1061 57 1 The Transfer Effects of Practice IN Cancellation Tests MELVTN ALBERT MARTIN. A. M. REPRINT OF ARCHIVES OF PSYCHOLOGY isro. 32 SUBMITTED IN PARTIAt. FLTLiFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF JPHILOSOPHY IN THE FACULTY OF PHILOSOPHY. COLUMBIA UNIVERSITY NEW YORK 1016 The Transfer Effects of practice IN Cancellation Tests MELVIN ALBERT MARTIN. A. M. REPRINT OF ARCHIVES OF PSYCHOLOGY NO. 32 SUBMITTED IN PARTIAL FUI.FILMENT OF THE REQUIREMENTS FOR THE DEGREE OFDOCTOR O^ Pmi^OSOPHY IN THE FACUI-iTY OF PHlLOSOPli i: , COLUMBIA UNIVERSITY NE'W YORK 1015 Gift Th" Uuiverslty KAR ♦ 1916 TABLE OF CONTENTS Introduction Chapter I. Discussion of Previous Investigations 1 Chapter II. Description and Discussion of Materials and Procedure 23 Section 1. Preliminary Investigations 23 A. Reaction Time. B. Cancellation. C. Cancellation. Section 2. The Present Investigation 24 A. Practice Material. B. Test Material. C. Procedure. Section 8. Treatment of Data 28 A. Data of the Practice Series. B. Data of the Test Series. Chapter III. Interpretation of Results 46 Section 1. In Terms of Speed and Accuracy of Performance 47 Section 2. When Corrected for Errors 62 Section 3. Summary of Results 65 Section 4. General Conclusions 67 INTRODUCTION The problem of Formal Discipline, or the transfer effects of training, or the spread of special training, is no longer the question whether the special training of a function or a func- tion-group increases its efficiency with all sorts of materials and in all kinds of situations, but it is, rather, under what conditions, to what extent, and in what direction, may we expect training of a special kind to produce transfer effects. In planning the present investigation the author had these questions in mind. No one realizes more than he the difficul- ties involved in conducting and completing such an experiment under conditions that make for scientific accuracy. Such a research requires quite a large number of subjects divided into two groups — one for the control, to determine the improve- ment in the tests themselves, and one for practice. To secure a sufficient number of subjects for the control is compara- tively easy, but to secure a suitable practice group and hold them to systematic and prolonged practice under conditions that favor a maximum of measurable improvement from prac- tice period to practice period is not an easy task. The investigation is purposely limited to a narrow field in order to avoid possible sources of error due to complicating the procedure as well as the results by attempting too much in one experiment. It is a straightforward attempt to discover the transfer effects of prolonged practice in cancelling in one situation upon cancelling in other situations. The materials in both the practice series and the test series, as well as the methods of procedure were such as to insure exact quantita- tive scoring; and both groups of subjects were sufficiently large to prevent chance errors and individual variations from invali- dating the results. Chapter I contains a survey and criticism of previous in- vestigations, with suggestions concerning the necessity and im- portance of standardizing procedure in this field of research. In chapter II the materials and procedure used in the present investigation are presented and discussed. This chapter con- tains also the data of both the practice and the tests, with a brief discussion of the same. Chapter III deals with the trans- fer effects. The results of the experiment are interpreted and explained and general conclusions drawn. The writer gratefully acknowledges assistance from sev- eral sources. He is indebted to Professors R. S. Woodworth, J. McK. Cattell, E. L. Thorndike, and Dr. A. T. Poffenberger, Jr., for many helpful suggestions and criticisms; and to the Hebrew Orphans Home, of New York City, for the services of the subjects and for the use of the school rooms of that insti- tution. To the subjects themselves he is indebted for unfailing devotion to the work involved, especially in the practice. VI THE TRANSFER EFFECTS OF PRACTICE IN CANCELLATION TESTS CHAPTER I. DISCUSSION OF PREVIOUS INVESTIGATIONS Ever since James announced the results of his experiment on the transfer effects of memory-training, and Thorndike and Woodworth published their conclusions based on the results of an elaborate series of transfer experiments, this field of re- search has proved a fruitful source of experimentation. While it is true that James found no evidence of transfer, yet, it is to James we must make acknowledgement for the inception of investigations which have already modified our psychological and educational concepts, and which bid fair to revolutionize our educational procedure- However much investigators may disagree as to whether special training is general in its effects, they are agreed that there is no function which through special training is made equally effective in all sorts of situations. It is hardly neces- sary to mention that conclusions concerning transfer based on experiments on cross-education should be rejected, for Pro- fessor Thorndike has already pointed out that "they have been improperly used as evidence on our question."^ In James's^ experiment on the transfer effects of memory training there was not only no evidence of transfer, but not even of improvement in the tests themselves. This, together with the fact that no control group was used, invalidates James's conclusions. Peterson^' repeated this experiment with nine subjects, using two in the practice and seven in the con- trol. The results indicate clearly the insufficiency of the num- ber of subjects, as well as the faulty method of conducting the practice. In the training series one subject lost about as much 'Thorndike, Educational Psychology, 1913, Vol. 2, p. 365. ^William James, Principles of Psychology, 1908, Vol. 1, pp. 667-668 'Psychological Review, 1912, 19, 491-492. 1 2 TRANSFER EFFECTS OF PRACTICE. as the other gained, and yet gained more in the tests. The two together, however, gained in the tests more than twice as much as the control group. In the training the two subjects alternated as experimenter and subject, one training with "The Coming of Arthur" and the other with "Guinevere." Thus, while each subject was primarily trained in memorizing one of the poems, he was incidentally trained with the other. These conditions seemed to facilitate the progress of one subject and interfere with the other; the latter subject made greater prog- ress in the tests. Surely the results of such an experiment cannot be accepted as valid. The best known and most elaborate of the earlier investi- gations of the transfer effects of memory-training was that of Ebert and Meumann.^ They sought an answer to the follow- ing question : "Is there a general memory function which can be perfected upon any material involving the use of memory, or, on the other hand, must we posit related or unrelated spe- cial memories"? Six subjects were tested with a wide range of material involving the use of memory- They were then trained in memorizing four lists of twelve nonsense syllables a day for eight days. After which they were retested with a series of tests similar to those used before the practice. There was another eight days of training with nonsense syl- lables, followed by a third series of tests. The results indicate considerable improvement in both the training and the tests. The authors conclude that "there are no doubt related memory functions which can be perfected upon any material involving the use of memory, the development taking place proportion- ately to the degree of relationship between the practice and the test material." In addition to certain irregularities in con- ducting one of the tests, the validity of every test has been called in question. There is no guarantee that the three test series were of equal difficulty. Too few subjects were used — only six — and in three out of the seven tests in immediate learning there were only two. However, the fact that no con- trol group was used is sufficient to invalidate the conclusions of the authors. Dearborn- sought to remedy this defect by re- peating the tests only, and concluded as follows: "The re- sults indicate that a considerable part of the improvement 'Ueber einige Grundfragen der Psychologie der Uebungsphanomene im Bereiche des Gedachtnisses. Archiv. fur die gesamte Psychologie, 1904, 4, 1-232. ^W. F. Dearborn, Psychological Bulletin, 1906, 6, 44. DISCUSSION OF PREVIOUS INVESTIGATIONS. 3 found must be attributed to direct practice in the test-series, and not to any 'spread' of improvement from the practice ser- ies proper. There is further, at times, lack of correlation be- tween the amount of improvement made in the practice and that made in the test series ; occasionally a larger percentage of gain is made in the latter than in the practice itself. This again indicates the presence of direct practice in the test- series. Some at least of the remaining general improvement found is to be explained simply in terms of orientation, attention, and changes in the technique of learning. These results seem to render unnecessary the hypothesis proposed by Ebert and Meumann to account for the large extent of the general influence of special practice, which their experiments seem to indicate." The most elaborate of the more recent investigations of the transfer effects of memory-training is that made by W. G. Sleight,^ of England. The general plan was as follows : Three series of ten memory tests were prepared in advance, and as far as possible the three series were equally difficult. The sub- jects were divided into four groups. One group took the tests only, and was known as the control group, or group number one ; the other three were trained in memorizing. Group num- ber two was trained with poetry; group number three with arithmetical tables, and distances from London to the chief towns of England, etc.; and group number four with prose substance. The training occupied thirty minutes a day, four days in a week, over a period of six weeks. The four groups were tested before the practice began, and again at the end of three weeks, and finally at the conclusion of the practice. Sleight says: "Care was taken that each group should work under similar conditions, and that the group which underwent no memory training was never allowed to have the impression that it was in any way handicapped or under conditions not similar to those of the others." The tests were as follows : 1. Points in Circles; an adaptation of a test used by McDougall and Burt. 2. Dates. Two series, each consisting of six dates and their corresponding events, were repeated by the subjects after the experimenter a given number of times. The event was then read out, and the subjects wrote the date. 3. Nonsense Syllables. The syllables were printed in ^British Journal of Psychology, 1911, 4, 386-457. 4 TRANSFER EFFECTS OF PRACTICE. white chalk upon a blackboard disc which revolved at a con- stant rate behind a screen in which there was a rectangular opening through which the exposure was made. Each of the eight couplets was exposed five times, the subjects repeating them aloud as they appeared, with emphasis upon the second syllable. The experimenter repeated the first syllable of each pair and the subjects wrote the associated syllable. 4. Poetry. A stanza of from eight to twelve lines was read to the class; after which the group repeated each line after the experimenter. After several such repetitions (Sleight does not say how many), the subjects wrote what they could remember. This was covered and the stanza given a few more repetitions, followed by another attempt to reproduce it in writing. The correct items in each attempt constituted the test result. 5. Literal Prose. The method of procedure in this test was the same as that in the poetry test; the repetitions being for the first attempt six, and for the second, three. 6. Prose Substance. The selection was read twice to the subjects ; after which they were told to write the substance. 7. Map Test. A large map of the world was exposed. Each subject had a corresponding outline map on his desk. The experimenter as he indicated a position on the wall map would announce its name. The wall map was then covered, the name announced, and the subjects required to locate with a cross the position on their own maps. Forty positions were used ; the first sixteen being given out two at a time, and the remaining twenty-four, three at a time. 8. Dictation. A prose selection divided into intelligible and grammatically complete portions, beginning with eight and increasing gradually to nineteen words. Each portion was dictated once. The subjects immediately wrote what they remembered. 9. Letters. Consonants only were used. There were 16 series; the first and second contained four letters each; the third and fourth five letters each ; the fifth, sixth, seventh, and eighth, six letters each ; the ninth, tenth, eleventh, and twelfth, seven letters each ; and the remaining four, eight letters each. Each series was dictated once, the subjects immediately re- producing it in writing. 