Poorest Best 
 
 Test No. 12 (Av. of (Av. of 
 
 Knox Cube Average S. D. lowest 5) best 5) 
 
 Group I 9.20 1.56 5.8 11.4 
 
 lines 
 Group II 8.82 1.64 4.8 12.0 
 
 lines 
 
 Test No. 13. Digit Span 
 
 To measure ability to reproduce with accuracy disconnected and 
 non-logical material, the digit span test was employed. It tests the 
 subject's power to concentrate her attention upon the series of 
 digits as they are read aloud to her by the experimenter and to so 
 retain said series in her mind that she may reproduce it with abso- 
 lute accuracy immediately after the experimenter has ceased 
 speaking. It affords an opportunity also to observe individual 
 differences. 
 
 Materials: Digit Span blank. Stop watch. 
 
 Method of procedure : These instructions were given : "I am 
 going to read some numbers to you and as soon as I have finished 
 saying them, I want you to repeat them in exactly the same order." 
 The smallest number ol digits given was five. Three trials were 
 
30 Psychological Examinations of College Students 
 
 given for each number. The attempt was made to repeat the num- 
 bers without rhythm. 
 
 Method of scoring: The score represents the highest number of 
 digits correctly repeated two trials out of three. 
 
 Results: Table XIII indicates the results of this test. 
 
 TABLE XIII 
 
 Range 
 
 Poorest Best 
 
 Test No. 13 (Av. of (Av. of 
 
 Digit Span Average S. D. lowest 5) best 5) 
 
 Group I ....*.... 7.39 digits 1.31 5 digits 10.2 digits 
 Group II 7.67 " 1.29 5.2 " 10.2 " 
 
 Test No. 14. Word Memory 
 Test No. 15. Logical Memory 
 
 Both of these tests call into play functions similar to those 
 demanded in the digit span test. However, here the material to be 
 reproduced has meaning, consisting in Test 14 of a series of con- 
 crete words and in Test 15 of a list of familiar proverbs. 
 
 Materials: Cards containing a list of 25 words and a list of 25 
 proverbs, respectively. Also two blanks containing 50 words and 
 50 proverbs, respectively. The cards and blanks were those em- 
 ployed by Edith Mulhall Achilles. 41 
 
 Method of procedure: Instructions were given as follows: "I am 
 going to let you look at a list of words (or proverbs as the case 
 might be) for one minute, after which I am going to ask you to write 
 as many of the words (or proverbs) as you remember." The subject 
 was allowed one minute in which to write down the words she 
 remembered and two minutes to write the proverbs. After record- 
 ing the words remembered the subject was given a second list in 
 which there were 25 words previously seen and 25 new words, and 
 was asked to mark "y" all the words she recognized as having seen 
 before and "n" those she thought she had not seen. Similar pro- 
 cedure was followed for the test with proverbs. 
 
 Method of scoring: For Recall the number of words or proverbs 
 written constitutes the score. No account was taken of the order 
 in which they were recalled, or any false recollections recorded. 
 
 In scoring Recognition this formula was employed to derive the 
 score : 
 
 tt Achilles, Edith Mulhall. Archives of Psychology, No. 44, 1920. 
 
Discussion of Tests 31 
 
 50 (which is the total number of words or proverbs) minus 2 x 
 number of errors = score. 
 
 Results: Tables XIV and XV indicate the results of these tests. 
 
 TABLE XIV 
 
 Test No. 14 
 Word Memory 
 
 Recollection Average S. D. 
 
 Group I 1 1. 59 words 2.70 
 
 Group II ....... 10.91 " 2.79 
 
 Word Memory Recognition 
 
 Group I 35.84 " 7.44 
 
 Group II 35.07 " 8.33 
 
 Range 
 
 Poorest Best 
 
 (Av. of (Av. of 
 
 lowest 5) best 5) 
 
 6.6 words 17.4 words 
 6.2 " 18.0 " 
 
 2O.O 
 14.8 
 
 47.2 
 48.4 
 
 Test No. 15 
 
 Logical Memory 
 
 Recollection 
 
 Group I 
 Group II . . 
 
 Logical Memory Recognition 
 
 Group I 
 
 Group II 
 
 TABLE XV 
 
 Average 
 
 6.19 
 
 6.50 
 
 proverbs 
 
 36.75 
 
 3747 
 
 S. D. 
 
 1.74 
 1.76 
 
 8-95 
 7.69 
 
 Range 
 
 Poorest Best 
 
 (Av. of (Av. of 
 
 lowest 5) best 5) 
 
 3-2 
 
 proverbs 
 
 17.2 
 
 18.4 
 
 9.6 
 
 9.8 
 
 proverbs 
 
 47.6 
 
 48.4 
 
 Test No. 16. Substitution 
 
 For a test which would measure speed of learning new associa- 
 tions the Substitution test was employed. In this test a key is 
 constantly referred to and as the test proceeds it is gradually learned, 
 the subject depending less and less upon it. Comparison between 
 the time taken to complete the first and second halves of the blank 
 gives a measure of the amount of time saved from learning the key. 
 
 Materials: Substitution test blank. The blank with 5 geometrical 
 forms was used. Stop watch. 
 
 Method of procedure: The key was explained to the subject and 
 then the blank was placed face downwards before her and she was 
 instructed to turn over the Substitution blank at the signal "go" 
 and to begin with the first form and take each one as it came, going 
 across the paper from left to right, and to write the proper number 
 in each form according to the key at the top. 
 
32 Psychological Examinations of College Students 
 
 Method of scoring: Three scores were taken, representing the 
 time for the first half of the blank, the second half and the whole 
 blank, respectively. Errors, being rare, were counted separately. 
 
 Results: The data for this test are found in Table XVI. 
 
 TABLE XVI 
 
 Range 
 Test No. 16 Average S. D. Poorest Best 
 
 Substitution i Half .... seconds seconds seconds 
 
 Group I 64.33 9-69 87.68 46.8 
 
 Group II 66.68 12.14 97-6o 46.0 
 
 Substitution 2 Half 
 
 Group I 59-io 11.62 86.2 37.0 
 
 Group II 61.51 13.15 91.8 38.4 
 
 Substitution Whole . . . 
 
 Group I 123.09 19.61 167.72 86.48 
 
 Group II 128.19 23.89 187.0 87.40 
 
 Test No. 17. Completion 
 
 For measuring correctness and facility in the use of words, readi- 
 ness in perceiving and comprehending situations and affording 
 some indication of creative ability, the Completion test was em- 
 ployed. To quote Trabue, "On the whole it will be found that 
 ability to complete these sentences successfully is very closely related 
 to what is usually called 'Language ability .' " ^ 
 
 Materials: Trabue Language Seal A. Stop watch. 
 
 Method of procedure: The standard procedure suggested by 
 Trabue was followed, a time-limit of four minutes being employed. 
 
 Method of scoring: In general, the method was to follow Dr. 
 Trabue's scoring; "A score of 2 being given each sentence if perfectly 
 completed, a score of I if almost but not quite perfectly completed, 
 and a score of o if not attempted at all or if imperfectly done." 
 Total of 48 points is the maximum score attainable in Scale A. 
 
 Results: Table XVII represents the performance of the freshmen 
 in this test. 
 
 TABLE XVII 
 
 Range 
 Poorest Best 
 
 Test No. 17 < Av - f ( Av - of 
 
 Completion Average S. D. lowest 5) best 5) 
 
 Group I 36.08 4.33 26.8 44.8 
 
 Group II 3578 4.36 25.2 44.4 
 
 Trabue. Completion-Test Language Scales. 
 
Discussion of Tests 
 
 33 
 
 Test No. 18. Information 
 
 To measure range of information and obtain some conception 
 of the number and kind of objects known and the degree to which 
 they are known, the information test was used. It tests the individ- 
 ual's knowledge rather than her ability. 
 
 Material: The information test blank as specified in Whipple's 
 Manual, containing 100 words and directions for marking them. 
 
 Method of procedure: The subject followed the directions at the 
 top of the blank, marking each word with a certain letter which 
 indicated the degree to which it was known to her. There was no 
 time-limit in this test, the subject being allowed all the time she 
 desired to finish the blank. 
 
 Method of scoring: The score represents the number of words 
 marked "D," "E" "F," and "N," respectively. As no check was used 
 in this test, the score probably shows over-estimation. The total 
 score was obtained by assigning these values: D = 3; E = 2; 
 F = I ; and N = o, and taking their sum. 
 
 Results: The table following indicates the results of this test. 
 
 TABLE XVIII 
 
 Test No. 1 8 
 
 Average 
 
 Information D 2 1.47 words 
 
 Information E 13. 70 " 
 
 Information F 14.81 " 
 
 Information N 50.01 " 
 
 Total Score: 
 
 Group I 106.63 
 
 Total Score: 
 
 Group II 104.71 
 
 s. D. 
 
 9.71 
 
 6.16 
 
 6-43 
 
 10.35 
 
 Range 
 Poorest Best 
 
 3. 6 words 41 .6 words 
 
 3 
 
 1.8 
 69.6 
 
 25-51 59-8 
 
 26.79 554 
 
 28 
 
 26.2 
 
 29 
 
 158.2 
 161.8 
 
 Test No. ip. Vocabulary 
 
 This test merely indicates the number of words in the individual's 
 vocabulary. 
 
 Materials: Vocabulary test blank as specified in Whipple's 
 Manual. 43 
 
 Method of procedure: The subject was asked to follow the 
 directions given at the top of the test blank and to mark the words 
 carefully according to the directions. 
 
 'Op. cit. Vol. 2, p. 310. 
 
34 Psychological Examinations of College Students 
 
 Method of scoring: The score represents the number of words 
 marked plus (+). This number indicates the vocabulary-index; 
 the index taken as a per cent, is multiplied into 28,000. 
 
 Results: Table XIX shows the results for this test. 
 
 TABLE XIX 
 
 Range 
 
 Poorest Best 
 
 Test No. 19 (Av. of (Av. of 
 
 Vocabulary Average S. D. lowest 5) best S) 
 
 Group I 74.8 1 words 6.86 59.6 86.6 
 
 Group II . 73.90 " 7.60 59.4 87.4 
 
SECTION V 
 
 NORMS OF PERFORMANCE AND THEIR PRACTICAL 
 APPLICATION 
 
 To summarize the results of the preceding section, Table XX 
 shows the norms of performance for the two hundred Barnard 
 freshmen (Groups I and II), in all the various tests. The average, 
 probable error, and range from the poorest to the best score are 
 shown for each test. To avoid misrepresentation of facts by undue 
 weight being given extreme cases, the average of the five poorest 
 scores is in each case taken as the poorest score, and the average 
 of the five best scores as the best score. 
 
 The following is a comparative table comparing our results with 
 those ofother investigators who have employed some of these tests 
 with freshmen. Only those cases are considered where the tests are 
 identical, and the method of scoring the same. 
 
 Test 
 
 Cancellation 
 Color Naming . 
 Hard Directions 
 Opposites 
 Word Building . 
 Digit Span . 
 Information 
 
 Barnard Norm Bingham 
 
 76.6 sec. 48.3 sec. 
 
 57.2 sec. 56.2 sec. 
 122.9 sec- 
 
 50.9 sec. 
 
 1 6.2 words 
 7-53 digits 7 digits 
 
 20.4 words 
 
 Kitson 
 69.2 sec. 
 
 Other Investigators 
 
 Washburn, 153 sec. 
 
 1 1 0.9 sec 
 52. 6 sec. 
 
 21.4 words Sunne, 1 8 
 
 8.4 digits Cattell, 7.6 
 
 Waugh, 24 
 King& 
 
 M'Crory, 25 
 
 Smith, 10.9 
 
 Figures I to 23 inclusive, show graphically the dispersion of 
 measures about the average in the case of the Barnard freshmen. 
 To secure uniformity and facilitate comparison, the charts are 
 constructed with the average in each case as the mid-point and the 
 scores expressed in terms of P.E. units from the average as a center. 
 The P.E. was taken as the unit because it is a convenient and 
 familiar measure. The vertical scale is also kept constant except 
 in three tests where it is changed for reasons to be specified later. 
 Inspection of these figures reveals many interesting features. 
 
