The Long‐run Effect of 9/11: Terrorism, Backlash, and the Assimilation of Muslim Immigrants in the West


THE LONG-RUN EFFECT OF 9/11: TERRORISM, BACKLASH,
AND THE ASSIMILATION OF MUSLIM IMMIGRANTS IN THE

WEST*

Eric D. Gould and Esteban F. Klor

This article investigates whether the 9/11 attacks affected the assimilation rate of Muslims in the US.
Terror attacks by Islamic groups are likely to induce a backlash against Muslims, thereby raising their
costs of assimilation. We find that Muslim immigrants living in states with the sharpest increase in
hate crimes also exhibit: greater chances of marrying within their own ethnic group; higher fertility;
lower female labour force participation; and lower English proficiency. These findings shed light on
the increasing use of terror and concurrent rise in social tensions surrounding Muslim immigrants in
the West.

The terror attacks on the US in 2001 impacted the world in many ways. The shock and
the loss of life were the most acute and immediate effects. Soon after, the economy was
affected by the damage, disruption of air travel and increasing uncertainty in the
security situation and world financial markets. Major military campaigns were
subsequently launched in Afghanistan and Iraq, creating new tensions and alliances
between countries. It is safe to say that the 9/11 attacks transformed the economic and
diplomatic landscape on a global scale. In this article, we examine how the 9/11 attacks
affected the Muslim community within the US.

In particular, we investigate the general idea that terror attacks by radical Islamic
groups are likely to induce a backlash against the Muslim community as a whole,
raising their costs of assimilation. Evidence for a backlash after 9/11 is supported by
the data on hate crimes against Muslims, which went from 28 to 481 reported incidents
from the year 2000 to 2001. A similar backlash against Muslims took place all across
Europe (�Aslund and Rooth, 2005; Hanes and Machin, 2012; Sch€uller, 2012). We
empirically examine whether this backlash slowed the rate of assimilation by exploiting
variation across states in the number of hate crimes against Muslims in the wake of the
9/11 attacks. Using individual-level data from the Census and American Community
Surveys from 1990–2010, our results show that Muslim immigrants living in states which
experienced the sharpest increase in hate crimes after 9/11 also exhibit:

(i) greater chances of ‘intra-marriage’ (marrying someone who also originates
from a Muslim country);

(ii) higher fertility;
(iii) lower female labour force participation; and
(iv) lower English proficiency.

* Corresponding Author: Esteban F. Klor, Department of Economics, Hebrew University, Mt. Scopus,
Jerusalem, 91905, Israel. Email: eklor@huji.ac.il.

We thank the editor, Kjell Salvanes, two anonymous referees, Moshe Hazan, Daniele Paserman, Jacob
Shapiro and participants at numerous seminars and conferences for helpful comments and suggestions. We
thank The Maurice Falk Institute for Economic Research and the Israeli Science Foundation (grant No. 408/
12) for financial support.

[ 2064 ]

The Economic Journal, 126 (November), 2064–2114. Doi: 10.1111/ecoj.12219 © 2014 Royal Economic Society. Published by John Wiley & Sons, 9600

Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.



Interestingly, the higher rate of intra-marriage is coming at the expense of marrying
outside of the ethnic group, rather than a general increase in the marriage rate. All of
these patterns are consistent with a less-assimilated outcome, since Muslim countries
are characterised by very low rates of female labour force participation and high
fertility rates compared to natives in the US.

Overall, our findings show that the 9/11 attacks induced a backlash which made the
Muslim community in America more cohesive and traditional. In order to attribute a
causal interpretation to our results, it is important to note that the state-level increase
in hate crimes against Muslims after the 9/11 attacks was not correlated with the pre-
existing state-level trend in any of the assimilation outcomes, or with the characteristics
of Muslim immigrants living in the state prior to the 9/11 attacks. In addition, the
results are robust to the inclusion or exclusion of a rich set of personal and state-level
characteristics, including state-level hate crimes against blacks, Jews and homosexuals.
These findings support our identifying assumption that variation in the extent of the
backlash across states can be considered exogenous. Moreover, we conduct a placebo
analysis by showing no effect of hate crimes on the marriage decisions of older Muslim
immigrants, who had largely already made their marriage decisions before the 9/11
attacks, and by showing that no other immigrant group exhibited a slower rate of
assimilation in response to hate crimes against Muslims. These findings lend even
more support for a causal interpretation for our findings that the 9/11 attacks
increased the ethnic identity of the Muslim community in the US.