10. Names. Forty-four common Christian names and surnames were used, dictated first in two series of two pairs DISCUSSION OF PREVIOUS INVESTIGATIONS. 5 each, then in eight series of three pairs each, and lastly in four series of four pairs each. After the experimenter had read two pairs, or three pairs, or four pairs, he repeated a surname and the subjects wrote down the corresponding Christian name. The names were not given in the order in which they were first read. The subject's score was deter- mined by the number of correct names. In only two of the ten tests were errors scored, and these were arbitrarily penalized. Attention is called to the following precautions used in giving the tests : 1. All answers were written. 2. Plenty of time was allowed for every answer. 3. No test was begun or carried on unless every subject appeared to be giving attention. 4. Every unfamiliar test was preceded by a short practice. The subjects were 84 Sixth Standard girls of average age 12 years, 8 months. Sleight says that no attempt was made to estimate numeri- cally the direct effect of the practice, because the conditions under which the practice took place did not admit of this. He simply assumes that improvement occurred. The three kinds of practice were conducted orally. In training with poetry the experimenter would read a line and the subjects would repeat it after him. From 20 to 30 lines were used in this way each day until the average child could repeat them without help. The same procedure was used in practicing with arith- metical tables. In the training with prose substance the se- lection was read twice to the subjects who then reproduced the substance in writing. In the final results Sleight indicates three scales of certainty of significance: First, those where the superiority or inferiority of a practice group is at least five times the probable error; second, where the difference is at least three times the probable error; and third, where the difference is between two and three times the probable error. The remaining numbers he considers of no significance whatever. According to this method of determining the signi- ficance of the results, the practice groups show neither superi- ority nor inferiority to the control group in four tests : Dates, Poetry, Letters, and Names. In the other tests there are some significant differences; in some cases positive, in other cases negative. The significant cases are indicated in the following table, or Table A. 6 TRANSFER EFFECTS OF PRACTICE. TABLE A. Superi- ority over Practice Groups Tests Control P.E. Poetry Nons. Sylls 66 11 Prose Subst -22 11 Map Test 50 14 Dictation -32 12 Tables Points in Circles 48 18 Nons. Sylls 85 11 Prose Subst. Literal Prose 21 11 Prose Subst 31 11 An examination of Table A shows that the Poetry Practiced have a superiority in the Test with Nonsense Syllables, and in the Map Test; and an inferiority in the Tests with Prose Substance, and Dictation. The Tables Practiced have a superi- ority with the Test Points in Circles, and Nonsense-Syllables. The Prose Substance Practiced have a superiority in the Tests with Prose-Substance, and Literal Prose. A short time after the completion of the above investiga- tion Sleight conducted a somewhat similar experiment with Women Students of the average age 18-19. There were four groups as before. Group number one took the tests only, while groups 2, 3, 4, practiced with Poetry, Tables, and Prose-Sub- stance, respectively. Instead of ten tests there were only six. They were as follows: 1. Dates. A series of ten dates, each with its associated event was repeated six times, after which the subjects repro- duced the dates as the events were announced. 2. Nonsense Syllables. The procedure was the same as that used in the first experiment. However, there were twelve couplets instead of eight. 3. Poetry. The subjects repeated after the experimenter line by line a stanza containing about eighty words. The repe- titions were preceded by one complete reading of the poem by the experimenter. 4. Prose (Literal). One complete reading of a prose ex- tract by the experimenter was followed by four repetitions on the part of the subjects according to the method used with the poetry. 5. Prose-Substance. A prose extract was read twice by the experimenter. The subjects then reproduced the substance in writing. DISCUSSION OF PREVIOUS INVESTIGATIONS. 7 6. Letters. This test differed from the Letter Test used in the first experiment in that it was extended to nine letters at one dictation. The training was carried on for twelve consecutive days except Sunday, for half-an-hour each day. After which the second series of tests was given. The training differed from that of the first experiment. Selections of verse were handed to the subjects of group number two, and they were told to memorize for thirty minutes by whatever method they pleased. A similar plan was followed with group number three. The material used included population, import and export tables, coinage systems, and other similar data of a somewhat irreg- ular form. In the training with prose substance the proced- ure was the same as that in the first experiment, except in the length and difficulty of the selections. The results which Sleight regards as significant are presented in Table B. TABLE B. Superi- ority over Practice Groups Tests Control P.E. Poetry Nons. Sylls 33 13 Poetry 33 lb Tables Dates 59 24 Prose Subst. Nons. Sylls -62 13 Prose Subst oZ rfi Letters -27 1^ The significant results in the two experiment are summed up in Table C. This will enable the reader at a glance to com- pare them. TABLE C. Younger Group Older Group Superi- Superi- Practice ority over ^"^^.^^^^tj i:^ Groups Tests Control. P.E. Tests Control. P.E. Poetry Nons. Sylls.... 66 11 Nons. Sylls 33 13 Prose Subst.... -22 11 Poetry 33 16 Tables Nons. Sylls.... 85 11 Dates 59 24 Prose Subst. Literal Prose.. 21 11 Prose Subst.... 52 31 Prose Subst.... 31 11 Letters -27 13 Nons. Sylls -62 13 Sleight proceeds to interpret these results and the differ- ences in the transfer effects between the two experiments 8 TRANSFER EFFECTS OF PRACTICE. by using the introspections of the older subjects. The Poetry Practiced in both experiments have a superiority in the Tests with Nonsense Syllables. While the method of training dif- fered, the method of testing was the same. The element of rhythm was emphasized, for in repeating the syllables the accent was on the second syllable of the couplet. With the younger group poetry training interfered with the Prose- Substance Test, and with the older group failed to affect it. This was probably due to the fact that the method of training the younger subjects emphasized rhythm more than the in- dividual methods used by the older group. This is shown in the superiority of the younger group in the Test with Nonsense Syllables, which is 66, with a probable error of 11, while for the older group it is only 33, with a probable error of 13. Any tendency to apply rhythm to the reproduction of the gist of a prose passage would very likely cause interference. With the older group training with poetry facilitated the test with poetry, but failed to affect it with the younger group. This may have been due to the fact that while the method of testing was the same for the two groups, the methods of training were different. Sleight says that the poetry used in the prac- tice was chosen for its simplicity, while that in the tests was not so chosen, some of it being unnecessarily difficult. He men- tions other facts which probably "led to a different distribu- tion of attention." The Tables used with the younger group contained the element of rhythm, while those used with the older group did not. This together with the fact that the method of training the younger group admitted of emphasizing rhythm, while with the older group each subject memorized the tables as she pleased, probably accounts for the fact that the Tables Practiced younger group has a superiority with Nonsense Syllables while the older Tables Practiced group seems to be unaffected in this test. On the other hand, the older group has a superiority in the Dates Test, while the younger group appears unaffected. According to Sleight the difference lies in the fact that the older group instead of making use of rhythm as the younger group did, employed a special kind of visualiz- ing power in the practice with the tables which they were able to use in the Dates Test. Practice with prose substance had a positive effect upon the Prose-Substance Test with both groups. However, with DISCUSSION OF PREVIOUS INVESTIGATIONS. 9 three other tests there was a difference. With the younger group Nonsense Syllables were unaffected, while with the older group there was interference. After examining the introspec- tions of the older group Sleight reaches the conclusion that the interference was caused by the strongly contrasted feel- ing-tone accompanying the exercise, while with the younger group this feeling-tone was absent. He says : "One was audi- torially presented, the other visually; one consisted of con- nected logical speech, the other of disconnected meaningless words; one was arhythmic, the other rhythmic." In spite of the fact that the older group has an inferiority of 27, with a probable error of 13 in the test with Letters, Sleight concludes thus: "With regard to other subjects, the group practicing prose substance reproduction remains in the same position it occupied in the first cross-section." It seems to the writer that the differences in age of the two groups, together with the fact that the methods of practice and the methods of conduct- ing two of the tests were very different, are sufficient to account for the differences in results between the two groups. Sleight is to be commended for conceiving and executing a memory experiment so elaborate. No one appreciates more than the writer the many difficulties involved. Although the conclusions may be true, yet one would feel surer of them if the practice had been conducted under measurable conditions and identical methods had been used in both investigations. It is always hazardous to leave it to subjects to practice as they please. Some of his conclusions are very interesting. They may be summed up as follows : 1. Specific memory-training is specific in its results. 2. In some cases practice precludes new adjustments. 3. There is no general memory function which can be sharpened upon any material. 4. Differences in the midst of great similarity in the mental processes involved may lead to a loss of "Transfer" or even to reciprocal interference. G. C. Fracker^ sought to discover the transfer effects of memory-training in an ingenious and unique experiment by so arranging the test series and the practice series "that the elements concerned in the transference might be determined by analysis of the final results." He had twelve subjects — ^Psychological Review, Monograph 38, 56-102. 1908. 10 TRANSFER EFFECTS OF PRACTICE. eight in the practice group and four in the control. The two groups were tested as follows : 1. Memorizing stanzas of poetry. 2. Memorizing the order of four shades of gray according to the method used in the practice series. 3. Memorizing the four intensities of sound used in the practice series, but in a series of nine instead of ten. 4. Memorizing the four grays used in the second test, but in a series of nine. 5. Memorizing four tones (major chord on piano) pre- sented according to the method of the practice series. 6. Reproduction of nine geometrical forms exposed all at once on a card for ten seconds. 7. Reproduction of nine two-place numbers in their order after one hearing lasting 13.5 seconds. 8. Memory of extent of arm movements. The practice consisted in memorizing four different sound intensities of the same tuning-fork. The four intensities were presented in a certain order, each intensity lasting a half- second, with an interval of a half-second before the next in- tensity was presented, and so on until the series of four had been presented. Then followed an interval of four seconds, after which the four intensities were presented as before except in a different order. During the next interval of four seconds the subject had to recall the order of the first four intensities presented; then followed another presentation of the four in- tensities in a still different order; after which the subject recalled, during the interval of four seconds, the order of the presentation in the second series. Thus the training pro- ceeded. One subject took 3600, six took each about 3000, while one took only six hundred of these trials. The two groups were then re-tested with the eight tests used before the prac- tice began. This investigation has been thoroughly analyzed and dis- cussed by Sleight,^ in England, and Thorndike' in this country. Between the two, the possibihties of criticism have been so exhausted that hardly anything is left to the writer except to indicate the defects pointed out by them. Thorndike, by omit- ting from consideration two subjects of the practice group — one who did worse at the end of the training, and one who ^British Journ. of Psychology, 1911, 4, 395-398. 