 We may divide the tests roughly into five groups. 44 The first 
 group contains the two motor tests Coordination and Tapping. 
 
 44 Justification of this division of the tests will be given in Chapter VI. 
 
Psychological Examinations of College Students 
 
 o- 
 
 O NO oO vO vO d vO vO QOOOOOOOOOOQOO 
 C v JOpI s ^>-'ONO\ONvO(N|pTt'NOocrN|T$-ooo<(NiOfNiO 
 ^" t^ l~^ t> <NI OoOod O od fC> 
 
 'T t^ (N ONI-I >-l HH rl-HH Tl-Tj-( 
 
 ^HOO MOO 
 O ^r^f 
 
 OOOOVONOOOOOOO QOOOOO 
 
 tOc tCMOOVO fO 
 >H M O Ch O^ M IO 
 
 OOOO IOI-H 
 
 t>.t^O roiO-^-vO t^"-" 
 
 I* 
 
 (i o 
 
 * 00 VO 00 
 > IO 00 ^ C> 
 
 00 00 
 
 t>. 00 t^. O 
 
 t^O 1^00 
 
 o 
 
 \O t-i oo 
 vo VO <M 
 
 rh 
 O 
 
 > 
 
 <o 
 
 M OON\O 
 
 IOVO fiM 
 
 ON t^. -i 10 vO 
 
 CO ^O VO 
 
 5 o a 
 
 S d - -2 
 
 :ls-s 
 
 81 J 
 
 g g> 
 
 | 
 (3 (2 
 
 w s^ o 
 
 slii-s 
 
 E- 
 
 O O 
 
 " I S S 
 
 c fe c > 
 
 O O M <N| ^ 
 
 6 6 cs c c 
 
 8 1 . . I 
 
 It 
 
 1 -1.^ -S b t "2 - - M -M -o TJ 1 3 '5 * ^ 1 -S 
 
 O > 
 
 5 I ^ 2 
 
 < 
 
 t^ 00 ON O *H 
 
4 3 I -I A +1 Z 3 
 
 (16. I COORDINATION 
 
 3 2 -I 
 
 F I 6. 2 
 
 A +1 2 3 
 
 TAPPING 
 
 F 16. 3 
 
 -I A 41 2 3 4 
 CANCELLATION 
 
 XI 
 
 5 
 
 J1 
 
 4 3 
 
 F I 6. 4 
 
 2-1 A +1 2 3 
 
 NUMBER CHECKING 
 
7 654 3 2-1 A +1 234 
 
 F I 6. 5 COLOR NAMING 
 
 8 7 6 5 4 3 2 -I A +1 2 3 
 FIG. 6 DIRECTIONS 
 
 4 3 2 -I * +1 2 
 F I G. 7 P P S I T S 
 
 35 
 
 30 
 25 
 20 
 15 
 10 
 6 
 
 5 43 2-1 A +i 23 
 F I 5. 8 V [ R B G J t C T 
 
87 654 32-1 A +1 23 
 
 F I 6. 9 MIXED RELATIONS 
 
 _ r^ 
 
 it 
 
 mm 
 
 - 
 
 
 if 
 
 
 785432-IA+I 2 
 
 F I 6. I WORD BUILDING 
 
 54 3 2-1 Afl 2 3 
 F I 6. I 2 R H X CUBE 
 
6 4 Z * I 4 
 
 F I 8. I 4 WflROMEMOfir RECOLLECTION 
 
 U 
 
 6 5 4 3 2 - I 4 +I 2 3 
 F I 6- 15 WORD RECOGNITION 
 
 3 I - I A +1 Z 3 \ 5 
 
 Ft GIG LOGICAL II E M R Y - R E C L L E C T I N 
 
Norms of Performance and Application 41 
 
 Here we have fairly uniform distributions. The actual range for 
 Coordination is from 3^ P.E. to + 5> P.E. (skewed at the 
 positive end), and for Tapping from 5^2 P.E. to + 7 P.E. But 
 to take the actual range as the basis of our comparison is misleading. 
 A clearer conception of the facts is obtained by noting the closeness 
 with which the measures distribute themselves about the central 
 tendency. In these two motor tests we find a fairly uniform dis- 
 tribution, suggesting that the tests are adequate for selecting good 
 and poor subjects even in a group as homogeneous as college fresh- 
 men. 
 
 In the second group we may place those tests which involve 
 powers of perception and comprehension, namely, cancellation, 
 checking, color naming, word naming, and substitution. Here 
 again we find a distribution approximating the normal curve of 
 distribution. At first glance it would appear that in four of these 
 tests the curves are skewed toward the negative or poor end. In 
 both Fig. 3 and Fig. 4, (Cancellation and Number Checking), we 
 find a case at 7>^ P.E.; in Fig. 5 (Color Naming) we find one at 
 7 P.E.; and in Figures 18, 19, and 20 (Substitution), we find 
 cases at 9 P.E.; 7 P.E., and 8 P.E.; while at the good end 
 no case exceeds + 4 P.E. We must take care, however, not to let 
 these extreme cases mislead us as to the general character of the 
 distribution. If we count up the cases on either side of the average 
 we find 1 08 cases above the average in Cancellation, 109 in Number 
 Checking, 106 in Color Naming, 107 in Substitution, and 98 in 
 Word Naming. Thus we really have a more or less uniform dis- 
 tribution with a tendency of the number of scores above the average 
 to exceed the number below it. Disregarding the few extreme cases, 
 we find the majority of the scores contained within the normal 
 limits of the P.E. distribution, (- 4 P.E. to + 4 P.E.). 
 
 In the third group we may place the tests involving associative 
 relations, namely, Directions, Opposites, Verb-object, Mixed Rela- 
 tions, Word Building, and Completion. Here, likewise, as in the 
 two preceding groups, we find fairly uniform distributions with a 
 greater number of cases above than below the average, (except in 
 Word Building, where the distribution is about equal). The major- 
 ity of cases are likewise contained within the normal range of 8 P.E., 
 but there are a few extreme cases at the poor end in Completion, 
 Opposites, Verb-object, Mixed Relations, and an extreme case at 
 both the good and bad end in the Word Building test. 
 
 The fourth group contains those tests which call into play powers 
 
[U I 
 
 87 654 32-1 A + 1 23 
 FIG. 17 LOGICAL RECOGNITION 
 
 98 7654 32-1 A +1 23 
 
 F I G. I 8 S U B S T I T U T I N- F I B S T H A L F 
 
 7 6 5 4 3 2 - I A +1 2 3 4 
 f I 6. I 9 S U B S I I T U T I H- S E C N HALF 
 
 8 7654 32-1 A -H 234 
 F 1C. 20 SUBSTITUTION WHOLE 
 
Norms of Performance and Application 43 
 
 of learning, viz : observation and retention, namely: Word Memory, 
 and Logical Memory. 
 
 A word of explanation is needed here regarding the construction 
 of the chart for Logical Memory (Recollection). The categories into 
 which the scores fall are so few that the finest grouping possible is 
 in i P.E. units instead of $4 P.E. units as in the other tests. As 
 we said before, to secure uniformity we let the P.E. represent the 
 same interval along the base line in all tests. Now, in order to keep 
 the area of a given number of cases constant for all tests, it is neces- 
 sary where we have scores in terms of I P.E. units to reduce the 
 vertical scale proportionately. Therefore, we regard the measures 
 as distributed evenly over the P.E. intervals and reduce the vertical 
 scale one-half. In this test and in Word Recollection we find a 
 greater number of cases below the average than above. The curve 
 is skewed toward the poor end in Word Recollection, and toward 
 the good end in Word Recognition and Logical Recognition. 
 
 In our fifth group we have tests which depend on the subject's 
 knowledge rather than her innate ability, namely, Information and 
 Vocabulary. Here we find fairly uniform distributions with no 
 extreme cases. This suggests the tendency of education to make a 
 homogeneous group of individuals approach a general level of per- 
 formance in a test of mere learning. 
 
 We have, finally, a miscellaneous group which comprises the 
 Digit Span and Knox Cube tests tests which showed both a low 
 intercorrelation and low correlations with the other tests of the 
 series. In the Knox Cube test the small number of categories makes 
 it necessary to use I P.E. units and in the Digit Span test it is 
 necessary to use 2 P.E. units. 
 
 To sum up then, these surfaces of distribution are fairly symmet- 
 rical, if we disregard the few extreme cases. In addition, the fact 
 that the averages and surfaces of distribution for the first group of 
 one hundred freshmen (Group I) are approximately the same as for 
 the second group of one hundred (Group II), corroborates this con- 
 clusion and supports the view that the norms here presented are 
 reliable. 
 
 ACADEMIC GRADES 
 
 Besides their score in the psychological tests we have additional 
 information about the first group of one hundred freshmen (Group I) 
 in the form of university grades and records taken in the gym- 
 nasium. The college subjects may be grouped into five classes: 
 
ll 1 
 
 L 
 
 L, 
 
 654 32-1 A 1-1 234 5 
 
 F I G. 2 I COMPLETION 
 
 FIG. 22 INFORMATION 
 
 4 3 2-1 1*1 2 3 4 
 
 f I 6. 23 VOCABULARY 
 
Norms of Performance and Application 45 
 
 I. Language (including English, Latin, Greek, German, French, 
 Italian, and Spanish); 2. Mathematics; 3. Science (physics, chemis- 
 try, botany or geology); 4. Philosophy (including psychology); 
 and 5. History. Due to the freedom allowed the students in making 
 out their programs, the same subjects are not taken by all, and the 
 number of cases in each class therefore varies. The letter system of 
 marking is employed at Barnard, the letters A (excellent), B (good), 
 C (fair), D (Poor), and F (failure), being used. For the statistical 
 treatment of the data the letter grades were transformed into 
 numbers according to the scale: A = 90, B = 80, C = 70, D = 60, 
 and F = 50. Norms for these freshmen in their college work are 
 shown in Table XXI. 
 
 TABLE XXI 
 
 Academic Number of Range (Actual) 
 
 Record Cases Average P. E. Lowest Highest 
 
 1. Language 97 75.31 4.69 50 90 
 
 2. Mathematics .... 88 76.99 6.99 50 90 
 
 3. Science 41 72.26 7.74 50 90 
 
 4. Philosophy 27 78.15 3.15 60 90 
 
 5. History 26 72.88 2.88 60 90 
 
 The averages tend to be approximately equal for all subjects 
 with a nearly equal range of distribution. 
 
 PHYSICAL MEASUREMENTS 
 
 Table XXII gives averages, P.E.'s, and range from lowest to best 
 score of the physical measurements taken in the gymnasium. 
 