There is no systematic empirical research on the question of whether the backlash
against Muslim Americans after 9/11 affected their rate of assimilation in the US. The
lack of research on this subject is surprising, given the increasing social and political
tensions surrounding the assimilation of a large influx of Muslim immigrants to
Western European countries and North America, and a concurrent increase in the use
of large-scale terror attacks on Western cities. The most related literature has been
conducted by historians, sociologists, journalists and Islamic scholars, who claim that
Muslims and their communities in the US underwent substantial changes after the
9/11 attacks (Abdo, 2006; Barrett, 2008; Bakalian and Bozorgmehr, 2009).1 These
studies argue that American Muslims, who felt under attack by the government,
general public and the media in the aftermath of 9/11, sought refuge in their religion
and community to withstand the backlash. According to these authors, the 9/11 attacks
spurred a renewed sense of solidarity among Muslims and a religious revival in the face
of widespread criticism of Islam.

However, the existing literature suffers from several shortcomings. The empirical
evidence is based on a small number of selected interviews, rather than a systematic
analysis of a large, representative sample of Muslims in the US. Individuals who agreed
to be interviewed may not have opinions which reflect the average person’s experience,
and an individual’s thoughts and opinions may not match their actions. In addition,

1 There is also a large literature studying the immigration of Muslims to Western European countries and
the potential for clashes with the native population. Some of these studies speculate that the backlash in the
aftermath of terror attacks is associated with the radicalisation of Muslim immigrants but do not conduct a
systematic analysis of that connection (Sniderman and Hagendoorn, 2007; Murshed and Pavan, 2009).
Others studies focus on the increasing political power of Muslim immigrants in European countries as a
major source of conflict between immigrants and the native European population (Dancygier, 2010).

© 2014 Royal Economic Society.

[ N O V E M B E R 2016] TERROR AND ASSIMILATION OF MUSLIMS IN THE WEST 2065



the existing literature does not attempt to establish a causal connection between the
backlash and assimilation and, therefore, cannot determine whether the changes in
the Muslim community after 9/11 were due to the attacks themselves or were part of a
pre-existing trend in their assimilation patterns.

Our article is the first systematic empirical analysis of how the Muslim community in
the US. reacted to the 9/11 attacks. We present the first evidence that the backlash
against Muslims slowed their rate of assimilation, as reflected by higher rates of intra-
marriage and fertility, and lower rates of female labour force participation and English
proficiency. In this manner, our findings suggest that terror attacks against Western
targets may have a long-term political and socio-economic impact, by creating a more
ethnically cohesive Muslim community in this generation and also the next.

The idea that terror groups instigate a backlash is not new in the theoretical
literature on political conflict (Rosendorff and Sandler, 2004, 2010; Siqueira and
Sandler, 2006; Bueno de Mesquita and Dickson, 2007; Baliga and Sj€ostr€om, 2012).
However, this literature focuses on the backlash against the country or territory where
the perpetrators reside. In the context of 9/11, this is consistent with the US attacking
Afghanistan. Our findings raise the possibility that terror groups may also intentionally
induce a backlash on persons of a similar ethnic origin in the targeted country, in
order to decrease their rate of assimilation.

The remainder of the article is organised as follows. The next Section briefly surveys
the literature on Muslim immigrants to the US, focusing on their experiences before
and after the 9/11 attacks. Section 2 presents the data on Muslim assimilation and hate
crimes. Section 3 presents our empirical strategy, followed by our main results for
‘intra-marriage’ in Section 4 and other assimilation outcomes in Section 5. Section 6
examines other immigrant groups and Section 7 concludes with a discussion of the
broader implications of our results.

1. Background and Related Literature

According to a recent survey by the Pew Research Center (2011), the vast majority of
Muslims in the United States were born abroad, with only 37% native-born. (Peck, 2004
finds a very similar number – 66% are foreign born.) Among Muslim immigrants, 71%
arrived after 1990. So, the Muslim population in America is relatively new and mostly
foreign-born, with almost all of the immigrants arriving after the repeal of country-
based immigration quotas in 1965. This policy change created a strong positive
selection regarding the characteristics of immigrants that were allowed into the US.
Consequently Muslim immigrants to America, mostly from the Middle East and South
Asia, were substantially more educated and wealthier than their counterparts that
immigrated to Europe. These characteristics, along with their pragmatism and drive to
succeed, led to a relatively fast integration into the labour market and suburban
communities of the US (Abdo, 2006).

Although Muslim immigrants in the West tend to be less assimilated than other
immigrant groups (Bisin et al., 2008), several scholars noted that Muslims in the US
were on the path of assimilation or ‘Americanization’ before the 9/11 attacks
(Haddad and Esposito, 2000; Abdo, 2006). These immigrants, while still religiously
identified as Muslims, started to view traditional practices in their home country as out-

© 2014 Royal Economic Society.

2066 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



dated and out of place, and adopted customs more similar to their American peers
(Hasan, 2000). Some of them even anglicised their names, intermarried with non-
Muslims and began to speak Arabic less frequently (Goldwasser, 2000; Kulczycki and
Lobo, 2002).