'Educational Psychology, 1913, Vol. 2, 393-396. DISCUSSION OF PREVIOUS INVESTIGATIONS. H trained only two days — reaches the conclusion that "on the whole it seems safe to say that a gain in the peculiar ability to grasp the order of four un-named facts by naming them or otherwise, and to hold them while reporting a similar previous set and grasping another, carries over from sound intensities to pitches and grays to half of its own amount and improves the grasping and holding of a series of nine grays, forms or numbers to one-fifth of its own amount— all the tests being subject to the same general conditions of a laboratory experi- ment." Attention is called also to the fact that the two ob- servers mentioned made great gains in the test series most closely allied to the practice. Sleight calls attention to several irregularities, and espe- cially to the fact that the withdrawal of subject number five, who made rather spectacular gains in six of the tests, reduces the average percentage of improvement of the trained over the untrained from 19 and 10, to 13 and 10, in the test with the nine grays. He points out also the fact that by omitting the sub- ject of the control group who showed a large retrogression in some of the tests, the practice group has a decided superiority in only two tests— the four grays and the four tones. The facts pointed out plainly indicate the insufliciency of the num- ber of subjects used, and, consequently, the unreliability of some of Fracker's conclusions. On the basis of his subjects' introspections Fracker concludes as follows : "We are able to say that transference depends upon the nature of the imagery employed in the practice rather than upon any other factor." "If in the mind of the observer, the imagery is capable of adjustment to different tasks, it can be used in both improve- ment and transference, for the elements of the training act are thereby made the same as those of the test act. If it is adapted, in the mind of the observer, to the training task only, it may assist in improvement, but it may interfere with trans- ference." Mr. W. H. Winch, of England, has made several experi- ments on the problem of transfer. Most of them deal with memory functions. Since none of them is satisfactory, and so many are similar, a discussion of the more typical and im- portant will suffice. One^ of the first has to do with the effect of memorizing poetry upon rote memory for historical passages and descriptions about places. Although the practice was very ^British Journ. of Psychol., 1908, 2, 284-293. 12 TRANSFER EFFECTS OF PRACTICE. brief — from forty to sixty minutes plus the time for repro- duction — there was considerable improvement. In the final tests the practice group gained much more than the control group. However, it should be borne in mind that it was rote memorizing in both the practice and the tests. In another experiment' Winch investigated the transfer effects of training in rote memory upon substance memory. The practice group was trained 20 minutes a day for three days — one day in each successive week — in rote memory. The control group was occupied during the same period in drawing difficult geometrical designs. The two groups were then re- tested with prose substance. The practice group gained 21 per cent, and the control group 10 per cent, in the tests, while the practice group gained 13 per cent, in the training. Winch arbitrarily assumes that 2 per cent, of this was due to growth, and concludes as follows: "About as much or more improve- ment reckoned in percentages, as has been made in the practice medium itself — rote memory for meaningless things — has been transferred to substance memory." The resijlts are as- tonishing when one considers the very limited amount of train- ing the practice group received. In the tests less than six lines of very simple prose was read three times to the subjects. This would constitute rote memory for many of the subjects. Then, too, scoring such tests accurately is always difficult. A|; least two or more disinterested judges ought to score such material independently. Such an experiment, instead of es- tablishing anything conclusive about transfer, rather accen- tuates the importance of greater care in conceiving and con- ducting a transfer experiment. The very fact that the superi- ority of the practice group is about the same as its improve- ment in the training series, together with the fact that very simple test material was read three times to the subjects, leads one to suspect that the function tested was the function trained. With another group Winch made a somewhat similar experiment. Instead of one practice period in each of three weeks, there was one in each of thirteen weeks. The superiority of the practice group was only 6 per cent. Taking the two ex- periments together, one is inclined to say that the more prac- tice there is in rote memory the less improvement there is in substance memory. Winch also investigated the transfer effects of practice in 'Ibid., 1909-1910, 3, 386-405. DISCUSSION OF PREVIOUS INVESTIGATIONS. 13 substance memory upon productive imagination with school boys of average age 13 years, four months.^ The practice con- sisted in giving the substance of short stories which the sub- jects had studied visually for five minutes. The tests in pro- ductive imagination consisted in producing a story which in- cluded certain words given to the subjects. The average im- provement in the prose substance practice was 37 per cent., while the superiority of the practice group in producing stories from the words given was 15 per cent. Winch concludes : "We know then that an improvement in imagination has resulted from an improvement in memory due to practice; but we do not know what percentage of the improvement in memory due to practice has been transferred." In a second experiment of this kind with school girls of average age 12 years and eleven months the practice group trained until it ceased to improve and gave evidence of being tired of the work. The results in- dicate an inferiority of 12 per cent, on the part of the practice group, which leads Winch to conclude as follows: "The view that memory can be overtrained and thus have prejudicial results on imaginative work is confirmed." He holds that there is a community of function between substance memory and productive imagination, and yet insists that a limited amount of training in the practice has a positive effect, while over- training has a negative effect. Scoring in such experiments is not only very difficult but often accompanied by a constant error. He says also that the teacher who administered the tests and exercises in the second experiment "was a strong believer in the cultivation of memory functions, and was not aware that I intended to endeavor to produce a decline in the inventive function." The very fact that Winch endeavored to bring about certain results in the second experiment may have influenced him in the selection of the material, in the proced- ure, and in the scoring. If three competent judges had scored the tests in both experiments without knowing to which group a subject belonged, the results might have been different. It seems rather strange that a small amount of practice should have produced a superiority of 15 per cent., and much practice an inferiority of 12 per cent. It will be recalled that in another experiment three practice periods produced a superiority of 13 per cent., and in a similar experiment 13 practice periods caused a superiority of 6 per cent. Surely no one is justified ^British Journ. of Psychol, 1911, 4, 95-125. 14 TRANSFER EFFECTS OF PRACTICE. in venturing a conclusion on the basis of these experiments, except to point out the need of greater caution. The most comprehensive and influential of the earlier in- vestigations of transfer was made by Thorndike and Wood- worth.^ They examined: 1. The influence of certain special training in the esti- mation of magnitudes on the ability to estimate magnitudes of the same general sort. 2. The influence of training in observing words containing combinations of letters or some other characteristics, on the general ability to observe words. 3. The influence of special training in memorizing on the general ability to memorize. In a part of these investigations the authors made use of for the first time a control group. Their results and conclu- sions were so radical that widespread interest in the problem was aroused, and, as a consequence, many investigations were inaugurated. Possibly the most significant and striking part of their conclusions is to be found in these words: "Improve- ment in any single mental function rarely brings about equal improvement in any other function, no matter how similar, for the working of every mental function-group is conditioned by the nature of the data in each particular case." In his recent survey of the literature on the problem of transfer, Professor Thorndike has this to say: "The change is simply the neces- sary result upon the second function of the alteration of those of its factors which were elements of the first function. Train- ing is not totally general; neither is it totally specialized."- One of the most interesting experiments in connection with this problem is the one on the perception of illusions by Judd.^ The method was that of the Miiller-Lyer Illusion. There were two observers, one of whom was Judd, who says he "was trained somewhat irregularly and with a background of abstract knowledge and expectation." He practiced with a figure 54mm. long, oblique lines at 90^ with each other, keeping the stand- ard on the right, both figures being held in a horizontal posi- tion. The iflusion which was 44.8 mm. (i. e-, this was the judged length) at the first trial disappeared after 980 trials, "not by any process of judgment or of indirect correction, but ^Psychological Review, 1901, 8, 247-261, 384-395, 553-564. ^Educational Psychology, 1913, vol. 2, 359. ^Psychol. Review, 1902, 9, 27-39. DISCUSSION OF PREVIOUS INVESTIGATIONS. 15 by coming to look differently than it did at first." In the test before practice the illusion of figure 54mm., 45°, was 45.1mm., after practice, 53.0mm. With figure 68mm., 90°, before prac- tice 58.7mm., after practice, 67.3mm. In these tests it should be remembered that the figures were kept in a horizontal position with the standard on the right. Again using figure 54mm., 90°, 175 determinations were made with the standard on the left. At first the illusion was almost as strong as in the beginning of the practice series. However, the curve rose rapidly, the illusion not quite disappearing with the 175th trial, "thus indicating clearly," says Judd, "the transfer of practice." Turning the figures into a vertical position with the standard above, and again with the standard below, gave al- most perfect results in 80 trials. A comparison was made also between the standard 54mm., 90°, and a simple straight line placed somewhat below and to the left. Judd insists "that in spite of the varying conditions there was a transfer of prac- tice." Observer E. was especially prepared for a study of the inversion of the standard figure. He was tested with ngure 53mm., 90°, the standard being on the left. The illusion was 45.6mm. E. then practiced with figure 68mm., 45°, with the standard on the right. After 750 trials he was again tested with figure 53mm., 90°, the standard at this time being on the right. The illusion was 50.4mm. After the whole practice series of 950 trials was completed, the illusion having not quite disappeared, he was tested again with figure 53mm., 90°, the standard being on the left, as in the original test. Professor Judd at this point emphasizes the fact that "E. did not know anything about the disappearance of the illusion in the prac- tice series, and that no information was given him in regard to the change in figure or the difference in position of the standard hne. The results were astonishing. The illusion was 41mm., and not even 1500 trials made any marked change in it." Judd says : "In spite of change in the length of the lines perceived, and in spite of a change in the degree of obhquity of the additional lines, and, finally, in spite of a new arrange- ment of the figures, the effects of the practice were obviously carried over to the new conditions. Furthermore, it is evi- dent from the curve that the observer started on this second series of measurements with a thoroughly established habit of interpretation. The subject who had cultivated in a purely empirical way this fixed mode of interpretation, misapplied it, 16 TRANSFER EFFECTS OF PRACTICE. to be sure, and by this misapplication of the mode of interpre- tation be increased the strength of the illusion under the new conditions." Scholkow and Judd' sought to discover what effect an under- standing of the principle of refraction would have upon prac- tice in hitting a target placed under water at different depths. One group of boys in grades 5 and 6 practiced with such knowl- edge, while another practiced without