 TABLE XXII 
 
 Number Range (Actual) 
 
 Test of Cases Average P. E. Poorest Best 
 
 Height 97 159.92 cm. 4.08 137 172.9 
 
 Weight 97 120.59 Ibs. 12.59 9 182 
 
 Lung Capacity 94 171.05 cu. cm. 13.50 118 230 
 
 Strength of Grip, r.h 97 30.02 kg. 4.02 13 43 
 
 Strength of Grip, 1. h 97 27.27kg. 4.27 16 38 
 
 Upper Back 97 20.60 kg. 3.4 12 42 
 
 Chest 97 19. 60 kg. 2.6 u 36 
 
 One of the main purposes of this investigation, as we remarked 
 in a preceding section, was to give the individual student a knowl- 
 edee of her strengths and weaknesses. Accordingly, at the com- 
 
Tdppiny 368 
 
 Ct^n-CcHa-tlon. 60 
 
 CktckCnfr 13.9 
 Co tot Naming SS' 
 
 Directions lOO 
 Opposite* 37 
 Vetb Objtct 43 
 Mixed fftbtionf 81 
 
 Woid Buitatina 10 
 Motet Namina, 15 
 Completion f/8 
 Kn.9X Cult 8 
 
 Diait Span '0 
 Woidfocoft. IS 
 
 LooU*.? ftectQ. M 
 Substitution /OK 
 Information //& 
 V*CA&UQSUJ 75 
 
 Language 10 
 v * 
 
 Pfatatopfy <IO 
 Hi-fhttt 10 
 
 Helot* /fS.Z 
 VIcL^t 120.5 
 
 m 
 
 G^ip Ctft Z5 
 Uppet t>AcK -23 
 Ckeft ZZ 
 
 IS 
 
 fe 
 
 loo 
 Sa 
 
 Iw 
 
 h 
 
 |s 
 
 3 
 
 N 
 
 OS 
 
 5 
 
 N 
 
Norms of Performance and Application 47 
 
 pletion of the entire series of examinations each year, an individual 
 report was sent to each student who took the tests. This consisted 
 of two blanks giving a description and interpretation of the various 
 tests, with whatever significance. each test was known to possess 
 from a vocational standpoint. In addition to these explanatory 
 blanks, there was a third blank which indicated the standing of the 
 individual student in each of the tests, together with the average 
 standing, (with the P.E.), in each test for the entire group of one 
 hundred freshmen, so that the individual could compare her own 
 record with that of the average in every case. 
 
 The ideal plan would have been for the experimenter, after 
 sending each student her report, to have had a personal interview 
 with her. In this she could have cleared up any difficulties the 
 student might have had in interpreting her results and under- 
 standing their significance. She could also have rendered distinct 
 aid by suggesting means whereby the student could make the best 
 use of her abilities, or strengthen her weak points. Where the girl 
 was doing academic work of a grade below the level her test record 
 showed her capable of, the experimenter could have sought to 
 determine the cause of the girl's academic failure whether due to 
 too many distractions, outside work, or what not and given advice 
 accordingly. Lack of time made it impossible to do this, however. 
 We therefore have no record of these girls in their last three years of 
 college to show whether they benefited from their test results. It 
 is worth while at this point, nevertheless, to indicate how one may 
 proceed to make practical use of these tests. 
 
 Charts I to 6, inclusive, represent the psychographic records 
 of six students from Group I. They are constructed as follows: 
 Reading along the heavy horizontal base line, we have the names 
 of the nineteen psychological tests, (Substitution First Half and 
 Substitution Second Half are omitted since ability in this test is 
 adequately measured by Substitution Whole), the academic 
 subjects varying from two to four, according to the programs of 
 study, and seven physical measurements. Opposite the name of 
 each test, subject, and physical measurement is the individual's 
 score, and below this, the amount of her plus or minus deviation 
 from the average scores expressed in P.E. units. To make the 
 individual's relative standing more concrete, her score in P.E. 
 units is also expressed in terms of what her position would be in a 
 group of one hundred freshmen, selected at random. 
 
 The vertical line (reading up from the base line) is divided into 
 
Coo zdi nation 81 
 
 
 Is 
 
 CancetfatCo* 6/ 
 
 
 
 Cfackin.fr 81 
 
 
 
 Cot** nasnino 57 
 
 h 
 
 "*^ 
 
 Directions 157 
 
 R 
 
 
 Opposite* 87 
 VetbQbjiU 61 
 Mindfota* 224 
 
 
 
 
 1* 
 
 |OO 
 
 WoidBuMino 13 
 
 1* 
 
 
 Compfttion 30 
 
 h 
 
 
 KM* Cu$e 8 
 
 r 
 
 
 Digit Sf>A* S 
 
 
 
 
 V/Qtci Kttoa . 32, 
 
 |S 
 
 
 LQiica? Rttca. 38 
 a (T 
 
 
 Is 
 
 Infvtmc&An 58 
 VocAbutaiij 61 
 
 K 
 
 H 
 
 
 Lan.quQ.ae. $5 
 
 k, 
 
 
 Science 50 
 
 lT 
 
 
 Hiftot*/ 60 
 
 HeiA&t 160.5 
 o 
 
 Weiete lc3.S 
 
 N 
 
 Is 
 
 Utno Capacity HZ 
 $T,ip left Z3 
 
 N 
 N 
 
 N 
 
 N 
 
 00 
 
 co 
 
 ISO 
 
Tappina 
 
 4/i 
 
 C&eckina 
 
 73. 
 
 6otoz N 
 Dittcttms I35 
 Ofpositts M. 6 
 Vfi&O&ject (J. 
 f?etat. 
 
 4 Qict Naming S3 
 
 37 
 10 
 7 
 * 
 
 tnox 
 
 fad 
 
 tfttd /? 
 
 A f 
 
 7 
 
 a 
 /c?V.8l 
 
 76 
 
 6Z\ 
 
 /72| 
 
 00 
 
 c 
 
 o 
 
 20 
 
 ** 
 
 Co 
 
 CO 
 
 i* 
 
 3/ 
 
Cooidin&tion Sl 
 
 ina 3/U t 
 
 Ca.nceeea.tSon 
 
 Cotoi. Nantino 
 
 ** ?m r 
 
 60 
 Hixtd ftekt. /o8\ 
 
 7 
 
 Wotd Naumtn* S6 
 
 37 
 
 Diait Span. S 
 
 focotf. 7 
 
 LoqitoJt fttcoo 
 Substitution 
 
 ^n lo l 
 7 
 
 60 
 
 60 
 
 Luna C&faciLulbO 
 
 I* 
 
 ON 
 
Norms of Performance and Application 51 
 
 equal divisions, indicating position in a group of one hundred fresh- 
 men selected at.random, using the norms of Table XX as the basis. 
 No. i is considered the poorest individual in each case, No. 100 the 
 best. The heavy horizontal black line in the center represents the 
 average individual or the 5oth individual in the group. To illus- 
 trate the use of these charts let us consider Chart I, A.M.'s record. 
 In coordination this individual scores 96. Referring to Table XX, 
 we see that the average freshman score for this test is 83.42 with a 
 P.E. of 7.5. A.M.'s deviation from the average score is, therefore, 
 + 12.58 (96-83.42) -f- 7.5 (the P.E.) or + 1.67 P.E. units above 
 the average. We know from the normal curve of distribution that 
 between the average and + I P.E. are found 25% of the cases, or 
 25 cases in a group of one hundred individuals. Between I P.E. 
 and + 2 P.E. there are approximately 17% more cases, or 17 in a 
 group of one hundred individuals, so that if a girl made a score 
 of + 2 P.E. she would rank 50 (average) + 25 + 17, or 92 in the 
 group. A.M., however, does not quite reach this score. Her score 
 reaches only. 67 of the interval between -f I P.E. and + 2 P.E., or, 
 .67 of the 17 cases contained within these limits. Now .67 X 17 = 
 11.39, * e -j A.M.'s score is that of the nth individual in this group. 
 This is only her approximate position, of course, since the scores are 
 not distributed evenly over the interval. To secure her exact 
 position we would transform her P.E. score into rank according to 
 proper table. 5 She therefore stands 50 + 25 + 1 1 , or 86 in a group 
 of one hundred freshmen in coordination. In Tapping her score is 
 368 taps. The average freshman score in this test is 372.4 taps with 
 a P.E. of 27.6. A.M.'s deviation from the average, accordingly, is 
 4.4 (372.4 368); her deviation in terms of P.E. is 4.4 -5- 
 27.6 (the P.E.), or she is .15 P.E. units below the average. Her 
 score therefore reaches .15 of the 25 cases in the interval between 
 the average and i P.E. Now, .15 X 25 = 3.75. Her score there- 
 fore gives her a rank 3.75 or approximately 4 places below the aver- 
 age or 5Oth individual, i. e., she stands 46 in a group of one hundred 
 freshmen. A similar method was employed in finding out the 
 psychographic records of the other five students. Considering the 
 net scores in the psychological tests, A.M. ranked 97 in Group I, 
 only three individuals surpassing her. When we group the tests 
 under the five divisions suggested above, we see that although she 
 would stand well above the average in a random group of one hun- 
 dred freshmen, she makes her highest rank (88 average rank for 
 
 Thorndike, E. L. Mental and Social Measurements. 
 
SI 
 
 317 
 
 *. 
 
 CoCoz 
 
 a 
 J77.& 
 
 Vet,b Objtct 6<? 
 Mixed, fiekt. //S4 
 
 If/Old Bui&kni 13 
 
 lft } 
 
 Knot Cu&e 9 
 Dijit S fA n 6 
 Moid, facott 9 
 V/otd. ftecoa. 3*+ 
 Logicol focott. 4 
 LoaicaC Rtcoo . 3Z 
 Substitution /3$ 
 
 : 113 
 71 
 
 88.33 
 30 
 
 PJutorophy 90 
 Heitfo /53.<? 
 
 Q 
 
 Luut Ca^dta 
 
 15 
 
 h 
 
 |8 
 
 i 
 
 Uc 
 o 
 
 r 
 i 
 
Norms of Performance and Application 53 
 
 this group) in the group of tests which involve the association 
 processes, i. e., in those tests involving more complex and higher 
 abilities. Moreover, she made the highest standing in academic 
 marks of any freshman in Group I, being the only one to secure 
 grade A in all the subjects she pursued during the year. It is of 
 interest to note also that the subject's score in physical measure- 
 ments is above the average. The tests therefore give an adequate 
 measure of this student's ability. 
 
 Chart 2. L.H.C. This freshman presents the other extreme of 
 ability. With the lowest academic standing of Group I, (having 
 no mark higher than D grade), she also ranks only 26 in net test 
 score. She is especially deficient in the association tests. In a 
 random group of one hundred freshmen she would rank only I in 
 Opposites and Mixed Relations, showing poor powers of associating 
 ideas and perceiving relationships among logical material, and 8 in 
 Completion, which measures readiness in perceiving and compre- 
 hending situations. She is also poor in the memory tests. In the 
 second group of tests which involves ability to perceive what is 
 wanted and to carry out simple instructions, she ranks above the 
 average, suggesting that she would do well at simple types of 
 clerical or stenographic work, though she lacks ability to perform 
 work requiring a higher level of intelligence. In Information and 
 Vocabulary her low rank of 4.5 is what we would expect. Having 
 no aptitude for study, it is only natural that she should be unin- 
 terested in it. Her physical report was also below average. All 
 indications confirmed her psychological report that she was unfitted 
 to pursue college work. Her failure to meet the academic standard 
 set for freshmen necessitated her withdrawal from college at the 
 end of the year a course justified by her psychological record. 
 
 Chart 3. G.S. Although in academic work this individual ranked 
 only 21 in the group of one hundred, her net score in the psychologi- 
 cal tests gave her a rank of 74. Her record in Group 3, i. e., in the 
 tests requiring the highest mental abilities, indicated that she was 
 doing work of a grade far below her ability. Her net score in the 
 tests of Group 4 suggested, and her record in the Information and 
 Vocabulary tests, which depend chiefly on knowledge acquired, 
 corroborated the hypothesis that she was neglecting her college 
 work. In her case interest in athletics furnished the explanation 
 for her college record. Not only was her physical record the highest 
 in the class, but G.S. was a prominent figure in all college athletic 
 events, especially in the swimming meets and in basket-ball games. 
 
t7S 
 
 TcLopina 3*W 
 Caacelt*tn. 5</.2 
 CfutckCnq 1( 
 
 CotA/amtA4 4j 
 
 / 
 
 Oppo^ter 37.6 
 Vtib Object St.S 
 Mixed fotat. //6 
 
 VJord /Vamino 58 
 
 Completion. V3 
 
 /Tnj Ca^e ^ 
 
 ^^ J^MZ/t.. 9 
 
 h/^.^eco ? . 40 
 
 Loqico.1 Kecoa. W 
 
 a 
 
 Sufatitution W.Z 
 
 7V 
 
 Lanauait 7/.2f 
 <F 
 
 itcy 7J.33 
 /67.^ 
 
 Lua* 
 
 34, 
 
 Cfust 
 
 Go 
 
 Is 
 
 Is 
 
 N 
 
 Ui 
 
 ha 
 
 00 
 
 S 
 
 H 
 
 K 
 
 c 
 
 k 
 
 I 
 
Norms of Performance and Application 55 
 
 Chart 4. I.E. This case parallels L.C.'s. I.E.'s net academic 
 rank was only 3 and her rank in the tests was also below freshman 
 standard. Like L.C., also, I.E.'s withdrawal from college at the 
 end of her first year was fully justified. 
 