it. When the target was twelve inches under water, both groups did equally well. However, when the depth was changed to four inches the in- formed group readily adjusted themselves, while the unin- formed group was much confused, their errors being per- sistent. It is evident that practice in hitting a target at a depth of 12 inches caused facilitation in the one case and inter- ference in the other when the depth was changed. Kline- tested seventeen subjects in marking parts of speech in English prose. Eight of these subjects were retained as a control group, while nine were trained in cancelling e's and t's, from 30 to 45 minutes daily, for fourteen days. The time- limit method was used, the scores being in number marked per minute. The group made considerable improvement, the gains ranging from 31 per cent, to 168 per cent, in the number marked per minute. When the two groups were retested, both made gains, but the control group gained the more. In other words the practice caused interference in the tests. Of all the stimuli presented in the practice situation, only e's and t's were to be cancelled, while such stimuli as parts of speech were to be neglected. In the test situation the conditions were reversed ; the stimuli to be neglected in the practice situation were now to be cancelled, while the stimuli determining the cancelling in the practice situation were now to be neglected. Kline offers the following explanation : "The meaning of the relatively in- ferior work of the practice group is best made out from the reports of the members of the group. One says: (1) 'In crossing out parts of speech one always had to think what part of speech the word was.' (2) 'The crossing out of the letters became a habit and instead of crossing out words one wanted to cross out e's and t's. These seemed to be seen so much more clearly than the parts of speech.' Another writes, 'The prac- tice with e's and t's hindered me in dealing with the parts of 'Educational Review, 1908, 36, 28-42. 'Bulletin of the State Normal School, Duluth, Minn., Feb., 1909. DISCUSSION OF PREVIOUS INVESTIGATIONS. 17 speech. I think it was because I was accustomed to looking for e's and t's and the tendency was to cross out those letters rather than the parts of speech.' " Coover and Angell^ tested four subjects in discriminating shades of gray, each test consisting of thirty-five judgments, made on three separate days. Three other subjects took the tests also, but made their judgments on two separate days. The four subjects were trained in discriminating sound-inten- sities for seventeen days, over a period of fifty-seven days, making forty judgments a day. In the tests the judgments were given in the categories of "lighter," "darker," "like," and "undecided." In the training the judgments were made in terms of "louder," "softer," "like," and "doubtful." At the conclusion of the training the control group of three and the practice group of four were re-tested in discriminating shades of gray. Results are given in per cents. The practice group gained 32 per cent, in right judgments, while the control group lost 7 per cent. About all that one can make out of these re- sults is that the practice group did a little better in judging differences of brightness after training in judging differences of sound-intensities, while the control group did worse in the second seventy judgments in discriminating brightness than in the first seventy. How reliable the figures are is not stated. The authors conclude as follows : "That efficiency of sensible discrimination acquired by training with sound stimuli has been transferred to the efficiency of discriminating brightness stimuli, and that the factors in this transfer are due in great part to habituation and to a more economic adaptation of at- tention, i.e., are general rather than special in character." The investigation is not reported in sufl^icient detail to enable one to go into a thorough analysis of the procedure so as to test the rehability of the results. The very fact that the control group did worse in the second test should be regarded as signi- ficant, for it plainly indicates too few subjects and, perhaps, some irregularity in giving the tests. Coover and Angell- report also an experiment in which they sought to discover the effect of practice in card-sorting upon typewriter reactions. There were four subjects in the prac- tice group and three in the control. The training consisted of about fifteen exercises in sorting cards, distributed over a ^American Journ. of Psychol, 1907, 18, 327-336, 'Ibid, 336-340. 18 TRANSFER EFFECTS OF PRACTICE. period of forty days. Previous to the training the practice group was tested for five days in typewriter reactions, and after the training, for three days ; while the control group was tested for only three days before and two days after. To make the results comparable both groups should have been tested in exactly the same way and under the same conditions. Since there is no common basis of comparison conclusions based on such an experiment are unreliable. The practice group im- proved more in the tests from the first three days before train- ing to the last three days before training, than from the last three days before training to the three after training. Inves- tigations of this kind, when conducted with so few subjects and so little regard for scientific accuracy, only emphasize the necessity and importance of consistent, accurate procedure. They certainly do not justify any such conclusion as that "training the activity of Reaction with Discrimination and Choice by sorting cards into compartments has increased the facility of a like activity in both speed and regularity in 'type- writer-reaction' (a) noticeably, in two cases, after the latter had become automatic, and (b) markedly in two others, in the course of practice." Gilbert and Fracker^ sought to determine what effect prac- tice in reaction and discrimination with stimuli in one sense has on the same process in other senses, the latter not having been practiced at all. There were three subjects, each of whom was tested in reacting simply, and reacting "with choice," to the following stimuli : Color, pressure, and electricity. Each day thereafter for twelve days two of the subjects practiced reacting simply, and also "with choice" to sound. The other subject practiced eleven days in reacting to simple sound only. The three were then tested again as in the beginning. In the opinion of the authors the results justified the following con- clusions : a. That practice in reaction to sound reduces the time of reaction to other forms of stimuli by amounts almost equal to the reduction of the time of sound reaction itself. b. That such practice alone does not reduce the time of discrimination and choice. c. That practice in discrimination of sounds reduces also the time of discrimination for other forms of stimuli. The second conclusion is based on the fact that the subject ^University of Iowa Studies in Psychology, 1897, 1, 62-76. DISCUSSION OF PREVIOUS INVESTIGATIONS. 19 that practiced in simple reaction only, made scarcely any gain in reaction "with choice." The figures show that his record in the first test was almost as good as the records of the other two in the second test. Consequently, we should expect him to make little improvement, for the first test shows him to be much nearer the physiological limit than either of the others. There were too few subjects and not sufficient practice to make the conclusions of any scientific value. Then, too, the functions tested and the functions trained contained many identical ele- ments. Bair^ made several experiments on the influence of prac- tice in forming certain associative habits upon the ability in certain different habits. We shall notice two that bear some- what upon the problem of transfer. 1. Taking six symbols, letters or figures, Bair made a series of fifty-five; with six different symbols he made an- other series of fifty-five ; and so on until he had twenty sets of these series. Having labeled six keys of a typewriter with the six symbols of the first series, he exposed the fifty-five sym- bols of this series, in chance order, one by one, while the sub- jects upon seeing a symbol tapped the corresponding key. The time required to tap out the series was recorded. In the same way six other symbols were used with the series composed of them. The time required to tap out this series was recorded. This was kept up until the twenty sets had been used. It will be noted that the symbols were changed from test to test, thus changing somewhat the conditions of the experiment. Four subjects took the tests. The first improved from 62 to 52; the second, from 95 to 85; the third, from 71.5 to 58; and the fourth, from 65 to 56. Bair claims that the major part of this gain could not have been due to merely getting used to the machine or the general features of the experiments, for the fourth subject was already used to these and still gained nearly as much as any one of the others. It is hardly necessary to call attention to the fact that the conditions of each succeeding test were the same, with the exception of the six symbols. Thus there were many identical elements, and these were being practiced from test to test. Probably the slight improvement made was due to the practice of these elements. 2. The other experiment "consisted in taking daily records, for twenty days, by means of a stop-watch, of the time 'Psychol. Review, Monogr. 19, 1902, pp. 25, 28, 64-67. 20 TRANSFER EFFECTS OF PRACTICE. required to repeat the alphabet from memory. Each day's experiment was as follows : First, the alphabet was repeated as rapidly as possible forward; secondly, the letter n was inter- cepted between each of the letters; thirdly, the alphabet was repeated as rapidly as possible backward ; and lastly, the alpha- bet was repeated backward intercepting n between each of the letters. At the end of twenty practices in each order, the sub- ject repeated the alphabet, first, forward, intercepting, in- stead of n, the letter x, and repeating three times; secondly, intercepting r, and repeating three times ; then lastly, repeating backward, and in like manner intercepting x and r and repeat- ing three times." In the test series there was an improvement equivalent to that of three days' practice in the training series. Here, as before, there were many identical elements. Ruger^ in an "experimental study of the processes involved in the solution of mechanical puzzles and in the acquisition of skill in their manipulation," also studied the transfer effects, and attempted to isolate the factors involved. The materials and method of procedure are too intricate for an intelligible and detailed description without some study of the puzzles themselves. With the aid of his subjects' introspections in con- nection with objective measurements Ruger made a classifi- cation of the transfer factors into (1) general factors, and (2) special factors. He says: "It has seemed advisable to the writer to use the term transfer in a very broad sense to in- clude the effect of any given experience on any subsequent one whether the effect results directly or by means of an idea, whether the transfer is one of method, or of material, or of motor processes, and whether it is positive or negative," And, again, "to the writer the problem consciousness considered in itself and as to conditions of eflficiency seems to have many characteristics in common irrespective of the degree of related- ness of the material concerned." Foster- studied the effects of practice upon visualizing and upon the reproduction of visual impressions, and reached the conclusion : "That training in these experim^ents has made the observers noticeably better observers or memorizers in general, or given them any habits of observing closely or reporting correctly, or furnished any ability to meet better any situa- tions generally met with, neither we nor the observers them- selves believe." ^Archives of Psychology, 1910. No. 15. 'Journal of Educational Psychology, 1911, 2, 11-21. DISCUSSION OF PREVIOUS INVESTIGATIONS. 21 Whipple^ tested the effect of practice upon the range of visual attention and of visual apprehension and concluded that the improvement was due to "habituation to experimental con- ditions" and th'^ use of "assimilative devices." In his "Educative Process," page 42, Bagley reports an experiment, made by Squire, on the transfer effects of special training in neatness in arithmetical v^ork, on neatness in other school work. In his report Bagley says: "At the Montana State Normal College careful experiments were undertaken to determine whether the habit of producing neat papers in arith- metic will function with reference to neat written work in other studies; the tests were confined to the intermediate grades. The results are almost startling in their failure to show the slightest improvement in language and spelling pa- pers, although the improvement in the arithmetic papers was noticeable from the very first." On the other hand, Ruediger- reports that neatness culti- vated in connection with one school subject did improve neat- ness in other subjects. However, it would seem almost im- possible for a teacher to follow the instructions given without suggesting to the pupils the importance of neatness in other school subjects as well as the activities specially mentioned. Attention is called to two of the instructions in which the writer thinks these suggestions are involved. "1. Talk freely with the class (not to) on the importance of neatness in dress, business, the home, hospitals, etc., con- necting it as far as you can with the subject under experiment. Guard against overdoses. "2. Do not bring up the subject of neatness with the other studies of the school. If the pupils bring up these studies quietly substitute something else. Talk of neatness only in that class, not to the school in general." The writer feels confident that the fact that neatness was demanded in one school subject, and emphasized as an ideal in everything else than the other school subjects, aroused sus- picion on the part of the pupils, and perhaps, was more effec- tive in influencing their behavior in the tabooed subjects than if such subjects had not been so conspicuously isolated and ne- glected. ^Journal of Educational Psychology, 1910, 1, 249-262. "Educational Revieiv, 1908, 36, 364-371. 22 TRANSFER EFFECTS OF PRACTICE. Jastrow^ found Hermann and Keller, sleight-of-hand per- formers, not quite as good in reaction-time experiments as the average college student. Quick as they were in performing their tricks, they were not quick enough to escape detection when required to perform them in slightly different ways. Raif- tested expert pianists and found that they could not manipulate a telegraph key any more rapidly than persons of average intelligence. Thorndike' calls attention to the fact that experiments on the amount and rate of improvement with practice "shows negatively that the practice in the tasks of school and life which an earnest graduate student has had leave him still far below his possibilities — so far below that a very small amount of time devoted to any special function improves it greatly." In presenting this survey the writer has followed the same general plan used by Thorndike in his excellent presentation and discussion of the investigations of the problems of trans- fer in his Educational Psychology, Vol. 2, chapter 12. Although the writer acknowledges his indebtedness for the free use of Thorndike's presentation, yet, wherever possible, the presen- tations and discussions are the writer's own. In concluding his discussion Thorndike says: "These experimental facts as a whole leave a rather confused impression on one's mind, and resist organization into any simple statement of how far the improvement wrought by special practice spreads beyond the function primarily exercised. They do, however, at least put out of court the old doctrine of a very wide spread of a very large percentage of the special improvement." To the present writer these investigations emphasize also the necessity and importance of a standard method of procedure to which inves- tigations of the problem of transfer should conform. Surely sufficient experimentation has been done to make clear what constitutes a standard method of procedure. Thorndike very aptly says: "The experimental facts now at hand are in a sense trivial in comparison with the very great variety of facts which must be measured in order to describe justly the spread of improvement in the work of the schools, trades, and the like."^ 'Science, N. S., 1896, 3, 685-689. 'Zeitschrift fur Psijchologie, 1900, 24, 352-356. ^Educational Psychology, 1913. Vol. 2, 415-416. 'Op. cit. Vol. 2, p. 417. CHAPTER II. DESCRIPTION AND DISCUSSION OF MATERIALS AND PROCEDURE Section 1. Preliminary Investigations. In three preliminary investigations the writer sought to discover, within a limited field, an answer to some of the ques- tions involved in the problem of the transfer effects of practice. A. Reaction Time The first experiment had to do with reaction time. Three subjects were tested in reacting to a soft sound rather than a loud one ; to a light pressure rather than a heavy one ; and to blue rather than red. They were then practiced for eight weeks in reacting each 200 times daily to a sound of medium intensity. At intervals of two weeks the three tests were re- peated. There was considerable improvement in the test series, but not so much as in the practice series. It was impossible to determine how much of the improvement in the test series, if any, was due to transfer, as there was no control group by which to measure the improvement due to direct practice in the tests themselves. B. Cancellation The second experiment was made with cancellation tests. Nineteen subjects were engaged — nine for the control group, and ten for the practice group. Both groups were tested at the same time, by the time limit method, with five cancellation tests. The practice group was trained 30 minutes a day for 10 days in cancelling in EngHsh prose words containing both a and t. On account of eye strain four of the subjects were forced to withdraw from the practice. When the practice was completed, the two groups — six of the practice group and nine of the control group — were reassembled and retested with the five tests used before the practice began. In only one test did 23 24 TRANSFER EFFECTS OF PRACTICE. the practice group give reliable evidence of superiority to the control group, and that v^as the test in Spanish prose, in which words containing both a and t were cancelled. The elements which determined the cancelling in this test were the same as the elements which determined the cancelling in the practice series. C. Cancellation In the third experiment the materials of the practice series and of the test series were the same as those used in the second experiment. In fact these two experiments were in progress at the same time. The subjects in the second were boys and girls in the fifth grade of the Speyer School, New York City ; while the subjects in the third were girls in the Scudder Secretarial School of New York City. They ranged in age from 18 to 19 years. Thirty-five girls were tested with the five tests. Four- teen of them volunteered to practice regularly every day as they had time. At the end of three weeks only seven had practiced, while only three of these had done the amount re- quired. However, these seven were accepted as the practice group. The two groups were reassembled and tested as before. The practice group had a decided superiority over the control group in the Spanish a — t Word test. In the other tests there were no reliable differences as measured by the probable error. Section II. The Present Investigation. As a result of these preliminary experiments, seven can- cellation tests, to be described later, were decided upon; and in order that the investigation might be thorough the writer determined to secure as large a group of subjects as possible under conditions which would enable him to carry the experi- ment through to a successful conclusion. The general plan has been set forth in the Introduction. Briefly, it was as follows : Having decided that the practice should consist in cancelling, in English prose, words contained both a and t, seven cancella- tion tests more or less similar to the practice material, were selected. Eighty boys, ranging in age from 11 to 13 years, of the Hebrew Orphans Home,^ of New York City, were tested This Institution is more like a large boarding school than like an orphanage. Individuality in manner and dress, spontaneity and freedom of expression, are manifest characteristics of the children. The Super- intendent and one of the Governors cooperated with the writer in arous- ing interest in the experiment and in conducting the practice and tests with systematic regularity. MATERIALS AND PROCEDURE. 25 twice with each of these tests. Thirty-six of these eighty- subjects were trained for four ten-minute periods a day dur- ing sixteen days in cancelhng a — t words in Enghsh prose. At the conclusion of the practice all of the subjects (with the ex- ception of four) were retested twice with each of the seven tests used before the practice. A. Practice Material The material for the practice was selected with special reference to the problem to be solved. One hundred copies of a pamphlet called the "Gold Standard," pubhshed by Ginn and Co., in 1896, were used in the practice series. These books contained 143 pages of prose printed on excellent paper. There were no pictures and the reading matter was of no special in- terest to the subjects. B. Test Material The test materials were selected with special reference to the task involved in the practice. 1. The Spanish a — t Word Test. This test consisted of a paragraph of Spanish prose in which all words containing both a and t were to be cancelled. 2. The Spanish e — s Word Test. This test consisted of the same paragraph of Spanish prose used in Test 1, but now the words containing both e and s were to be cancelled. 3. The a and t Letter Test. This test consisted of a dis- tribution of small letters in which the letters a and t were to be cancelled. 4. The e and s Letter Test. This test consisted of the same distribution of small letters used in Test 3, but now the letters e and s were to be cancelled. 5. The Columbia A Test. This test consisted in a distri- bution of capital letters in which the A's were to be cancelled. 6. The Thorndike B Test This test consisted in a distri- bution of capital letters different from that used in Test 5, in which the B's were to be cancelled. 7. The Woodworth and Wells Number-Group Checking Test. This test consisted in a distribution of six-place number- groups, in which groups containing both 4 and 7 were to be cancelled. 26 TRANSFER EFFECTS OF PRACTICE. C. Procedure Each subject was supplied with a drawing pencil, which had been properly sharpened. Before the blanks were distributed one was shown to the subjects and the task to be performed was carefully explained. The subjects were then asked ques- tions concerning the task, and, in turn, were invited to ask ques- tions of the writer, if they did not understand clearly what the task was. They were told that the test blanks would be dis- tributed face down and that they were to write their names on the back of them, and that just three seconds before they were to begin cancelling, the signal "ready" would be given, which would mean that the blanks were to be turned over and pencils held in readiness for the command to "start" cancell- ing. They were also told that the cancelling would last exactly one minute, and that at the signal "stop," they were to lay the pencils on the table. Promptly at the conclusion of each test the blanks were collected and another set distributed. The same method was used with each one of the seven tests. Im- mediately upon the completion of the series, the seven tests were given again in the same order with the same instructions as before. The subjects were urged at the beginning of each test to cancel only the designated symbols and to cancel as rapidly as possible. All of the work, both the practice and tests, was done at night between 7:15 and 8:30, in one of the large school rooms of the Hebrew Orphans Home. The seven tests described above were given on October 22, 1914 and will be known in this investigation as the Initial Test Series. On October 26 thirty- seven of the eighty subjects began the practice, and will be referred to as the Practice Group; while the remaining sub- jects will be known as the Control Group.^ The practice per- iods were each 10 minutes long and there were four of them each night for 16 nights, distributed over a period of 22 days. There was no practice on Friday and Sunday night. At the be- ginning of each practice period, the task to be performed was carefully explained and the subjects urged to do their best in speed and accuracy. Each subject was provided with a draw- ing pencil properly sharpened. These pencils along with the ^As a matter of fact, on the 12th day of the practice, one subject of the practice group was expelled on account of cheating; and three of the control group failed to take the final test, on account of sickness. Con- sequently the practice group consisted of 36 and the control group of 40 subjects. MATERIALS AND PROCEDURE. 