 Chart 5. L.J.H. Here we have a case of a girl with a physical 
 record above the average, and a rank of 95 in academic standing, 
 but whose net score in the psychological tests is only 17. Having 
 no other information about this girl besides the test data and her 
 school marks, we cannot definitely explain this case. In only six 
 of the tests does she rank above average, but two of these Mixed 
 Relations and Completion involve the most complex mental 
 functions, powers of understanding, and reasoning. It may be that, 
 lacking powers of immediate recall, this girl was willing to devote 
 long hours to grasping the subject matter of her studies so that by 
 extra effort she was able to make high grades. Her score in Infor- 
 mation and Vocabulary also suggests her attention to her studies. 
 
 Chart 6. M.M. This case presents the other extreme. Here 
 we have a freshman who is in fine physical condition and has a net 
 score of 77 in the psychological tests, but whose net academic 
 standing is only 26. Inasmuch as she stands well above the average 
 in all the tests involving the higher mental processes, her academic 
 failure is probably due to lack of interest in her studies, or to too 
 many outside activities. 
 
SECTION VI 
 
 INTER-TEST CORRELATIONS AND THEIR 
 SIGNIFICANCE 
 
 The psychographic charts showed that a freshman rarely did 
 equally well in all the psychological tests. Whereas she tended to 
 make approximately the same standing in all her academic subjects, 
 she showed no such uniformity in the psychological tests. There 
 were, of course, a few extreme cases where a good student scored 
 above average in the majority of the tests, (for example, A.M.), or 
 a poor student scored below average, (for example, L.H.C.). This 
 raises the interesting question: Just what is the nature of the 
 relationship existing between these tests? Are some more closely 
 related than others? Is there any evidence to support our division 
 of the tests into the groups suggested in the preceding section? 
 For determining the relationship between the tests the particular 
 method of correlation used in this investigation was one suggested 
 by Professor Woodworth for combining the results of several tests. 46 
 By the use of his method it is possible to assign each individual her 
 position in the distribution of the group; she stands, in other words, 
 "above or below the group average and so and so much above or 
 below as compared with the average variation of the group." The 
 method of procedure is as follows: The average of any test is regard- 
 ed as zero, and the individual's standing is expressed as a deviation 
 above or below the average. Then the measure of variability (in 
 this case the S.D.) is taken as the unit of deviation from this zero, 
 and all deviations are expressed as fractions or multiples of the unit. 
 Each individual deviation, then divided by the S.D. of the series, 
 gives a resulting quotient called the "reduced measure." Having 
 obtained the reduced measures, by appropriate substitution in the 
 Pearson formula for correlation, we may easily obtain the correla- 
 tion of two given tests A. and B., for, given the reduced measures 
 of two arrays, the coefficient of correlation between them is the 
 average of the products of the various reduced measures. The 
 advantage of using this method is that the net position of an in- 
 dividual in a group of tests, for example, in the twenty-three tests 
 
 "Woodworth, R. S. Combining the Results of Several Tests; A Study in Statistical Method. 
 From Psychological Review, March, 1912. 
 
Inter-Test Correlations and their Significance 57 
 
 here used, may be easily obtained by dividing the sum of her reduced 
 measures in those tests by the number of tests, (twenty- three in this 
 instance). 
 
 Table XXIII gives the inter-test correlations computed according 
 to this method. The test records used in obtaining these correla- 
 tions are those of the one hundred freshmen of Group I. Inspection 
 of this table reveals many interesting features. The correlations 
 range from -f .77 (between Cancellation and Digit Span) to .00 
 (between Tapping and Word Recollection, and between Mixed 
 Relations and Word Recollection). The highest correlations are 
 + .77 (between Cancellation and Digit Span) ; + .58 (Word Recol- 
 lection and Word Recognition); + .57 (Opposites and Mixed 
 Relations); -f- .56 (Logical Recollection and Logical Recognition); 
 + .51 (Cancellation and Checking); + 48 (Coordination and 
 Tapping) ; + .48 (Mixed Relations and Completion) ; + .44 (Oppo- 
 sites and Verb-object); and + .40 (Cancellation and Word Nam- 
 ing). That the Cancellation test furnishes the highest single corre- 
 lation is interesting because it contradicts the old compensation 
 theory and McCall's finding of a negative correlation ( .28) 
 between this and the Trabue Completion test. All our correlations 
 with Cancellation are positive, ranging from + .03 to -f .77. 
 Especially noteworthy are the correlations of + .40 with Word 
 Naming, -f .30 with Word Building, and -f .31 with Substitution 
 all tests calling into play the higher thought processes. The fact 
 that the correlations are all positive is suggestive of a definite rela- 
 tionship between cancellation and these various tests. 
 
 Checking and Word Naming show the highest average correlation 
 (+ .25) with the other tests (omitting Information, Vocabulary, 
 Word Recollection, and Word Recognition). Then, in order, 
 Opposites, Verb-object, and Cancellation; Color Naming, Direc- 
 tions, Mixed Relations, Word Building, and Completion; then, 
 Logical Recollection and Substitution Whole; Knox; Tapping, and 
 Digit Span; Coordination; Logical Recognition. The Information 
 and Vocabulary tests were omitted because they showed no correla- 
 tion with the other tests. The Vocabulary test has an average 
 correlation with the other tests of .00, indicating chance relation- 
 ship. The correlations of Information with the other tests were not 
 worked out because inspection of the scores showed that approxi- 
 mately the same result would be obtained as for the Vocabulary 
 test. 
 
 On the whole, the inter-test correlations, although mostly posi- 
 
58 Psychological Examinations of College Students 
 
 CM/5 f5 a -<t T)-^ r-ce -^rOOO * * * \r>mO M o 
 
 h !?9999999 c !l99 ". o ! >: 9 <? o 9 
 
 I+I++++++++I + + I 11+ 
 
 
 S 8 
 
 uopn W sqn S _:_: j. ,' 
 
 ,'++ f + + 4. ++ f 
 
 8 
 
 \o O vO Oi 
 
 UOPTU80D3H 
 
 jora3 ^ p jo 
 
 UOP33HOD3H S 
 
 V 
 
 > 
 
 t^- T}- ooo Tt-vO O 00 o 
 
 . - " *! ^ ! 9 
 
 +++++++++++ + I + + ++++ ) 
 
 
 MOiOOMOi^POrO irtOO O 00 O M -<toO * O 
 
 Suipnng PJOM 19 r ? p J c ? t ?T t> ; ^09 1 9 *? 9^^^ 
 
 H +++++++++ +++ + + + I +++ ' ' 
 
 j~j M M roo O v) t^oo c<5Nioo O^or^Noooo Ov 
 
 rS suoi^Bia^H paxipv "99 h J1 r ? l ? r ? 1!?" 9 9 ^ "9^9 
 
 X I I ++++++ ++++ + I + ++++ 
 
 WfJOO 
 
 9 9 
 + + 4- 
 
 i^r~oo O f* O inM-^-t^ N * 1C 
 
 * M MPr * 
 
 j 
 
 ++++++ +--++++ +"+'+' +'+++ 
 
 OfOrOvOOOOOHOO>W5 tf) t^OO vONNTj- M N ^. 
 
 "f?^^ Nror^ciNfoiHN M O O OfONO PJ w<5 
 
 ++++ ++++++++ + + + f +'+"+' 
 
 OMM OiOOOOvO>t-0 t- * t^ a l^oo O N '" 
 
 f?^ 1 ^ NNNrOMNrONM C5 O MNMQ 
 
 ' ' 
 
 +++ +++++++++ + + ++ 
 
 UOpBUlpJOO3 
 
 S ff 
 
 m ^. v2 
 
 "o 
 
 00 O O O MOO row M l^N O N O r^ t^/5i-iO\ O >o *O 
 
 -trprowOqiiwMNi-iM M N w MNq"-" M M - 
 
 +++++++ I +++'+' + + + 4+1 \ .2 
 
 
 il1isffJig^lJlJi!ifl 8 si a 'hms B 
 
 i 
 
 >4w4diio6di6iw4 t o ^ ood6>* J03 ' ;3 >^' 5l *- Soi a 
 
Inter-Test Correlations and their Significance 59 
 
 tive, are low. This would indicate that we are testing here different 
 mental abilities. The fact that we can group certain tests together 
 on the basis of relationship shown by the correlation coefficients 
 further supports this view. It is possible to find several groups of 
 tests which correlate closely among themselves, but loosely with 
 the other tests. The following table gives the various groupings 
 with their correlations: 
 
 TABLE XXIV 
 
 GROUPING OF TESTS ON THE BASIS OF THEIR CORRELATION COEFFICIENTS 
 Group I. Coordination and Tapping, Correlation +.48 with each other. 
 
 Group II. Cancellation, Checking, Color Naming, Word Naming, Substitution. 
 
 Average Correlation of tests within group +-3 2 
 
 " u Cancellation with all others +.35 
 
 Checking " +.36 
 
 u u Color Naming " " " +.27 
 
 Word Naming " +.34 
 
 Substitution " " +.30 
 
 Group III. Directions, Opposites, Verb-object, Mixed Relations, Word Building, 
 
 and Completion. 
 
 Average Correlation of tests within group -H.3 2 
 
 " Directions with all others +.25 
 
 Opposites u " " +.40 
 
 Verb-object " " " +.31 
 
 Mixed Relations " " " +.40 
 
 Word Building " u " +.25 
 
 * Completion " " +.30 
 
 Group IV. Word Recollection, Word Recognition, Logical Recollection, Logical 
 
 Recognition. 
 
 Average Correlation of tests within group +.38 
 
 Word Recollection with all others +.39 
 
 Word Recognition +.37 
 
 " " Logical Recollection " u " +.40 
 
 tt " Logical Recognition " u u +.35 
 
 Group V. Information and Vocabulary. 
 
 Miscellaneous: Digit Span, Knox Cube. 
 
 Thus Tapping and Coordination correlate + 48 with each other, 
 but both tests show a much lower correlation with the other tests. 
 (The correlations outside of the group range from + -33 to + .01). 
 This agrees with Thorndike's theory that tests of the motor sensory 
 level correlate rather closely with each other, but only loosely with 
 tests of other levels. In Group II, Checking has an average corre- 
 lation of + .36 with the others of the group, and also a much lower 
 
60 Psychological Examinations of College Students 
 
 correlation with tests outside Group II, (ranging from + .30 to 
 - .04). Similarly, in Group III, Opposites and Mixed Relations 
 both have an average correlation of + 40 with the other tests in 
 this group, but a lower correlation with any test outside the group, 
 again conforming to Thorndike's contention that tests on the 
 associative level correlate closely with each other, but rather loosely 
 with tests on other levels. (The average correlation of Opposites 
 with the tests outside Group III is + .15; the average correlation of 
 Mixed Relations with tests outside Group III is + .10). In Group 
 IV, also, Logical Recollection has an average correlation of + .40 
 with the other tests in the group, but a lower correlation with any 
 test outside this group. (The correlations outside the group run 
 from + .30 to -1- .01). Information and Vocabulary differ from the 
 other tests of the series in that they are indicative of one's learning 
 rather than one's innate ability. There is only a chance correlation 
 between them and the other tests. A more detailed discussion of 
 this relationship we will postpone till the following section. As for 
 Knox Cube and Digit Span, perhaps the best plan is to consign 
 them to the miscellaneous class. Knox Cube shows on the whole 
 the closest correlations with the tests in Group II, but the average 
 group correlation is not high enough to warrant us definitely placing 
 it in this group rather than in Group IV. In like manner, aside from 
 its surprisingly high correlation with Cancellation (+ .77), Digit 
 Span shows no close relationship with any other test. If we omit 
 these four tests, (namely, Information, Vocabulary, Knox Cube, 
 and Digit Span), we do get very definite groupings of the other 
 tests, as shown in Table XXII above, indicating that we are measur- 
 ing different abilities. The rather high intercorrelations between 
 the tests of each group, together with their low correlations with 
 tests outside their own groups would support this view. There is 
 no evidence from these results to support Spearman's theory that 
 correlations are produced between all sorts of performance, the 
 amount of the correlation being simply proportional to the extent 
 that the performances concerned involve the use of a general com- 
 mon factor or "general ability." Our data give evidence neither of 
 a common factor nor of a hierarchial arrangement of the correla- 
 tions. Attempts to arrange the correlations to form a hierarchy 
 met with even greater failure than Simpson reports. 
 
 The simplest and clearest way to explain the existing relation- 
 ships between our tests seems, therefore, to arrange them in the 
 groups indicated in Table XXIV a grouping supported by the 
 
Inter-Test Correlations and their Significance 61 
 
 * Ji S> Sj 3 * ) ? ) N S IT 
 
 r ++++++++++++ + r + + ' 
 
 S CO 
 
 f 
 
 
 II|P|||^SSP|S|*||||| 
 
 lllllllilllHtifllilll 
 
 S^asSsfilSlliall I I ^ s 
 
 u 
 
 od 6, 
 
62 Psychological Examinations of College Students 
 
 actual correlation coefficients. The tests within each group seem 
 to be closely related to each other because they possess elements in 
 common elements serving to bind them closely to each other, but 
 loosely to tests without their own groups. Thus, Group I involves 
 motor capacity and skill; Group II powers of perception and com- 
 prehension; Group III associational relations; Group IV pure 
 memory. Though there is some slight overlapping in the qualities 
 called into play in the various groups, nevertheless it is not sufficient 
 to spoil our classification. 
 
 Table XXV gives the inter-test correlations corrected for attenua- 
 tion. The correlations are all higher but show in general the same 
 relationship. They range from + i.oo (Cancellation and Digit 
 Span; Word Recollection and Word Recognition; Word Recollec- 
 tion and Logical Recollection) to + .00 (Tapping and Word Recol- 
 lection; Mixed Relations and Word Recollection). When the corre- 
 lations are corrected for attenuation, Logical Recollection shows 
 the highest average correlation ( + .39) with the other tests (Omit- 
 ting Information and Vocabulary). Then, in order, Word Naming; 
 Substitution, Word Recollection and Cancellation; Opposites, 
 Verb Object and Word Building; Checking, Directions and Mixed 
 Relations; Completion and Color Naming; Word Recognition and 
 Logical Recognition; Coordination, Digit Span, Knox and Tapping. 
 
 The corrected coefficients of correlations also support the group- 
 ings of tests given in Table XXIV. It is possible to arrange the 
 attenuated correlations in the same groups as those given by the 
 raw correlations. The corrected coefficients of correlation are higher 
 than the raw correlations but the relationship between the tests is 
 similar. 
 
 To determine the reliability of the test scores, an investigation 
 was conducted three years after the testing of the first group of one 
 hundred freshmen (Group I). Two trials of the tests were given 
 to a group of 45 freshmen during the period extending from March 
 14 to May 15, 1919, inclusive. The two trials occurred in every 
 case on the same day and required approximately 45 minutes of the 
 student's time. Table XXVI gives a list of the tests employed in 
 two trials. 
 
 The method of procedure in conducting these tests with the 45 
 freshmen was the same as that employed with the 200 freshmen 
 in Groups I and II. Moreover, all the tests were conducted individ- 
 ually just as was done in testing the freshmen in Groups I and II, 
 and the room employed for the testing was the same as in the former 
 
Inter-Test Correlations and their Significance 63 
 
 TABLE XXVI 
 
 1. Coordination Trials I and 2 identical, same as with Groups I and II. 
 
 2. Tapping Trials I and 2 identical, same as with Groups I and II. 
 
 3. Cancellation First half of Woodworth- Wells' blank used in Trial I, 
 
 and second half in Trial 2. 
 
 4. Checking First half of Woodworth- Well's blank used in Trial I, 
 
 and second half in Trial 2. 
 
 5. Color Naming Trials I and 2 identical. 
 
 6. Directions Woodworth- Wells' blank used in Trial I ; Wells' alterna- 
 
 tive form used in Trial 2. 
 
 7. Opposites f The first half of each of these Wood worth- Wells' 
 
 8. Verb-object { blanks was used in Trial I, and the second half in 
 
 9. Mixed Relations [Trial 2. 
 
 10. Word Building Letters a e i I p r used in Trial I. (Same as in groups 
 
 I and II). Letters a e o b m t used in Trial 2. 
 
 11. Word Naming Trials I and 2 identical. 
 
 12. Knox Cube Trials I and 2 identical. 
 
 13. Digit Span Trial I as in Groups I and II; equivalent form used 
 
 in Trial 2. 
 
 14. Word Recollection / 
 
 15. Word Recognition I Trial I the same as in Groups I and II; equivalent 
 
 16. Logical Recollection 1 Mulhall form used in Trial 2. 
 
 17. Logical Recognition * 
 
 1 8. Substitution Given only once. (The closeness with which the cor- 
 
 relations of the first half of the test with the other tests 
 agreed with the correlations of the second half of the 
 test with the other tests, measures the reliability of 
 this test.) The correlation between the score in the 
 first half of the blank and the score in the second half 
 of the blank was taken as the measure of reliability. 
 
 19. Completion Given only once. The correlation between the score in 
 
 the odd numbered sentences and the score in the even 
 numbered sentences was taken as the measure of 
 reliability. 
 
 investigations. Just as we found the average and P.E.'s for the 
 various tests to be approximately the same for both groups I and II, 
 so the norms for this group of 45 freshmen are approximately the 
 same as those obtained for Groups I and II. Thus, since one group 
 of Barnard freshmen appears very similar to any other group of 
 Barnard freshmen selected at random, we may fairly assume that 
 the coefficients of reliability secured with any one group will also be 
 indicative of the relationship that would exist between two trials 
 with any other group selected at random. If, then, we find the 
 reliability of the tests high for this group of 45, it is fair to judge 
 that it would have been equally high with the group of 100 fresh- 
 
64 Psychological Examinations of College Students 
 
 men, (Group I), whose test scores were used in computing the 
 correlations given in Table XXIII. 
 
 TABLE XXVII 
 
 TEST CORRELATIONS BETWEEN TRIAL i AND 
 TRIAL 2 GROUP OF 45 FRESHMEN 
 
 1. Coordination +.66 
 
 2. Tapping +.77 
 
 3. Cancellation . +.60 
 
 4. Checking +.88 
 
 5. Color Naming +.88 
 
 6. Directions +.76 
 
 7. Opposites +.79 
 
 8. Verb-object +.70 
 
 9. Mixed Relations +.60 
 
 10. Word Building +.70 
 
 11. Word Naming +.71 
 
 12. Knox Cube +.69 
 
 13. Digit Span +.83 
 
 14. Word Memory Recollection +.18 
 
 15. Word Memory Recognition +.33 
 
 1 6. Logical Memory Recollection +.48 
 
 17. Logical Memory Recognition +-73 
 
 1 8. Substitution +.70 
 
 19. Completion +-77 
 
 Table XXVII shows the correlation between the first and second 
 trial for each of the 19 psychological tests. With three exceptions 
 Word Recollection (+- .18), Word Recognition (+- .33), and Logical 
 Recollection (+- .48) the correlations are high enough to indicate 
 a high degree of reliability. These reliability correlations range 
 from -+ .88 in the case of checking and Color Naming to + .60 in 
 the case of Cancellation and Mixed Relations. If we disregard 
 Word Recollection, Word Recognition, and Logical Recollection 
 on the ground that their low reliability coefficients suggest that 
 their correlations with the other tests do not give us an exact 
 measure of the existing relationship, we have remaining a series of 
 1 6 reliable tests. The inter- test correlations based upon the scores 
 in these 16 tests are accurate indicators of the true relationship 
 existing between these tests. Our conclusions drawn from these 
 inter-test correlations are, moreover, strengthened by our knowledge 
 that they are based upon reliable test scores which give an accurate 
 measure of the freshman's ability in these tests. 
 
SECTION VII 
 
 CORRELATIONS BETWEEN THE TESTS AND 
 ACADEMIC MARKS 
 
 TESTS VERSUS MARKS AS MEASURES OF MENTAL 
 
 ABILITY 
 
 The charts discussed in Section V showed that the freshman 
 scores in the psychological tests were distributed according to the 
 normal probability curve. Tables XXVIII to XXXII inclusive, 
 show the distribution for the five groups of academic marks, based 
 on grades of freshmen in Group I. 
 
 TABLE XXVIII TABLE XXIX TABLE XXX 
 
 LANGUAGE MATHEMATICS SCIENCE 
 
 Grade Frequency Frequency Frequency 
 
 F (50-60) 2 i 4 
 
 D (60-70) 14 14 6 
 
 C (70-80) 49 33 1 6 
 
 B (80-90) 30 30 12 
 
 A (90-100) 2 10 3 
 
 TABLE XXXI PHILOSOPHY TABLE XXXII HISTORY 
 
 Grade Frequency Frequency 
 
 F (50-60) o o 
 
 D (60-70) I 4 
 
 C (70-80) 10 1 6 
 
 B (80-90) 12 4 
 
 A (90-100) 4 2 
 
 Not only is there a coarse grouping (only five units) as compared 
 with the fine grouping of scores in the various psychological tests 
 (15 to 20 units), but the distributions fail to follow the normal error 
 curve as is the case in the test scores. With the academic marks 
 there is a decided skewing of the distribution curves toward the 
 good or positive end. It seems as though instructors made a delib- 
 erate effort to avoid failing their students. As for the passing grades, 
 inspection of the marks suggests that there is insufficient care in 
 rating students according to their relative abilities in various courses. 
 
 Observation of the uniform surfaces of frequency obtained when 
 these one hundred freshmen were given the twenty-three psycho- 
 logical tests, compared with the decidedly skewed distributions for 
 the same students in academic marks, prepares us for correlation 
 tables XXXIII and XXXIV. 
 
66 Psychological Examinations of College Students 
 
 Table XXXIII shows the correlation between the scores of all 
 the psychological tests (excluding Information), and the marks in 
 each of the five academic groups for the freshmen in Group I. 
 Language shows a fair positive correlation with Mixed Relations 
 (+ .20), Word Building (+ .31), Completion (+ .30), and Vocabu- 
 lary (+ .41), i. e. t with the tests in which the language factor per- 
 forms a significant role. Mathematics shows a fair positive correla- 
 tion with Cancellation (+ .28), Checking (+ .22), tests involving 
 simple mathematical processes, and Knox Cube (+ .24). Science 
 shows positive correlations with Opposites (+ .33), Verb-object 
 (+ .23), Mixed Relations (+ .30), tests involving the higher 
 thought processes needed in understanding the science courses 
 given at Barnard, Knox Cube (+ .34), a test involving powers of 
 perception and observation which are necessary in scientific labora- 
 tory work, and Logical Recollection (+ .21), which is also an im- 
 portant factor in scientific work. 
 
 The correlations of Philosophy with Cancellation + .37, Word 
 Naming (+ .29), Knox Cube (+ .28) and Digit Span (+ .22) are 
 unexpected. 
 