27 books were collected promptly at the conclusion of the practice each night. Precautions were taken against cheating. One of the Gov- ernors of the Home, and two trustworthy monitors, remained in the room during the practice. Promptly upon the signal from the writer to stop cancelling, the monitors, under his di- rection, rubber-stamped the place reached by each subject. The books were closed and the pencils placed on the desk. Careful watch was kept to see that no subject tampered with his book during the relaxation interval allowed between the periods. On the 12th night of the practice one subject opened his book during this interval and began cancelling. He was promptly expelled with the approval of those in authority. The interval between practice periods was from six to eight minutes. Realizing the importance of securing a maximum of im- provement in the practice, the following incentives Wr»re em- ployed: (1) On the second night of the practice, two prizes were offered, one to the subject making the greatest improve- ment, and one to the subject cancelling the largest number of words with the fewest errors. (2) Two entertainments were promised ; one during the practice, and another at the conclu- sion if the practice proved satisfactory. (3) Before begin- ning the practice each night, individual scores of the previous night were read to the group, and each subject urged both to excel the other subjects and to beat his own record. (4) During each relaxation interval, the subjects were given candy or entertained with short stories or music. (5) During each practice period, the writer was constantly among the subjects encouraging them by his presence and interest. In the mean- time, the members of the control group were assured repeat- edly that they were just as capable and important as the prac- tice group. They were promised an entertainment if they did their best in the final test. Their behavior indicated an interest in the experiment equal to that of the practice group. It should be said that these promises were fulfilled. Both groups were fully informed as to the nature of the investigation. The practice was concluded on Monday night, November 16. On the following night, the two groups — the control group of forty and the practice group of thiry-six — were reassembled, and again tested with the same seven tests used before the practice began, each test being used twice. All of the material, 28 TRANSFER EFFECTS OF PRACTICE. both of the practice series and of the initial and final test series, was scored carefully by the writer himself. The scor- ing consisted in counting the number of cancellations and er- rors, the errors being mostly those of omission. The total number of a — t words cancelled in the practice series was 452,995 and the total number of errors was 35,719. The data of the experiment will be found in the next section. Since more than 99 per cent, of the errors were errors of omission, the total ground covered, or gross speed, is repre- sented by the sum of the cancellations and errors. Conse- quently the per cent, of accuracy per period can be obtained by dividing the number of cancellations per period by the total ground covered, or gross speed. For instance, the average number of cancellations for the first practice period was 102.6 a — t words, and the average number of errors (mostly omis- sions), 27 a — t words. Therefore, the total ground covered, or gross speed, was 129.6 a — t words. By dividing 102.6 by 129.6 the per cent, of accuracy is obtained. Section III. Treatment of the Data A. Data of the Practice Series Instead of presenting in detail all of the data of the practice series, the data of the group performance in each practice per- iod will be presented. It will be found in Table I. The reader should bear in mind that each practice period was ten minutes long, and that there were four of them each night for sixteen nights. Consequently, there are sixteen first-periods, sixteen second-periods, sixteen third-periods and sixteen fourth-periods. The table is so constructed that one can not only follow the progress of the group both in speed and accuracy of cancellation through each successive practice period, but from each first-period to the next first-period, and from each second-period to the next second-period, and so on with the sixteen third-periods and sixteen fourth-periods. The first double column contains the data for the sixteen first- periods ; the first half containing the average number of can- cellations, and the second half the corresponding percents of accuracy. The remaining columns of the table are con- structed in exactly the same manner. If one wishes to follow the progress of the practice group in successive practice periods, he reads across the table thus : In MATERIALS AND PROCEDURE. 29 the first period the group has an average performance of 102.6 cancellations, and an accuracy of 79 per cent. ; in the second period, 101.0 cancellations and 82 per cent, of accuracy, and so on to the fourth period. In the fifth period (the first period of the second day) an average of 132.3 cancellations and an accuracy of 86 per cent., and so on through the table. If, on the other hand, one cares to follow the progress in each suc- Figure I — Showing Progress of the Practice Group in Average Num- ber of Cancellations per Period, through the Sixty-four Practice Periods. Figure II — Showing Improvement in the Percent of Accuracy. 30 TRANSFER EFFECTS OF PRACTICE. TABLE I. DATA OF THE PRACTICE SERIES (The figures represent the average for the thirty-six subjects. Each of the four periods per day is averaged separately and occupies a sepa- rate column in the table. Accompanying each average is the correspond- ing per cent, of accuracy for that period.) Days First Period Second Period Third Period Fourth Period Av. Can. %Acc. Av. Can. %Acc. Av. Can. %Acc. Av, Can. %Acc. 1 102.6 79 101.0 82 126.4 86 128,9 84 2 132.3 86 137.7 86 145.8 85 151.3 85 3 134.9 88 149.1 87 153.4 87 161.2 87 4 146.0 90 152.2 89 156.6 88 159.2 89 5 145.1 91 153.0 89 182.2 90 189.1 93 6 160.9 93 174.3 92 182.0 93 166.3 94 7 176.2 93 186.1 92 196.1 93 192.6 94 8 211.8 93 203.2 93 218.4 93 202.8 94 9 212.9 94 216.8 93 190.0 95 206.1 94 10 216.6 95 226.4 94 211.1 94 225.5 95 11 225.2 94 234.3 93 218.3 95 223.4 94 12 235.2 94 236.2 94 232.3 95 241.3 95 13 234.6 94 238.3 93 240.1 94 231.9 96 14 236.5 95 237.0 95 234.6 96 228.9 95 15 231.3 96 233.6 95 234.2 94 243.4 94 16 239.5 96 260.7 96 257.6 96 266.5 96 ceeding first-period, or second-period, and so on, he reads down the double columns. If the practice is followed from period to period as it occurs and the changes observed, one will notice that out of the sixty-three possible changes there are forty- four in which there is an increase in the number of cancella- tions ; ten of these are accompanied by an increase in accuracy, twenty by a decrease and fourteen by no change. There are eighteen cases in which there is a decrease in the number of cancellations, two of which are accompanied by a decrease in accuracy, thirteen by an increase and three by no change. The one case in which there is no change in the number of cancellations is accompanied by a decrease in accuracy. As the writer deemed it unwise to attempt to get introspections from subjects so young, he has no explanations of these facts to offer from the point of view of the subjects themselves. However, he noticed after the first week of practice, that the subjects were striving more and more for accuracy. During the second and third weeks more than half of the errors were made by fewer than eight subjects. MATERIALS AND PROCEDURE. 31 Figure III — Showing Improvement in Speed of Cancellation in Terms of the Four Daily Practice Periods. ^ 3 -V 5- 9 10 11 /A n ;y js Ik Figure IV — Showing Improvement in Percent, of Accuracy in Terms of the Four Daily Practice Periods. 32 TRANSFER EFFECTS OF PRACTICE. The data in Table I are presented in graphic form in figures I, II, III and IV. Figure I shows the progress of the practice group in average number of cancellations per period for the sixty-four ten-minute periods of the practice series, while figure II shows the corresponding per cents, of accuracy. The numbers along the horizontal axis indicate the practice periods, while those along the vertical axis indicate, in figure I, average number of cancellations per period, and in figure II, per cents, of accuracy. In figure III the progress in average cancellation per period is presented, not through each successive period, but through each of the sixteen first-periods, each of the sixteen second-periods, and so on. The continuous line represents the progress through the first-periods, the broken line through the second-periods, the dash line through the third-periods and the dotted line through the fourth-periods. The corresponding per cents, of accuracy are presented in the same way in figure IV. An inspection of these figures shows that the practice group made considerable pro- gress both in speed and accuracy of cancellation. This im- provement is evident both in the graphs showing the consecu- tive practice periods, and also in those showing the four sepa- rate practice periods of each day in separate curves. B. Data of the Test Series Table II presents in detail the data of the seven tests for the practice group. In the first column the subjects are indi- cated by numbers 1 to 36. Under Spanish a — t Word Test are four columns ; the first two contain the data of the initial test series, while the last two contain the data of the final test series. Under "Can." will be found the number of actual can- cellations per minute for each subject, and under "Er." the corresponding errors. The data of the other six tests are re- corded in exactly the same way. It should be borne in mind that in each series of tests — the initial and the final — each of the seven tests was given twice, and that the numbers re- corded in Table II are the averages of these two tests. The table reads thus: In the Spanish a — t Word Test subject No. 1 has an initial performance of 7,0 cancellations per minute and 1.5 errors, and a final performance of 15.0 cancellations per minute and 2.0 errors; in the Spanish e — s Word Test an initial performance of 10.0 cancellations per minute and 1.0 errors, and a final performance of 11.0 cancellations per min- MATERIALS AND PROCEDURE. 33 ute and 0.0 errors, and so on with each of the other tests. The performances of the other subjects are read in the same way. TABLE II. DATA OF THE PRACTICE GROUP IN THE TEST SERIES Spanish a-t Word Test Initial Final Sub. Can. Er. Can. Er. 1 7.0 1.5 15.0 2.0 2 9.0 1.5 14.0 0.0 3 8.0 0.0 16.0 0.5 4 9.0 1.5 17.5 3.5 5 9.5 1.0 19.5 0.0 6 12.5 1.0 19.5 1.5 7 10.0 3.0 15.5 1.0 8 11.5 0.0 18.0 0.0 9 11.0 0.0 18.0 4.0 10 11.5 0.0 23.0 1.5 11 11.0 1.0 17.5 1.5 12 11.5 0.0 19.5 0.5 13 10.5 1.5 19.5 1.0 14 10.5 1.0 18.5 2.0 15 9.0 1.0 18.0 1.0 16 9.0 2.5 19.5 0.5 17 11.5 1.0 25.5 1.0 18 10.0 0.0 18.5 0.0 19 8.5 2.0 20.0 0.5 20 10.0 0.5 19.5 1.5 21 9.0 2.0 14.0 2.5 22 11.0 0.5 17.0 3.0 23 11.5 0.5 19.0 2.5 24 9.5 1.5 20.0 1.0 25 10.5 0.5 17.5 3.5 26 9.5 6.5 20.5 2.5 27 8.0 2.5 15.5 0.5 28 7.0 4.5 17.0 1.5 29 10.5 1.0 21.5 1.5 30 11.0 1.5 20.5 0.5 31 11.0 0.5 17.5 0.5 32 10.0 1.5 16.5 0.5 33 9.5 0.5 14.0 0.5 34 12.0 12.0 21.0 4.5 35 9.0 0.5 20.0 3.5 36 11.5 0.0 21.5 1.5 Spanish e-s Word Test Initial Final Can. Er. Can. Er- 10.0 1.0 11.0 0.0 9.5 2.5 11.5 1.5 10.5 0.0 12.0 3.0 11.0 0.0 12.0 4.0 11.0 1.0 11.5 3.0 11.5 0.0 15.0 1.5 10.0 1.5 8.5 3.0 12.5 0.0 12.5 1.0 11.5 0.0 14.0 9.0 11.0 0.5 16.0 7.0 11.0 1.0 10.0 5.0 12.0 0.0 12.0 1.5 12.0 0.5 12.5 4.0 11.0 0.0 10.0 3.5 10.5 0.0 11.0 2.0 10.5 0.5 11.5 1.0 15.5 2.5 20.5 8.0 10.5 0.5 12.0 1.0 10.0 0.5 12.5 3.0 11.0 0.0 15.5 4.5 10.5 1.0 10.0 10.5 12.0 2.5 12.0 4.5 12.0 0.0 17.0 3.5 7.5 4.0 15.0 7.5 11.0 0.5 12.5 7.0 11.5 1.5 11.5 2.0 11.5 1.5 12.0 5.0 11.0 2.0 10.5 4.5 11.5 0.5 16.5 3.5 14.5 6.0 15.0 10.5 11.0 1.0 12.0 2.5 10.0 1.0 9.5 3.0 10.0 0.0 11.0 0.0 12.5 1.0 11.0 10.0 10.5 1.0 11.5 16.0 11.5 0.0 13.0 2.5 This table shows the initial and final performance of each member of the practice group in each of the seven tests, in terms of number of cancellations and errors per minute. 34 TRANSFER EFFECTS OF PRACTICE. TABLE II (Continued). DATA OF THE PRACTICE GROUP IN THE TEST SERIES Letter Test a and t Letter Test e and s S„h r^^^'^'i /'""^^ ^"^*^^1 Final Sub. Can. Er. Can. Er. Can. Er. Can Er- 1 27.0 0.0 34.0 1.0 40.0 0.0 42.0 1.5 2 29.0 5.5 29.5 2.5 42.0 1.0 415 15 3 18.5 2.5 29.5 2.0 36.0 0.0 39*5 25 4 26.5 8.0 30.5 0.0 39.0 4.0 490 10 32.0 60 40.0 6.5 47.5 3.0 53 5 ! 7 Hi rn tli '•' ''■' '•' 52-5 2.0 8 275 li f.i '•' ''•' 2.5 41.5 1.0 9 So li Hi '•' "'-^ '-' 41.0 2.0 9 27.0 8.5 38.0 1.0 43.0 52 5 10 10 30.5 6.0 48.0 0.5 47.5 25 56 .0 2 ?0? oi ''•' '•' ''■' «-5 39.0 0.5 12 30.5 2.5 47.5 1.0 40.5 0.0 495 30 13 27.5 9.0 36.5 1.5 4I.O 4 5 ^0 oi 14 24.0 0.5 33.0 1.0 43.5 H 32. ! 6 32I In f«n '■' ''■' '■' 29.5 0.0 16 32.5 2.0 48.0 1.0 53.5 2.0 57 15 17 9.0 1.0 46.0 0.5 53.5 1.5 530 2.5 18 26.5 3.0 42.0 1.0 39.0 2.5 47 5 19 24.5 1.0 35.0 3-0 36.5 1.5 tli 25 2 2 10I tli '■' ''■' '■' 4^-« 20 22 240 9, Hi ^'' 29.5 8.0 33.5 10-0 23 So li !?n '•' ^'-^ ^-0 29.0 1.5 24 Iti ti til '•' ''■' '■' 46.0 0.5 ^4 33.5 5.5 34.5 2.5 44.0 4 45 9k 25 20.5 0.0 36.0 1.0 34.5 t'o til H fl 25*0 «'n f.i '•' ^'-^ 3-0 34.0 1.5 27 25.0 6.0 41-5 5.0 44.0 15 420 R^ 29 300 \\ '^l '-' ''-' '' 45:0 3: In m .i '^-^ ^'^ 45.5 1.5 52.5 2-5 30 43.5 5.5 53.0 1.5 70.5 2.0 66.5 2.5 32 lA li ''•' '•' ^^-5 1-0 36.0 2.0 33 ?«n A f •' '•' 32.0 0.5 26.0 4.0 34 36 110 ?A '•' 28.0 2.5 30.0 0-5 35 2?'o ]i I'/ ^-^ ''■' ^-5 55.0 7.5 36 lln li fA ^'^ 3'-^ 5-0 40.0 10.5 36 26.0 3.0 40.0 3.0 38.5 1.5 39 2 MATERIALS AND PROCEDURE. 