 TABLE XXXIII 
 CORRELATIONS BETWEEN TESTS AND ACADEMIC RECORDS 
 
 Language Mathematics Science Philosophy History 
 
 Coordination .12 +.05 . .03 +.03 +.15 
 
 Tapping .16 +.01 -.10 +.15 +.00 
 
 Cancellation v -+.14 +.28 +.04 +.37 +.10 
 
 Checking .01 +.22 +.06 +.10 +.02- 
 
 Color Naming +.11 +.07 +.12 .07 .05 
 
 Directions +.03 -.10 -.03 -.22 +.13 
 
 Opposites +.17 -.01 +.33 +.01 +.30 
 
 Verb-Object +.04 +.03 +.23 +.17 .05 
 
 Mixed Relations +.20 +.01 +.30 +.12 +.19 
 
 Word Building +.31 +.15 +.00 .17 +.24 
 
 Word Naming +.10 +.06 +.02 +.29 +.09 
 
 Knox Cube +.18 +.24 +.34 +.28 +.08 
 
 Digit Span +.19 +.19 +.05 +.22 +.33 
 
 Word Memory Recollection . . .01 .23 .07 .27 .03 
 
 Word Memory Recognition . . +.06 +.02 +.12 +.10 +.13 
 
 Logical Memory Recollection . . +.13 +.13 +.21 -.03 +.40 
 
 Logical Memory Recognition . . .03 +.06 +.03 .08 +.02 
 
 Substitution ist Half . . . . . -.08 +.11 +.09 -.19 +.18 
 
 Substitution 2nd Half -.05 +.08 +.06 -.14 +.26 
 
 Substitution Whole .10 +.11 +.00 .19 +.14 
 
 Completion +.30 +.02 +.05 +.17 +.14 
 
 Vocabulary +.41 -.05 +.12 +.09 +.23 
 
The Tests and Academic Marks 67 
 
 TABLE XXXIV 
 CORRELATION BETWEEN TESTS AND INTELLIGENCE QUOTIENT 
 
 Intelligence 
 Quotient 
 
 Coordination +.18 
 
 Tapping +.17 
 
 Cancellation +- 22 
 
 Checking +.20 
 
 Color Naming +.23 
 
 Directions +.20 
 
 Opposites +.24 
 
 Verb-object +.23 
 
 Mixed Relations +.20 
 
 Word Building +.22 
 
 Word Naming +.26 
 
 Knox Cube +.22 
 
 Digit Span +.16 
 
 Word Memory Recollection +.14 
 
 Word Memory Recognition ~Ki7 
 
 Logical Memory Recollection +.23 
 
 Logical Memory Recognition +.18 
 
 Substitution First Half +.27 
 
 Substitution 2nd Half +.25 
 
 Substitution Whole +.27 
 
 Completion +.21 
 
 Vocabulary +.03 
 
 History shows positive correlations with Opposites (+ .30), 
 Word Building (+ .24), Digit Span (+ .33), Logical Recollection 
 (H- .40), and Substitution (+ .26), i. e., with the tests involving 
 ability to memorize logical material and ability to perceive rela- 
 tionships between facts two essentials for successful performance 
 in the required first-year history course at Barnard. 
 
 In general, then, the five academic groups show positive corre- 
 lation with tests which we would expect to correlate with them. 
 Table XXXIV gives the correlations between the tests and the 
 composite score of all the academic groups. The correlations are 
 all positive, ranging from + .14 to + .27 (excluding Vocabulary), 
 suggesting a positive relationship. They are, however, too low to be 
 used for diagnostic purposes. Aside from a few correlations in Table 
 XXXIII previously mentioned, the correlations between the 
 various tests and each of the five academic groups are even less 
 susceptible to use for practical purposes. 
 
 In view of these low correlations and the wide variation in corre- 
 
68 Psychological Examinations of College Students 
 
 lations obtained between tests and marks by other experimenters, 
 the question arises: Do the academic marks or the psychological 
 tests give the more reliable estimates of the student's mental 
 ability? The present writer believes that the psychological tests 
 give the more adequate measures. 
 
 What meager experimental data there is relevant to this question 
 of the reliability of school marks, corroborates this view. The 
 skewed distributions in the case of the Barnard academic grades 
 were indicated before a fact which has been noted by investigators 
 in the case of other institutions. 47 
 
 Professor Max Meyer, 48 making a statistical study of all the 
 marks of forty instructors given during a period of five years at the 
 University of Missouri, found a striking lack of uniformity in the 
 standards of grading used. So striking was the non-uniformity 
 that the college authorities were moved to establish a definite 
 system of marking in 1908, with the aim of overcoming the ten- 
 dency of the instructors to distribute grades according to personal 
 opinion. Following Meyer, a study of the distribution of marks 
 at the University of Wisconsin was made by Dearborn, 49 and of 
 the marks at Harvard University and the University of California 
 by Foster. 50 These, and studies made at the University of Chicago, 
 Amherst College, and Columbia University, agreed in showing the 
 same wide variation in the standards of grading employed by 
 instructors. 
 
 Aikins 51 found a slight difference in the relative positions assigned 
 to 17 students in a philosophy class by the students themselves on 
 the basis of several ten-minute tests, and the positions he assigned 
 them on the basis of four hour tests. Smith gives several plates, 
 illustrating clearly the great discrepancies and marked lack of 
 uniformity in marking systems at the University of Iowa. 52 
 
 Zerbe, in a detailed study of the distribution of grades assigned 
 for academic work and those assigned for shop work at the School 
 of Applied Industries, Carnegie Institute of Technology, found 
 that the grades as distributed for the shop work were based on a 
 much lower standard than the grades assigned for the theoretical 
 
 " Kelly in a monograph entitled "Teachers' Marks" has given a history of the standards of 
 marking in elementary schools, high schools, and colleges. 
 
 Meyer, Max. The Grading of Students, Science, 28; 243-252. 
 
 Dearborn, W. F. School and University Grades. 
 
 50 Foster, William T. Scientific vs. Personal Distribution of College Credits; Popular Science 
 Monthly, 78; 378-408. 
 
 si Aikins, H. A. The Reliability of "Marks," Science, N. S., 1910, 32; 18-19. 
 
 " Smith, A. G. A Rational College Marking System, Journ. of Educ. Psychol., 1911. 2; 383-393- 
 
The Tests and Academic Marks 69 
 
 subjects. 63 He also observed a marked lack of conformity to a 
 standard in the case of grades given by individual instructors. 
 When Jones M gave an opposites test and a memory test to each of 
 two elementary psychology classes, taught by different instructors, 
 he obtained these interesting results: 
 
 Instructor "A" Instructor "B" 
 
 (28 students) (33 students) 
 
 Class standing and opposites .... .09 .49 
 
 Class standing and memory .44 .07 
 
 These correlations were explained when further investigation 
 revealed that instructor A taught by the outline method, emphasiz- 
 ing the memory factor, whereas instructor B discouraged verbatim 
 statements taken from the text book. Both instructors were teach- 
 ing the same subject, but assigning grades according to entirely 
 different standards. 
 
 After an exhaustive study of the question at Harvard and other 
 institutions, President Foster of Reed College concluded that 
 "Not only are there extreme variations among different courses, 
 but there are variations in the same course from year to year that 
 cannot be accounted for, apparently, by any of our scientific studies 
 in the distribution of abilities among human beings. From Maine 
 to California the administration of college credits, although alike 
 in no other particular, agrees in this: "That its basis is personal 
 rather than scientific." 55 Recognition of this personal equation 
 factor has led Smith, Weiss, 56 Zerbe, Foster, Starch, and other 
 investigators to emphasize the need of a uniform system of grading. 
 They agree, moreover, in maintaining that the distribution of 
 college grades, when properly assigned, should conform to the 
 normal probability curve. In 1914, a committee on standardizing 
 grades at George Washington University made a similar proposal. 
 Definite attempts to enforce such systems of marking are now 
 being used at the University of Missouri, Reed College, and other 
 institutions. 
 
 Even in a more restricted and more objective situation when 
 instructors are asked to assign grades according to performance in 
 a definite task as for example, in a written examination paper, 
 there is great variability due to the widely different subjective 
 
 61 Zerbe, J. L. Distribution of Grades. Journ. of Educ. Psychol., 1917, 9; 575-588. 
 " Jones, E. S. A Suggestion for Teacher Measurement. School and Society, 1917, 6; 321-322 . 
 Zerbe, J. L. Distribution of Grades. 
 
 M Weiss, A. P. School Grades To what Type of Distribution shall they Conform? Science , 
 1912, 36; 403-407. 
 
70 Psychological Examinations of College Students 
 
 standards employed by the teachers in judging. 57 Jacoby found a 
 variation of 1.5 points out of 10 in the grades of six professors of 
 astronomy in marking eleven astronomy papers. 58 Starch and Elliott 
 had facsimile reproductions made of two first-year English papers 
 and a geometry paper, printed on the same kind of paper the 
 students had written them on. 59 These they then had rated by 
 142 high school teachers of these two subjects. The English papers 
 were also rated by a class in the Teaching of English in the Univer- 
 sity of Wisconsin and by a Summer School class of teachers in the 
 University of Chicago. They found that the grades assigned to the 
 two English papers by 142 English teachers ranged in the case of 
 one paper from 64 to 98, with a probable error of 4.0, and in the 
 case of the other from 50 to 98, with a probable error of 4.8. The 
 grades of the mathematics paper assigned by 118 mathematics 
 teachers ranged from 28 to 92, with a probable error of 7.5 points. 60 
 
 In a later investigation Starch had ten college freshman English 
 papers graded independently by ten instructors of the various 
 sections of freshman English. 61 He found as wide a range of marks 
 as he obtained with the English and Mathematics papers of his 
 former investigation. Moreover, when ten papers were regraded by 
 the same instructor after a certain interval of time, Starch found 
 an average difference between the first and second grading of 4.4 
 points. He also found a mean variation of the grades assigned by 
 teachers in different schools of 5.4 points, by teachers in the same 
 department and institution of 5.3 points, and of grades assigned at 
 different times by the same teachers to their own papers of 2.2 
 points. On the basis of all his data, he concluded that the best 
 marking scale is 100, 95, 90, 85, 80, etc., and that the distribution 
 of grades should follow the probability curve. 
 
 All the studies thus far made in this field indicate this same 
 variation in standards of grading. There are, moreover, additional 
 factors which render school marks absolutely unreliable measures 
 of a student's mental ability, and cause low correlations between 
 psychological tests and marks. 
 
 57 For illustrations of the variability of Civil Service examiners in rating the same papers, the 
 variation between the marks of teachers in New York State on the one hand, and the Regents on 
 the other, see Kelly's monograph. 
 
 48 Jacoby, H. The Marking System in the Astronomical Course at Columbia College, 1900- 1910, 
 Science, 31; 819. 
 
 59 Starch and Elliott, Reliability of Grading High School Work in English, School Review, 
 September, 1912. 
 
 60 Starch and Elliott, School Review, 21, 254-259. 
 
 81 Starch, D. The Reliability and Distribution of Grades, Science, 1913, 38; 630-636. 
 
The Tests and Academic Marks 71 
 
 James, from work done at Whitewater Normal School, gives 
 these three reasons for the low correlations obtained by him: 62 
 
 "i. The reluctance of nearly all teachers, and their inability 
 because of the limitations of our poor rating methods, to rate the 
 good students as high as they should be rated, or the poor ones as 
 low as they should be rated." 
 
 "2. The rather closer application to their studies made by the 
 less able, due to greater anxiety and more time at their disposal." 
 
 "3. The easy-going satisfaction displayed by many able minds 
 content with what is for them mediocre accomplishment, and the 
 greater drain on their time imposed by fellow-students for outside 
 activities of all kinds." 
 