35 TABLE II (Continued). DATA OF THE PRACTICE GROUP IN THE TEST SERIES Capital A Test Capital B Test Initial Final Initial Final Sub . Can. Er. Can. Er. Can. Er. Can. Er- 1 36.5 1.5 45.0 0.5 18.5 0.5 20.0 0.5 2 37.0 1.0 40.0 0.0 20.5 3.0 21.0 1.5 3 34.0 3.0 42.5 0.5 15.5 0.0 22.0 2-5 4 40.5 0.5 46.5 0.0 19.0 0.0 18.0 1.0 5 47.0 1.0 62-5 0.5 24.0 0.5 31.5 1.0 6 41.5 0.0 61.0 0.0 23.0 0.5 31.0 3.0 7 41.0 0.0 54.5 1.0 22.0 0.0 31.0 0-5 8 40.0 0.0 56.0 1.0 20.0 1.0 25.0 4.0 9 36.5 0.0 46-5 1.0 19.5 1.0 25.5 0.0 10 32.0 3.5 72.0 5.0 20.5 0.5 32.0 4.0 11 37.0 3.0 43.5 0.0 20.0 1.0 25.0 00 12 41.5 3.0 57.0 0.0 23.0 0.5 34.0 0.0 13 41.5 0.0 510 0.5 21.5 0.5 32.0 0.5 14 38.5 0.0 48.5 0.0 19.5 0.0 20.0 0.5 15 33.5 1.0 48.0 0.0 14.5 0.0 19.0 2-5 16 44.5 1.5 62.5 0.5 24.5 0.5 31.5 0.0 17 37.5 0.0 61-5 1.0 21.5 0.0 28.0 2.0 18 37.0 0.0 49.0 0.0 20.0 0.5 28.5 1.0 19 41.5 0.5 57.5 0.5 16.5 0.5 22.0 7-5 20 30.5 1.0 47.0 4.5 17.5 1.0 29.5 3.5 21 37.0 2.0 42-5 3.0 17.0 3.5 19.0 1.5 22 35.5 0.5 37.0 0.5 25.5 0.0 24.0 0.0 23 88.0 1.0 48.0 1.5 20.0 0.0 22.0 7-0 24 43.5 0,5 49.0 0.0 20.5 0.5 28.5 4.5 25 35.5 0.0 44-5 0.0 16.0 0.0 21.0 0.5 26 49.5 0.0 58.5 0.0 25.5 1.5 22.0 5.0 27 31.5 1.5 50.5 6.5 19.5 1.5 25.0 5.0 28 35.5 0.0 50-5 2.0 18.0 2.0 23.0 1.5 29 38.0 1.0 57.5 0.5 23.0 0.5 31.0 1.5 30 45.5 8.0 60.5 0.5 33.5 0.5 34.0 10 31 40.0 0.5 52.5 0.0 23.5 0.5 27.5 1.0 32 35.0 0.0 41-5 0.5 19.5 1.0 24.0 0.0 33 32.5 1.0 44.0 0.0 18.5 1.0 22.5 0.5 34 41.0 6.0 66.5 3.5 32.5 4.0 33.0 30 35 34.5 0.0 49.0 0.5 21.5 0.0 28.5 1.0 36 38.5 0.5 590 0.5 18.5 0.5 24.5 2.0 36 TRANSFER EFFECTS OF PRACTICE. TABLE II (Continued). DATA OF THE PRACTICE GROUP IN THE TEST SERIES 4 and 7 Group Test Initial Pinal Sub. Can. Er. Can. Er 1 11.0 0.0 16.0 1.0 2 18.0 0.5 16.5 1.0 3 9.0 0.0 17-5 0.0 4 14.5 0.0 18.0 2.5 5 16.5 0.0 21.0 0-5 6 19.0 0.0 22.5 0.5 7 14.0 0.0 16.5 20 8 11.0 0.0 18.0 0.5 9 16.0 0.5 20.0 2.0 10 20.0 0.5 29.0 0.0 11 13.0 0.0 17-5 5.0 12 19.0 0.0 25.0 0.0 13 18.0 0.5 17.0 1.0 14 16.0 0.0 15.0 7.0 15 9.0 0.0 17.5 0.0 16 21.0 0.0 28.0 0.0 17 21.5 1.0 34.0 1.0 18 11.0 2.5 16.0 1.0 19 15.5 1.0 240 1.0 20 14.5 1.0 22.0 5.0 21 10.0 0.5 15.5 4,0 22 20.0 0.0 19.5 1.5 23 15.5 0.0 22.0 10 24 12.0 0.0 19.0 2.0 25 11.0 0.5 16.5 4.5 26 13.0 0.0 20.0 0.0 27 14.5 0.5 17-5 3.0 28 15.5 1.0 19.5 2.0 29 14.0 1.0 19.0 0.5 30 29.0 0.5 34.5 0.5 31 18.0 0.5 20.0 00 32 14.5 0.0 20.0 0.0 33 9.5 0.0 14.5 0.0 34 26.0 0.5 26.0 4.5 35 16.0 0.0 24.0 3.0 36 12.5 1.0 18.5 1.0 Table III gives in detail the data of the Control group in the two test series. This table is constructed in exactly the same way as Table II. In the first column the subjects are in- dicated by the numbers 1 to 40. It reads thus : In the Spanish a-t Word Test Subject No. 1 has an itial performance of 10.5 cancellations per minute and 0.5 errors, and a final per- formance of 12.0 cancellations per minute and 0.0 errors- in MATERIALS AND PROCEDURE. 37 TABLE III. DATA OF THE CONTROL GROUP IN 1HE TEST SERIES Spanish a-t Word Test Spanish e-s Word Test Initial Final Initial Final Sub. Can. Er. Can. Er. Can. Er. Can. Er- 1 10.5 0.5 12.0 0.0 11.0 0.0 11.5 0.0 2 10.5 10 11.0 0.0 7.5 4.5 11.0 0.0 3 10.0 0.5 13.5 0.0 10.5 0.0 12.5 0.0 4 6.5 3.0 11.0 2.5 11.0 1.5 14.5 0.5 5 10-0 2.0 11.0 0.0 10.5 0.5 11.5 0.0 6 10.5 1.0 13.0 1.0 10.0 1.5 13.0 1.0 7 8.5 1-5 10.0 1.0 11.5 0.5 11.5 0.0 8 9.5 1.5 10.0 0.0 11.0 0.0 10.0 0.0 9 10.5 0.5 12.0 0.5 10.5 0.0 11.0 0.0 10 110 1.5 11.5 1.5 12.0 0.0 12.0 1.5 11 8.0 1.0 10.5 0.0 11.0 0.0 11.0 1.0 12 9.5 0.5 11.0 0.5 9.5 0.5 10.5 1.0 13 10.5 1-5 12.0 0.5 12.0 0.5 12.5 0.0 14 10.0 1.0 12.0 1.0 12.5 0.5 14.5 0.0 15 7.5 2.5 11.5 1.5 10.0 2.0 11.5 2.5 16 9-5 0.5 12.0 0.0 10.0 0.5 11.5 0.0 17 7.0 1.0 9.0 0.5 9.0 2.5 10.5 0.0 18 10.0 0.0 11.0 1.5 10.5 0.5 10.0 2.0 19 12.0 10 12.5 0.5 12.0 1.5 12.0 0.0 20 10.0 0.0 10.5 1.5 8.0 0.0 11.0 0.0 21 8.5 1.5 10.5 0.5 10.0 1.5 10.5 0.5 22 100 0.0 11.5 0.0 11.0 0.5 12.0 0.0 23 7.5 1.5 13.0 2.0 11.0 0.5 15.0 4.5 24 10.0 0.0 11.5 2.0 12.5 0.0 15.0 0.5 25 9.5 0-5 12.5 0.0 10.0 0.5 11.0 0.0 26 10.0 0.0 11.0 0.0 10.5 1.0 10.5 1.0 27 9.0 1.0 8.5 6.0 10.0 1.0 10.5 2.0 28 9-5 2.0 10.0 0.5 11.0 0.0 12.0 0.0 29 7.5 0.5 11.0 0.0 8.5 0.5 10.5 0.0 30 9.5 1.0 11.0 1.0 10.0 1.5 12.0 0.0 31 12.0 0-5 13.0 0.5 11.5 0.0 12.5 1.0 32 10.0 0.5 11.5 1.5 11.0 0.0 12.0 0.0 33 7.5 1.5 10.5 0.5 9.5 1.5 10.0 1.0 34 2-5 7.0 9.0 0.5 10.0 1.0 8.5 4.5 35 6.5 4.5 9.5 0.0 6.5 4.5 12.0 11.0 36 8.0 1.5 10.0 0.0 11.0 0.0 11.0 0.0 37 10.5 0-5 12.0 0.0 12.0 0.0 15.0 0.5 38 10.0 0.5 14.0 2.0 14.0 1.5 18.0 1.5 39 9.5 1.5 9.5 2.0 11.0 0.0 11.5 0.0 40 10-5 0.5 13.0 1.0 14.0 0.0 16.5 0.5 This table shows the initial and final performance of each member of the control group in each of the seven tests, in terms of number of cancellations and errors per minute. 38 TRANSFER EFFECTS OF PRACTICE. TABLE III (Continued). DATA OF THE CONTROL GROUP IN THE TEST SERIES Letter Test a and t Letter Test e and s Initial Final Initial Final Sub. Can . Er. Can . Er. Can . Er. Can, . Er- 1 29.5 1.5 32.5 1.5 41.5 1.5 38.0 4.0 2 28.5 1.0 26.5 2.5 36.5 0.0 38.0 0.5 3 22-5 1.0 34.5 1.5 36.5 0.5 42.0 0.0 4 29.0 4.5 34.0 2.0 46.5 1.5 54.5 1.0 5 19.5 0.5 22.0 0.0 32.0 0.5 35.0 0.0 6 29.0 10 37.5 2.5 40.0 0.5 49.5 1.5 7 32.5 2,5 34.0 3.0 42.0 2.0 55.5 1.0 8 25.0 1.0 24.0 0.5 41.0 1.0 36.0 0.0 9 26-5 0.5 30.5 0.5 38.0 0.5 34.0 0.0 10 32.0 1.5 32.5 0.5 40.5 2.0 41.5 0.0 11 24.5 2.0 32.0 2.5 36.0 1.5 45.0 0.5 12 27.0 00 27.0 1.5 38.0 0.5 42.0 0.0 13 29.0 0.0 30.5 2.0 40.5 0.5 49.5 0.5 14 32.5 1.0 38.5 0.5 45.5 2.0 58.0 0.0 15 230 5.0 31.0 9.0 37.0 1.0 47.5 0.5 16 25.5 2.0 27.5 0.5 32.5 1.0 33.5 1.5 17 18.5 2.0 16.5 1.0 30.0 1.5 31.5 0.5 18 17.5 2-5 25.0 3.5 32.0 0.0 42.5 0.0 19 41.0 0.5 41.5 1.0 62.5 3.5 63.5 0.5 20 20.0 0.5 23.5 0.0 31.0 1.5 34.0 0.0 21 27-0 1.5 28.0 2.5 36.0 1.5 35.5 1.0 22 22.0 1.5 27.0 2.5 35.5 0.5 41.5 1.0 23 30.0 2.5 39.0 3.0 40.0 2.0 55.5 2.0 24 29.5 10 40.5 3.0 55.0 3.0 64.5 1.5 25 25.5 1.0 31.5 3.5 43.5 2.5 42.0 0.5 26 23.0 0.5 26.5 0.0 39.0 0.5 35.0 0.0 27 240 3.0 33.5 5.5 35.5 1.0 46.5 0.5 28 20.0 3.5 29.0 7.0 35.0 0.5 45.5 1.0 29 22.5 0.0 31.5 3.5 34.0 1.5 48.0 0.5 30 30.5 60 36.0 1.0 41.0 0.5 49.0 2.0 31 25.0 1.0 35.5 0.5 34.0 0.0 39.0 0.0 32 25.0 3.5 29.0 0.0 34.5 2.5 43.0 1.5 33 230 6.5 26.0 4.0 33.5 3.0 38.5 1.5 34 19.5 3.5 30.5 2.5 35.0 2.0 36.0 3.0 35 22.0 5.0 31.5 1.5 29.0 4.5 40.0 2.0 36 22.5 0-5 23.5 1.0 34.0 0.5 42.0 0.5 37 24.0 2.5 30.5 0.5 42.0 1.0 45.5 0.0 38 34.5 2.0 44.5 1.5 56.0 3.0 74.0 1.5 39 27-5 4.5 31.5 3.0 45.5 0.0 52.0 3.5 40 31.5 5.5 33.5 2.5 43.5 1.0 61.5 1.0 the Spanish e— s Word Test an initial performance of 11 cancellations per minute and 0.0. errors, and a final perfor- mance of 11.5 cancellations per minute and 0.0 errors, and so MATERIALS AND PROCEDURE. 39 TABLE III (Continued). DATA OF THE CONTROL GROUP IN THE TEST SERIES Capital A Test Initial Final Sub. Can. Er. Can. Er. 1 27.5 2.5 42.5 0.0 2 37.5 00 37.5 0.5 3 31.0 0.5 43.0 0.5 4 35.5 0.0 48.5 0.5 5 35-0 0.0 40.0 0.0 6 40.5 1.0 48.5 0.5 7 41.5 2.0 40.5 1.5 8 34.0 10 43.5 0.0 9 44.0 0.0 46.0 0.0 10 38.5 3.0 39.5 0.5 11 34-5 0.0 41.0 0.5 12 36.5 0.0 44.5 0.5 13 43.5 0.0 47.0 0.5 14 36.5 0-0 53.0 0.0 15 31.5 0.5 44.0 0.0 16 29.5 1.0 36.0 0.0 17 300 0.0 33.5 0.5 18 39.0 2.0 41.0 0.5 19 42.5 2.0 57.5 1.0 20 33-5 0.5 45.0 0.0 21 29.0 0.0 38.0 0.0 22 32.0 0.0 42.0 0.0 23 50.0 0.5 56.0 1.5 24 39-5 0.5 62.0 1.0 25 36.0 5.5 42.0 0.0 26 35.0 0.0 43.5 0.0 27 34.5 0.5 47.5 0.5 28 270 0.0 38.5 0.5 29 33.5 0.0 45.0 0.0 30 46.0 1.5 54.0 1.0 31 38.0 0.5 47.0 0.0 32 340 1.5 42.5 0.5 33 32.5 4.5 47.5 0.0 34 34.5 5.0 46.0 0.5 35 31.5 1.5 45.0 0.5 36 300 4.5 36.0 1.5 37 37.5 0.5 46.0 0.0 38 44.0 3.0 62.0 1.5 39 39.5 2.5 54.0 1.5 40 39-0 0.0 55.5 1.0 Capital B Test Initial Final Can. Er. Can. Er- 21.0 0.5 24.5 0.0 20.0 0.5 19.0 0.0 19.5 0.0 28.0 1.0 23.0 2.5 27.0 4.0 19.0 0.0 20.0 0.0 21.0 1.0 28.5 1.5 23.5 1.5 28.5 4.0 19.0 0.5 20.5 0.5 21.5 0.0 20.0 0.0 18.0 0.0 • 17.0 3.5 16.0 1.0 19.5 0.0 20.5 0.5 23.5 0.0 18.5 0.5 26.0 0.5 20.0 0.0 24.0 0.0 18.5 1.5 21.0 0.5 17.5 1.5 19.0 0.0 15.5 0.5 18.5 0.0 16.5 0.0 22.0 0.5 24.5 0.5 23.0 1.0 17.5 0.0 19.0 1.0 16.0 1.0 17.0 1.0 18.5 0.0 20.0 00 25.0 2.5 32.5 0.5 23.0 0.0 33.0 0.0 20.0 0.0 24.0 1.5 19.5 0.5 22.5 00 17.0 1.0 19.5 1.5 14.5 4.0 23.5 1.0 15.5 0.5 19.5 1.5 22.5 1.5 26.5 1.5 16.0 1.0 23.5 1.0 18.5 0.5 21.5 0.0 18.0 2.0 20.5 1.0 19.0 0.5 25.5 0-5 17.5 1.5 26.0 5.0 20.5 0.0 21.5 0.5 24.0 0.0 28.0 0.0 27.5 2.0 38.0 1-5 23.5 2.0 29.5 2.5 26.0 0.5 29.5 2.0 on with the other tests. The data for each of the remaining subjects are to be read in the same manner. 40 TRANSFER EFFECTS OF PRACTICE. TABLE III (Continued). DATA OF THE CONTROL GROUP IN THE TEST SERIES 4 and 7 Group Test Initial Fina! Sub. Can. Er. Can. Er. 1 17.5 0.0 18-5 0.0 2 16.0 0.0 20.5 3 13.0 0.0 19.0 0.5 4 12.0 2.0 17.0 1.0 5 14.5 0-0 16.5 0.0 6 18.0 0.0 20.5 0.5 7 16.0 0.5 17.0 0.0 8 11.5 1.0 12.0 0.0 9 17.5 0.0 19-0 0.0 10 17.5 0.5 19.0 0.0 11 15.0 1.0 20.0 0.0 12 15.5 0.0 18.0 0.0 13 16.0 0-5 19.5 0.0 ' 14 18.0 0.0 21.5 1.0 15 16.0 0.0 18.0 0.0 16 11.0 0.0 16.0 0.0 17 9.5 0.0 13-0 0.0 18 14.0 0.5 17.0 0.5 19 22.5 0.0 20.5 0.5 20 11.0 0.0 15.5 0.0 21 14.5 00 18.5 0.0 22 14.0 1.0 17.0 0.0 23 17.0 0.5 21.5 0.5 24 16.0 1.0 24.0 0.0 25 18.5 0.0 20-5 1.5 26 14.0 0.0 17.0 0.0 27 12.5 1.0 17.5 1.0 28 9.5 1.0 15.0 1.0 29 10.0 00 17.0 0.5 30 15.0 0.0 20.0 0.0 31 17.5 0.5 210 0.0 32 15.5 0.5 17.0 0.0 33 13.0 0.5 16-5 0.5 34 8.5 6.0 17.5 0.5 35 11.5 0.5 18.0 2.5 36 14.5 0.5 19.5 0.0 37 18.0 00 23.5 0.5 38 24.0 0.0 31.5 1.0 39 14.0 0.5 21.5 1.0 40 15.5 1.0 18.5 0.5 MATERIALS AND PROCEDURE. 41 An inspection of Tables II and III shows the differences in performance between the subjects of the control group and those of the practice group. While performance is recorded in cancellations and errors, yet a discussion of change in per- formance requires that errors be treated with respect to their accompanying cancellation scores. When this is done, we can say whether a subject has changed not only in speed of cancel- lation but in per cent, of accuracy. Attention is directed to the fact that the change in the control group in both speed and accuracy measures the progress due to direct practice in the tests themselves. Therefore, whatever differences we may note between the performance of the practice group and that of the control group must be due to the transfer effects of the training in the practice series. While a complete discussion and explanation of the transfer effects is reserved for the next chapter, yet it may prove of interest to the reader to consider briefly some of the differences between the two groups without attempting to draw final conclusions. Such a study will indicate what one may reasonably expect when the data are more elab- orately treated. In the Spanish a — t Word Test 80 per cent, of the control group and 72 per cent, of the practice group gained in both speed and accuracy. Fifteen per cent, of the control and 28 per cent, of the practice lost in accuracy while gaining in speed. The gains of the control group range from 5.5 cancellations to a loss of 0.5, and those of the practice group from 5.0 to 14.0. In the Spanish e — s Word Test, on the other hand, 40 per cent, of the control and 8 per cent, of the practice gained in both speed and accuracy, while 5 per cent, of the control and 16 per cent, of the practice lost in both. Twenty-two per cent, of the control and 61 per cent, of the practice lost in accuracy while gaining in speed. Five per cent, of the control and 11 per cent, of the practice lost in accuracy without changing in speed. It is significant that only 32 per cent, of the control and 88 per cent, of the practice lost in accuracy, while 62 per cent, of the control and 69 per cent, of the practice gained in speed. In the a and t Letter Test 60 per cent, of the control and 80 per cent, of the practice gained in both speed and accuracy, while 2.5 per cent, of the control and none of the practice lost in both. Thirty per cent, of the control and 16 per cent, of the practice lost in accuracy while gaining in speed. Five per cent, of the control and 3 per cent of the practice gained in accuracy 42 TRANSFER EFFECTS OF PRACTICE. W dl o y w <1 o ^ > « iH tH O iH o < fa Q > hJ o m « <: u t» s s « < 2 Oh -h' i-H C- i-H 00 CO T-H (N tH •r^-tj'P -r: m'P ft > r/5 >^ C P. > 'Em s o c MATERIALS AND PROCEDURE. 