 From data obtained from a questionnaire sent to 127 delinquent 
 college freshmen and to their high school principals, Miner con- 
 cluded that such traits as "lack of purpose, laziness, and lack of 
 resistance to social and other distractions" often explain a student's 
 failure in school work. 63 Their marks in such cases are unreliable 
 measures of their ability. Scott manifested agreement with Miner 
 when he stated that: "Where students stood high in the tests, but 
 low or medium in estimates, their failure to succeed in class work 
 was usually due to laziness, timidity, or disgust for the idea of 
 struggling for marks." 64 
 
 Abundant statistical evidence, therefore, supports our conten- 
 tion that the striking lack of uniformity in standards of grading 
 among instructors, making for skewed distributions of marks, the 
 differences in grades assigned the same paper by teachers at dif- 
 ferent times, the personal equation in marking, the tendency of 
 many able students to neglect studies for outside distractions and 
 of poorer students to apply themselves. more assiduously, the role 
 played by such factors as lack of purpose or incentive, interest in 
 outside or in college activities, economic pressure causing students 
 to devote much time to earning money, etc., make college marks 
 totally inadequate measures of students' ability. All these factors 
 are influential, moreover, in making Barnard marks as unreliable 
 as marks given in other colleges. No attempt is made by Barnard 
 instructors to distribute their grades according to the normal 
 
 M James, B. B. Mutual Correlations of Intelligence, Scholarship, and Vocabulary. School & 
 Society, 1919, 9J 427. In School & Society, 1918, 7; 238-239, James gives similar factors as 
 influencing the correlations between marks and tests. 
 
 68 Miner, J. B. The College Laggard. Journ. of Educ. Psychol., 1910, i; 263-271. 
 
 "Scott, C. A. General Intelligence or "School Brightness." Journ. of Educ. Psychol., 1913, 
 41 500-524. 
 
72 Psychological Examinations of College Students 
 
 probability curve. Absolute freedom is permitted the teachers. 
 As a result, the personal bias of the teachers plays a large part in 
 the marks received by students. This, combined with the con- 
 tributory causes above mentioned, renders Barnard marks untrust- 
 worthy. 
 
 The psychological tests, on the other hand, have much to recom- 
 mend them as giving reliable estimates of freshmen's mental 
 ability. All the tests employed are standard tests. They were, 
 moreover, administered by one experimenter according to a care- 
 fully standardized method of procedure. All conditions were kept 
 constant the place of testing, the attitude of the experimenter, 
 the method of conducting the tests, and the method of scoring. 
 Every student undertook the examination with a determination to 
 do her level best. Whereas, in school subjects, lack of interest or 
 incentive often caused a girl to do a lower grade of work than she 
 was mentally capable of doing, here there was a definite incentive 
 impelling her to exert maximum effort. Each freshman expected 
 to receive vocational guidance based on her test scores. She accord- 
 ingly took the psychological test at an hour convenient for her 
 when she was feeling in good condition. Genuine interest in the 
 tests, (noted in the case of all students), coupled with a keen desire 
 to make a favorable record, renders their test scores reliable esti- 
 mates of their ability. The fact that the scores conform to normal 
 distribution curves further indicates the reliability of these measures, 
 
 We do^not claim, however, that we can_predict_a_student's 
 future_success_in_college from her psychological test record. The 
 psychological examination gives an adequate measure of what 
 each freshman can do. From it we can make an authentic psycho- 
 graph of her mental abilities indicating in which processes she is 
 strong, and in which she is weak. Whether she will make high 
 academic grades or attain success in later life depends not only 
 upon her mental capacity, but upon such other factors as interest, 
 incentive, will-power, economic stress, environmental conditions, 
 etc. The tests, not her academic marks, measure her mental capac- 
 ity; to predict her future performance in school or her success in 
 a particular vocation, we must also consider these other factors. 
 
SECTION VIII 
 
 CORRELATIONS BETWEEN PSYCHOLOGICAL TESTS 
 
 AND PHYSICAL MEASUREMENTS. THEIR 
 
 SIGNIFICANCE 
 
 There is one further problem to be considered the relation 
 existing between the psychological tests and the physical measure- 
 ments. The correlations shown in Table XXXIII, based on the 
 records of the one hundred freshmen in Group I, furnish an im- 
 portant contribution to our existing meagre data on this subject. 
 
 Most investigators who have hitherto reported correlations 
 between physical traits and mental ability have used school marks 
 or teachers' estimates as indicators of mental ability. Their sub- 
 jects, moreover, have been school children. Porter, Smedley, 
 De Buck, MacDonald, Gilbert, Baldwin, Pyle, King, Arnold, 
 Wilson, and Schuyten are some of the chief workers in this field. 
 Widely varying results have been reported, some experimenters 
 finding positive correlations between physical traits and school 
 progress, others negative, and still others indifferent or zero corre- 
 lations. Discussing the significance of these varying correlations, 
 Whipple says: "The trend of evidence is to the effect that all such 
 correlations, where found, are largely explicable as phenomena of 
 growth, i. e., as correlations with relative maturity. This makes 
 intelligible the fact that, in general, the positiveness of all such 
 correlations lessons with age, and that many of them, indeed, 
 become difficult or impossible of demonstration in adults." 65 
 
 Of the investigations in which adults have been used as subjects, 
 the work of Dr. Karl Pearson is perhaps the most extensive. He 
 made measurements of 1,000 Cambridge University students, 
 obtaining these correlations: 
 
 Mental ability and dolichocephaly . . . . +.03 =*=.<>3 
 Mental ability and short heads . . . . .08 =^.03 
 Mental ability and broad heads . . . . +.04 .03 
 
 His method of rating his subjects for mental ability was extremely 
 rough, consisting merely in grouping the men into two big classes 
 pass men and honor men. Similar correlations obtained by Pearson 
 between head measurements and mental ability as measured by 
 
 Whipple, G. M. Manual of Mental and Physical Tests. Part I, p. 71. 
 
74 Psychological Examinations of College Students 
 
 teachers in the case of 1856 school boys twelve years of age, lead 
 Galton to conclude "that there is no marked correlation between 
 ability and shape or size of the head." 66 
 
 In another investigation with Cambridge students, Pearson 
 found zero correlations between mental ability, determined roughly 
 as indicated above, and strength of pull, strength of squeeze, long 
 sight, weight, and ratio of weight to stature. 67 Continued testing 
 of Cambridge students and school children lead Pearson to conclude 
 in 1906 that "The results (of our investigations) confirm the previous 
 conclusion that: While there exists a slight but sensible relation 
 between size of head and intelligence, there is no possibility of 
 using this relation to make even rough individual predictions." 68 
 
 These investigations, although interesting, have no direct bear- 
 ing upon our problem, however, which is concerned with the rela- 
 tionship existing between the performance of college freshmen in 
 psychological tests and their physical measurements taken in the 
 gymnasium. 
 
 We have good reason to feel that these physical measurements are 
 fully as reliable and accurate estimates as are the psychological test 
 scores. The physical examinations were all conducted in the 
 Thompson Gymnasium of Teachers College. They were given 
 individually, the head of the Department of Physical Education 
 of Barnard College making all the measurements. These were then 
 immediately recorded on the student's physical record card by an 
 assistant. Thus any inaccuracy in taking the measurements would 
 be a constant one, and would not disturb the relative ranking of the 
 freshmen. 
 
 Experimental conditions were as uniform as in the case of the 
 psychological tests. Each girl came to the gymnasium at an hour 
 convenient for her and went through all parts of the examination 
 according to a standardized method of procedure. No clothing was 
 worn during the examination, save for two light cloth flaps which 
 were fastened loosely about the shoulders by means of a draw 
 string and two similar flaps fastened about the waist which could 
 easily be raised in taking measurements. These were provided by 
 the physical director for the occasion. 
 
 Pearson, K. On the Correlation of Intellectual Ability with the Size and Shape of the Head . 
 Proc. Roy. Soc. 1902, LXIX, 333-342. 
 
 Lee, A., Lewenz, M. A., and Pearson, K, On the Correlation of the Mental and the Physical 
 Characters in Man. II Proc. Roy. Soc., 1002, LXXI, 106-114. 
 
 68 Pearson, K. On the Relationship of Intelligence to Size and Shape of Head, and to other 
 Physical and Mental Characters. Biometrika. 1006, 5; 105-146- 
 
Psychological Tests and Physical Measurements 75 
 
 The physical records taken were: height measured in centi- 
 meters with a stadiometer; weight, measured in pounds with the 
 Fairbanks scale; lung capacity, measured in cubic centimeters; and 
 four other strength tests grip right and left hand, upper back and 
 chest, measured in kilograms with a dynamometer. The norms for 
 these measurements obtained for these one hundred freshmen were 
 given in Section V. 
 
 The curves of distribution for these seven measurements (which 
 lack of space prevents us from printing), conform approximately 
 to the normal probability curve. The subjects, moreover, with a 
 very few exceptions, were all eighteen years of age or over, so that 
 the factor of relative maturity does not affect the correlations. The 
 freshmen are a rather homogeneous group with respect to age. 
 These facts, coupled with the accuracy of both the physical and 
 psychological measures give us good reason to believe in the reliabil- 
 ity of the correlations in Table XXXIII. 
 
 It is interesting to note that six of the seven physical measure- 
 ments all except lung capacity manifest zero or chance correla- 
 tions with all the psychological tests. The average correlation of 
 each of these six measures with all the psychological tests is as 
 follows: Height with all the tests, -f- .05; weight + .06; strength 
 of grip, right hand, + .04; strength of grip, left hand, + .02; 
 strength of upper back, + .02, and strength of chest, + .05. As 
 these correlations are all less than the probable error ( .068) 
 they indicate clearly that there is no connection between these 
 physical measurements and a freshman's mental ability as indicated 
 by her psychological test records. In the case of lung capacity, all 
 the correlations (except with vocabulary) are positive. They are 
 markedly low, though, the average correlation between lung capac- 
 ity and all the psychological tests being only + .10. This is little 
 more than the probable error, indicating the existence of only a 
 chance relationship. 
 
 The uniformity of the single correlations in exhibiting this 
 tendency toward chance relationship is significant. In only eight 
 cases out of the total number of 154 correlations, or, in fact, we 
 might say in only six cases, since the correlations between Substitu- 
 tion First-half and lung capacity (+ .20) and Substitution Second- 
 half and lung capacity, (+ .26) duplicate information yielded by 
 the correlation between Substitution Whole and lung capacity 
 (+ .24) are there correlations of + .20 or over. The highest 
 correlation is only + .26 (Substitution Second-half and lung capac- 
 
76 Psychological Examinations of College Students 
 
 ity), which is too low to admit of diagnostic purposes. With these 
 few exceptions, all the correlations between physical measurements 
 and the tests 146 correlations in all show approximately zero 
 relationship. The large number of these correlations justifies us 
 
 TABLE XXXV 
 CORRELATIONS BETWEEN TESTS AND PHYSICAL MEASUREMENTS 
 
 1. Coordination . 
 
 2. Tapping . . 
 
 3. Cancellation 
 
 4. Checking . . . . '. . 
 
 5. Color Naming .... 
 
 6. Directions ' 
 
 7. Opposites 
 
 8. Verb-object 
 
 9. Mixed Relations .... 
 
 10. Word Building .... 
 
 11. Word Naming .... 
 
 12. Knox Cube 
 
 13. Digit Span 
 
 14. Word Memory Recollection 
 
 15. Word Memory Recognition 
 
 1 6. Logical Memory Recol- 
 
 lection +.18 +.22 +.22 -.05 +.01 +.10 +.09 
 
 17. Logical Memory Recogni- 
 
 tion +.09 +.ii +.14 -.05 +.04 -.01 +.05 
 
 18. Substitution First Half . +.19 +.05 +.20 +.02 +.07 .00 +.02 
 
 19. Substitution 2nd Half . . +.17 -.02 +.26 +.05 +.07 +.09 +.02 
 
 20. Substitution Whole . . +.19 +.00 +.24 +.04 +.07 +.06 +.01 
 
 21. Completion -.12 .05 +.04 -.02 .05 +.02 .03 
 
 22. Vocabulary -.02 +.07 -.17 +.07 .05 .04 .15 
 
 Average +.05 +.06 +.10 +.04 +.01 +.01 +.04 
 
 in concluding that the relationship between the physical measures 
 and the tests is one of chance only. 
 
 It is interesting to know that the only other experimenter who 
 has reported the results of a similar study with college freshmen 
 supports this view. Although Wissler in his study of the results of 
 the old Columbia freshman tests reports only two correlations 
 
Psychological Tests and Physical Measurements 77 
 
 between the physical tests and the psychological tests namely, a 
 correlation between length of head and logical memory of -f .21, 
 and between breadth of head and logical memory of - .05 the 
 observation of the records of freshmen in other physical tests com- 
 pared with their records in the psychological tests lead Wissler to 
 conclude : "That the physical tests show a general tendency to corre- 
 late -among themselves, but only to a very slight degree with the 
 mental tests." 69 
 
 Although the physical measurements exhibit only a chance 
 connection with a freshman's psychological test score, they should 
 be taken into consideration by an instructor or advisor whose duty 
 it is to give guidance to a student in planning her college course. 
 In Section V we pointed out the case of a freshman (Chart 3, G.S.), 
 whose net score in the psychological examination was well above 
 the average freshman record, but whose standing in academic work 
 was in the lowest quintile of the class. The fact that she made the 
 best record in the class in the physical measurements, together with 
 the information we later acquired concerning her athletic activities, 
 explained her academic failure. The more varied measures of a 
 student we have, the better qualified we will be to make an adequate 
 psychograph of a student's relative abilities and disabilities, in 
 various lines. 
 
 'Wissler, Clark. Psychological Review Monograph Supplement, June, 1901. 
 
GENERAL SUMMARY OF THE RESULTS WITH 
 
 SUGGESTIONS FOR THE PRACTICAL USE 
 
 OF THE TESTS 
 
 A series of nineteen psychological tests was given to two groups 
 of one hundred Barnard freshmen each with the aim first of estab- 
 lishing norms and standards of performance and giving students a 
 clear conception of their abilities and aptitudes along various lines 
 and second of determining the reliability of the tests and their 
 correlations with freshmen university grades and physical measure- 
 ments. 
 
 All the tests were given individually according to a standardized 
 method of procedure and under standard conditions. 
 
 The averages and surfaces of distribution for the first group of 
 one hundred freshmen (Group I) are approximately the same as 
 for the second group of one hundred (Group II) and for a third 
 group of forty-five freshmen showing that Barnard freshmen are 
 a homogeneous group, differing little from year to year. 
 
 The inter-test correlations range from + .77 (between Cancella- 
 tion and Digit Span) to .00 (between Tapping and Word Recollec- 
 tion and between Mixed Relations and Word Recollection). The 
 positive correlations between Cancellation and the other tests 
 (+ .03 to + .77) contradict the old compensation theory. The fact 
 that the correlations are all positive is suggestive of a definite 
 relationship between Cancellation and these various tests. 
 
 Checking and Word Naming show the highest average correla- 
 tion (+ .25) with the other tests (omitting Information, Vocabu- 
 lary, Word Recollection and Word Recognition); then, in order, 
 Opposites; Verb-object and Cancellation; Color Naming; Direc- 
 tions, Mixed Relations, Word Building, and Completion; Logical 
 Recollection and Substitution Whole; Knox; Tapping and Digit 
 Span; Coordination; Logical Recognition. 
 
 On the whole, the inter-test correlations, although mostly posi- 
 tive, are low, indicating that we are testing different mental abilities. 
 
 On the basis of the relationship shown by the correlation coeffi- 
 cients we may divide the tests into three groups: (i) motor tests 
 (Coordination and Tapping); (2) tests involving powers of per- 
 ception and comprehension (Cancellation; Checking, Color Nam- 
 
General Summary of Results 79 
 
 ing, Word Naming and Substitution) ; (3) tests involving associa- 
 tive relations (Directions, Opposites, Verb-object, Mixed Rela- 
 tions, Word Building and Completion); (4) tests which call into 
 play powers of learning, viz., observation and retention (Word 
 Memory and Logical Memory); (5) tests depending on the sub- 
 ject's knowledge more than on her innate ability (Information and 
 Vocabulary) ; (6) miscellaneous group (Digit Span and Knox Cube). 
 There is only a chance correlation between Information and Vocabu- 
 lary and the other tests. With the exception of this group and 
 Digit Span and Knox Cube, the remaining groups of tests correlate 
 closely among themselves but loosely with the other tests. 
 
 There is no evidence from these results of a general common 
 factor nor of a hierarchial arrangement of the correlations. 
 
 The tests within each group seem to be closely related to each 
 other because they possess elements in common elements serving 
 to bind them closely to each other but loosely to tests without 
 their own groups. 
 
 The coefficients of correlation corrected for attenuation are con- 
 siderably higher than the raw correlations but show in general 
 the same relationships. 
 
 The coefficients of reliability are low for Word Recollection 
 (-f .18), Word Recognition (+ .33) and Logical Recollection 
 (+ .48). For the other tests they range from + .88 (Checking 
 and Color Naming) to + .60 (Cancellation and Mixed Relations). 
 We have, thus, a series of sixteen reliable tests. Inter-test correla- 
 tions based upon the scores in these sixteen tests are accurate 
 indicators of the true relationship existing between these tests. 
 
 The psychological tests show low correlations both with each of 
 five academic groups (i) Language, (2) Mathematics, (3) Science, 
 (4) Philosophy and (5) History, and with the composite score of 
 all the academic marks (+ .14 to + .27). 
 
 Lack of uniformity in standards of grading among instructors, 
 causing skewed distribution curves of marks, the personal equation 
 in marking, the role played by such factors as lack of incentive, 
 interest in outside or college activities, economic pressure, etc., 
 make college marks inadequate measures of the students' ability. 
 
 There is evidence that the psychological tests give a true estimate 
 of each freshman's mental capacity. To predict her performance in 
 school or in a future vocation both her capacity and such other 
 factors as interest, incentive, will-power, environmental conditions, 
 etc., must be considered. 
 
8o Psychological Examinations of College Students 
 
 The correlations between the physical measurements and the 
 psychological tests show approximately zero or chance relationship. 
 
 Psychographic charts may be constructed, showing each student 
 her relative rank in the tests, academic grades and physical measure- 
 ments. Such psychographs may be put to practical use as for 
 example, in cases where a student is doing academic work of a grade 
 below the level her test record showed her capable of. 
 
 The results of this investigation make it possible to offer a few 
 tentative suggestions to college administrators who desire to insti- 
 tute a system of student guidance. The first step in such a plan 
 might well be to put each member of the freshman class through 
 a thorough physical examination to determine her physical fitness 
 for undertaking college work. This examination should be made 
 by the director of the Physical Education department or a com- 
 petent assistant in the department. Students with correctible 
 physical defects should be given proper treatment eyeglasses, 
 special physical exercises or what not, according to their needs. 
 Those suffering from a slightly run down condition might be advised 
 to take a light program until they regained their normal condition; 
 those too far below par might be advised not to enter college. 
 
 The second step might be to obtain an estimate of her mental 
 capacity on the basis of her score in a psychological examination. 
 A psychologist (who might also act as vocational advisor) with an 
 assistant might well be in charge of this work. If possible, each 
 freshman should be tested individually, the same experimenter 
 conducting all the tests according to a standard method of pro- 
 cedure. As for the particular tests to be used, they should be 
 varied in character, adapted to measure various mental abilities. 
 A series that may be divided into several groups, each group testing 
 a rather definite mental ability and such that tests within each 
 group correlate highly among themselves but loosely with all tests 
 outside their own group, as in the present investigation perhaps 
 represents the ideal type of tests. The particular series of tests 
 employed in this study is not, however, recommended as the best 
 series of tests that might be used. It is very probable that a series 
 could be found that will test more significant mental abilities and 
 such that the tests within each group will correlate more closely 
 with each other and more loosely with other tests. Only by empiri- 
 cally trying out different series can the ideal series be found. 
 
 Where lack of time or the size of the freshman class makes it 
 impossible to test each freshman individually, a comprehensive 
 
General Summary of Results 8 1 
 
 group test that has been found successful as for example, the 
 Army Alpha or the Thorndike Group test may be employed. 
 In view of the successful results secured with these group tests 
 and the speed with which they may be administered, it may well 
 be that such a comprehensive group test as the Thorndike test 
 would be the best to employ. In the case of students who barely 
 passed or who failed in this group test, such a series of tests as that 
 used in the present investigation might be used to supplement the 
 results of the group test. It would seem that a group test which 
 might be supplemented, where necessary, by an individual examina- 
 tion would be the ideal arrangement. 
 
 As we stated before, a psychologist and an assistant should pref- 
 erably be in charge of the psychological testing. Perhaps a group 
 of fifteen to twenty persons with some experience in scoring psy- 
 chological tests might be employed to score the tests immediately 
 after the psychologist has given them. In this way the examinations 
 might be easily scored within three or four days and the reports 
 made out for each student very soon after. The results of the 
 psychological examination and the physical examination together 
 with the student's academic entrance record, might then be sub- 
 mitted to the psychologist or vocational advisor. On the basis of 
 these records, psychographic charts might be made out for each 
 student indicating her strengths and weaknesses. The vocational 
 advisor might then have an immediate interview with such stu- 
 dents who showed any marked disabilities. In this personal con- 
 ference the advisor might try to obtain from the student pertinent 
 information concerning her interests, economic status, environ- 
 mental conditions, etc. All these supplementary items of informa- 
 tion would then enable him to form a comprehensive idea of the 
 student's mental and moral calibre. With this as a basis vocational 
 advice could be given the student regarding her choice of subjects, 
 study habits, participation in extra curricula activities, etc. Per- 
 haps such students might be asked to report at stated intervals for 
 further conference. Much the same procedure might be followed 
 with the other students except that here fewer conferences would 
 be necessary. 
 
 The advisor should be free to devote all his time to supervising 
 the academic career of the students and to rendering needed advice. 
 Obviously such a man should be a psychologist with both ability 
 to interpret the various measures secured of each student's ability 
 and tact in persuading students to follow his suggestions. From the 
 
82 Psychological Examinations of College Students 
 
 attempts that have thus far been made in certain institutions to 
 guide students' academic careers, it seems probable that with an 
 able vocational advisor aided by a competent assistant such a sys- 
 tem would be a distinct help in stimulating students to exert maxi- 
 mum effort in doing their college work. 
 
VITA 
 
 The writer was born in Norwich, Chenango County, New York, 
 on October I, 1895. She graduated from Girls' High School, Brook- 
 lyn. In 1916 she received the degree of Bachelor of Arts from 
 Barnard College, and in 1917 the degree of Master of Arts from 
 Columbia University. Her work in psychology has been done 
 under Professors R. S. Woodworth, H. L. Hollingworth, L. S. 
 Hollingworth, E. L. Thorndike and A. T. Poffenberger. She was 
 assistant in psychology in Columbia University 1918-20; assistant 
 in Summer School, 1918, 1919, and 1920; assistant and instructor, 
 Extension Teaching, 1918-19; and instructor, Extension Teaching, 
 1920. 
 
THIS BOOK IS DUE ON THE LAST DATE 
 STAMPED BELOW 
 
 AN INITIAL FINE OF 25 CENTS 
 
 WILL BE ASSESSED FOR FAILURE TO RETURN 
 THIS BOOK ON THE DATE DUE. THE PENALTY 
 WILL INCREASE TO 5O CENTS ON THE FOURTH 
 DAY AND TO $1.OO ON THE SEVENTH DAY 
 OVERDUE. 
 
 12Apr'55 
 
 apK 2 195S U 
 
 'D LD 
 
 MOV 7 
 
 MftVS 
 
 LD 21-100n-7,'33 
 
517566 
 
 UNIVERSITY OF CAUFORNIA LIBRARY