43 while losing in speed. In the e and s Letter Test 67.5 per cent, of the control and 30.5 per cent, of the practice gained in both speed and accuracy, while 2.5 per cent of the control and 25 per cent, of the practice lost in both. Seventeen and five-tenths per cent, of the control and 33.3 per cent, of the practice lost in accuracy, while gaining in speed. Twelve and five-tenths per cent, of the control and 11 per cent, of the practice lost in speed while gaining in accuracy. The reader will note that 85 per cent, of the control and 63.8 per cent, of the practice gained in speed while 20 per cent of the control and 58.3 per cent of the practice lost in accuracy. In the Columbia A Test 72.5 per cent, of the control and 67 per cent, of the practice gained in both speed and accuracy. Twenty-two and five-tenths per cent, of the control and 33 per cent, of the practice lost in accuracy while gaining in speed. In the Thorndike B Test 57.5 per cent, of the control and 33 per cent, of the practice gained in both speed and accuracy, while 5 per cent, of each group lost in both. Thirty-two and five- tenths per cent, of the control and 61 per cent, of the practice lost in accuracy, while gaining in speed. Attention is called to the fact that only 37.5 per cent, of the control, and 66 per cent, of the practice lost in accuracy. In the Number Group Checking Test 75 per cent, of the con- trol and 42 per cent, of the practice gained in both speed and accuracy, while none of the control and 6 per cent, of the prac- tice lost in both. Twenty-five per cent, of the control and 50 per cent, of the practice lost in accuracy while gaining in speed. Table IV is derived from Table II and shows the average performance in speed and accuracy of the practice group in the initial and final test series. The seven tests are listed in the first column. Under "Initial Test Series" are four columns ; the numbers in the first indicate the average number of cancel- lations per minute, while the numbers in the second represent the corresponding probable errors ; in the third and fourth col- umns the corresponding average per cents, of accuracy with their probable errors are presented. The fou^' columns under "Final Test Series" are similarly arranged. Under "Differ- ences" there are four columns. The first contains the differ- ences between the final averages and the initial averages, while the corresponding probable errors are in the second column. The third column contains the differences between the final and initial per cents, of accuracy, with the corresponding probable errors in the fourth column. 44 TRANSFER EFFECTS OF PRACTICE. ^O T-1 < pq <: Q o ^ Z Z > < o W <: oa (M C~ [^ (M (N D-; H 5 d^ d d d M 03 H H o 00 0-. H B o CO Ci 1 ^ >' (>i 1—1 r* < t> H W S OQ ^ w 00 ic 1-1 11^ d d d S °§ ^ £ ^ ^ !> CO CO a-. CO as 5 S 3 H S p > O ^ H • ■^ 00 t- |j^ i-H lo r^ < CM d d d « w a/ g O H _> E 3 o "t o 05 .2 •^ o 05 H (^ r-l c4 d S 5 _a) > 1— 1 I-H CO 8| 3 H H > W < c- ^ CVl W J H 1-H 1— ( lO H < CLh" d d d O H g g O o o ra >' < d d 1—1 d M w Eh 'S 1 ^ 'a 7'2 2- H c3 ■"1 -;^ Cs tH (N CC IT '—• ■— ' ;-i _ TO CD 1-3 I* a/ 3 J 10 f W !^ SO Ji- 30 J5- 1 1 l^ i f w rt /r 30 1 1 1 1 r-- ^ 1 1 1 1 1 eJ3 .... 1 1 1 1 ■ Figure XI — Showing the Relation Between the Practice and Control Groups in the Woodworth and Wells Number-Group Checking Test. 64 TRANSFER EFFECTS OF PRACTICE. of the practice caused an inferiority in speed of cancellation. In gross speed the practice group has a superiority of 3.32 words per minute. What, then, were the transfer effects in this test ? Certainly not an inferiority in speed of cancellation, as indicated by the corrected results. Taking the facts as they actually occur, we note that the training of the practice series caused a superiority in gross speed, and an inferiority in percent of accuracy, and no difference in speed of cancel- lation. Surely, no one would say that a gross speed of 3.32 words per minute and an inferiority in accuracy of 16 per- cent with no difference in speed of cancellation, means the same thing as an inferiority of 3.28 cancellations per minute. In the a and t Letter Test, corrected results give the practice group a superiority of 6.62 cancellations per minute, instead of 4.91 with a superiority of 6 percent in accuracy, and in gross speed of 3.25 words per minute. Thus, in this test, as in the Spanish e — s Word Test, corrected results tend to con- ceal the actual transfer effects of the practice, for instead of a superiority in cancellation only, the practice group has a superiority in gross speed and in speed and accuracy of can- cellation. In the e and s Letter Test, corrected results give the practice group an inferiority of 4.79 cancellations per minute, while the actual results indicate that the practice caused an inferiority, not only in the speed of cancellation, but in accuracy of performance and in gross speed. The trans- fer effects in these two tests are in contrast whether the results are corrected or not. However, it is plainly evident that the uncorrected results give a clearer and truer presenta- tion of the transfer effects than the corrected results, for instead of the practice causing an inferiority of 4.79 cancella- tions per minute, it caused an inferiority in gross speed of 2.94 words per minute, in speed of cancellation 3.98 per min- ute, and in accuracy of performance 3 percent. Corrected results in the capital A Test indicate that the practice group has a superiority of 3.5 cancellations per min- ute, while with the uncorrected results it has a superiority of 3.99. Since there is no reliable difference in the gains in accuracy made by the two groups, the transfer effects are manifested in speed of cancellation, and in gross speed which was 4.53 words per minute. In the capital B test, corrected results indicate a superiority of 0.14 cancellations per minute / INTERPRETATION OF RESULTS. 65 for the practice group, while uncorrected results give an actual though unreliable superiority of 1.13 cancellations, and a reli- able inferiority in accuracy of 3 percent. In these two tests corrected results indicate positive transfer effects in the Capital A Test, and none at all in the Capital B Test. On the other hand, the actual results indicate transfer effects in the Capital A Test in a superiority of actual cancellations, and in gross speed ; and in the Capital B Test an inferiority in accuracy of 3 percent., and a superiority in gross speed of 2.13 words per minute. In the Number Group-Checking Test corrected results indicate an unreliable inferiority of 0.32 cancellations per min- ute, on the part of the practice group, while the actual results show a reliable inferiority in accuracy of 7 percent, and in gross speed a superiority of 2.46 words per minute. The writer feels confident that this presentation and discussion of the results of correcting for errors justifies his objection to a procedure so arbitrary; a procedure which tends to con- ceal the actual transfer effects of the practice. Section 3. Summary and Conclusion. The transfer effects of the practice are revealed in the superiority or inferiority of the practice group. These effects may manifest themselves in three ways : In gross speed, that is, total ground covered cancellations plus errors ; in speed of actual cancellation; and in accuracy of performance. Pro- longed practice in cancelling in Enghsh prose words contain- ing both a and t resulted in an improvement in gross speed from an average of 129.6 words per ten minute period to an average of 277.0, or a gain of 113 percent; in speed of can- cellation from an average of 102.6 cancellations per period to an average of 266.5, or a gain of 159 percent ; and in accuracy of cancellation from 79 percent to 96 percent, or a gain of 17 percent. In gross speed the practice group has a superiority in six of the seven tests and an inferiority in one. In speed of cancellation the transfer effects that are reliable are posi- tive in three tests ; being 67 percent in the Spanish a— t Word Test, 15 percent in the a and t Letter Test, and 10 percent in the capital A Test ; and negative in one, being 9 percent in the e and s Letter Test. In percent of accuracy the transfer effects are positive in one test and negative in four tests, and without 66 TRANSFER EFFECTS OF PRACTICE. reliable influence in the other two. In only one test does the practice group have a superiority in both speed and accuracy of cancellation — the a and t Letter Test; and in only one test is it inferior in these two respects — the e and s Letter Test. In each of the seven tests there is evidence of the transfer effects of the practice, and these are pronounced in four tests ; in two there is facilitation and in two there is interference. The two Spanish Word Test situations were identical; in the one the stimulus provoking cancellation was identical in form and content elements with that in the practice situation. These were the conditions favoring the largest amount of positive transfer or facilitation. In the other, the stimulus provoking cancellation was one of the stimuli to be neglected in the practice situation, while the practiced stimulus was neglected. Such conditions favored interference. The two Letter Test situations were identical; in the one the stimuli provoking cancellation were identical with the content ele- ments of the stimulus cancelled in the practice situation. Such conditions favored facilitation. In the other, the stim- uli provoking cancellation were unlike the content elements of the stimulus cancelled in the practice, and like the content elements of many stimuli neglected in the practice. Such con- ditions favored interference. In the two Capital Letter Tests the situations were somewhat unlike and the tasks different. In the Capital A Test the stimulus provoking cancellation was the same in name as one of the stimuli determining the can- celling in the practice situation. The result was a slight facil- itation in the form of speed of cancellation. In the Capital B Test, the stimulus provoking cancellation was the same in name and rarely in form as one of the stimuli neglected in the practice situation, while there were many stimuli to be neg- lected that were the same in name and rarely in form as the stimuli determining cancelling in the practice situation. The result was a slight interference in the form of a decrease in accuracy. In the Number Group-Checking Test the stimulus provoking the cancellation response was the presence of 4 and 7 in a number group, while in the practice, it was the presence of a and t in a letter group. The result was interference in the form of a decrease in accuracy. The tendency toward speed acquired through the practice was stronger than the tendency toward accuracy, for in six INTERPRETATION OF RESULTS. 67 of the seven tests, the practice group has a superiority in gross speed. However, in only three of the six, does this gross speed eventuate in faciUtation ; in the other three it is accompanied by interference in the form of a decrease in ac- curacy ; and in the remaining one was not sufficient to prevent a loss in gross speed and in speed of cancellation, but was suf- ficient to cause a decrease in accuracy. Section 4. Generalized Conclusions. In this investigation the writer endeavored to discover under what conditions, to what extent, and in what direction, prolonged cancelling in one situation affects cancelling in other situations. The results justify the following conclu- sions : 1. That the extent and direction of the transfer effects depend, first, on the practice situation and the task performed ; and second, on the test situation and the task performed. 2. When the test situation is such that the task per- formed is identical with the task in the practice, the extent of the transfer effects is largest and in a positive direction in the form of speed. 3. When the test situation is such that the task per- formed is the reverse of that performed in the practice, the extent of the transfer effects is large, causing interference in the form of a decrease in accuracy. 4. When the test situation is such that the task per- formed is identical as to content elements, with that of the practice, the extent of the transfer effects is comparatively small and in a positive direction, both in speed and accuracy. 5. When the test situation is such that the task per- formed is the reverse as to content elements of that of the practice, the transfer effects are small causing interference both in speed and accuracy. 6. When the test situation is such that the task per- formed is somewhat similar as to content elements to that of the practice, the transfer effects are slightly positive in the form of speed. 7. When the test situation is Kuch that the task per- formed is somewhat the reverse as to content elements of that in the practice, the transfer effects are slightly negative in the form of a decrease in accuracy. 68 TRANSFER EFFECTS OF PRACTICE. 8. When the test situation contains none of the content elements of the practice situation but is such that the task performed is somewhat similar as to form with that of the practice, the transfer effects are negative, causing a decrease in accuracy. 9. Finally, this investigation plainly teaches the import- ance of scoring for accuracy as well as speed. VITA Melvin Albert Martin. Born at Petersburg, Virginia, October 18, 1871. Early education received in the Public Schools of Petersburg and at the Virginia Polytechnic Insti- tute A.B., Richmond College, Richmond, Virginia, 1898; A M Columbia University, 1905. Special student in the Uni- versity of Chicago, 1898-1900; Head Master, Mossy Creek Academy, Virginia, 1900-1901; Principal Southside Female Institute, Burkeville, Virginia, 1901-1902; professor of Philos- ophy and Physics, Woman's College, Richmond, Virginia, 1902-1904; graduate student, Columbia University, 1904- 1905- Dean and professor of Philosophy and Education, Wom'an's College, Richmond, Virginia, 1905-1913; professor of Psychology, in the Summer School of the University of Vir- ginia, 1911-14; graduate student, Columbia University, 1913- 1915; assistant in Psychology, Columbia University, 1914-1915. LIBRARY OF CONGRESJ 021 350 395 9