Several scholars argue that the 9/11 terror attacks brought this process of
Americanisation to a sudden halt. They suggest that the substantial increase in hate
crimes, harassment, racial profiling and discrimination against Muslims in the
immediate aftermath of 9/11 started a process of re-Islamisation. At the individual
level, Muslim Americans renewed their interest in the Quran and traditional Islamic
practices (Bryan, 2005; Abdo, 2006). One example is the growing number of young
women wearing the hijab (headscarf) as an icon of Islamic pride and defiance to a
hostile environment (Haddad, 2007). Another example is the growing demand for
Islamic educational and recreational programmes for children and young adults,
which led to the creation of numerous Islamic day schools and Sunday schools across
the US (Schmidt, 2004), as well as the spread of additional chapters of the Muslim
Students Association at numerous universities (Abdo, 2006).

These studies also claim that Muslims in search of community and a safe haven from
a more hostile environment started to attend mosque services more frequently
(Haddad, 2007; Bakalian and Bozorgmehr, 2009). In turn, mosques and Islamic
community centres took a more central role in the lives of Muslims after 9/11, as they
ceased to be only places of worship and became social centres. Women’s groups began
to flourish (Haddad, 2007). Mosques and Islamic centres also started to play a more
active political role, with the Muslim community increasing their participation in
electoral politics in every election since 2001 (Bakalian and Bozorgmehr, 2009).

While the studies cited above present a poignant picture of how 9/11 affected the
Muslim community in the US, the conclusions are based entirely on anecdotal
evidence collected from extensive interviews. It is possible that individuals who agreed
to be interviewed hold views and underwent experiences that were not representative
of the Muslim community at large. In addition, these studies do not attempt to
establish a causal connection between the backlash and assimilation, nor do they
address whether the changes in the Muslim community after 9/11 were due to the
attacks themselves or were part of a pre-existing trend.

Our contribution is to examine this issue with standard econometric tools and with a
large, representative sample of Muslim immigrants obtained from the US Census and
American Community Surveys. In particular, we control for the aggregate trends in the
assimilation rate of Muslim immigrants and for a broad set of personal and state-level
characteristics, and exploit the geographic variation in hate crimes across states to
explain the changes in the state-level assimilation rate of Muslim immigrants.

Although this article is the first systematic analysis of the assimilation patterns of
Muslims in the aftermath of 9/11, several articles have examined whether terrorism
increased discrimination against Muslims in the labour market. Overall, the evidence is
mixed. Some studies find that terror attacks affected the earnings and employment of
Muslims living in the targeted country (Rabby and Rogers, 2009), while others report a
significant effect only on earnings (D�avila and Mora, 2005; Kaushal et al., 2007). A
third group of studies finds no effect on any labour market outcome for Muslims
(�Aslund and Rooth, 2005; Braakmann, 2010). Importantly, this literature focuses

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2067



exclusively on labour market outcomes in the immediate aftermath of terror attacks. In
contrast, we focus on assimilation outcomes measured up to nine years after the
attacks, and since these outcomes are difficult to reverse (marital status, fertility etc.),
they are much more likely to have long-term implications.

2. The Data

2.1. Data on the Assimilation Outcomes of Muslim Immigrants

Information about the assimilation of immigrants from Muslim countries over time is
obtained from the US Census in 1990 and 2000, as well as the American Community
Surveys (ACS) for 2007–10.2 We used the four years of the ACS in order to obtain a
large sample of Muslim immigrants after the 9/11 attacks in 2001.3 Therefore, our
main analysis is broken down into two time periods: the pre-attack period of 2000 and
the post-attack interval of 2007–10 (which is referred to as the 2010 period throughout
the text). In addition, the 1990 Census is used to examine pre-existing trends. Since
many of our measures of assimilation concern marriage and fertility decisions which
are typically made in early adulthood, our sample of interest is restricted to immigrants
between the ages of 20 and 40.

Although it is possible to determine whether someone is an immigrant in the Census
and ACS data using information on their country of birth, it is not straightforward to
identify who is a Muslim. The Census and the ACS contain no information on a
person’s religious affiliation (see Ruggles et al., 2008). Therefore, we use information
on the respondent’s self-reported ‘country of ancestry’ to infer whether the person is
likely to be a Muslim. Specifically, individuals reporting an ancestry from the
predominantly Muslim countries in the Middle East, Northern Africa and Asia were
classified as Muslims. A similar set of countries was used by Kaushal et al. (2007) to
examine the effect of the 9/11 attacks on the labour market outcomes of Arab and
Muslim residents.4 However, Kaushal et al. (2007) based their definition on the
respondent’s country of birth, not the respondent’s ancestry. For foreign-born
individuals, these two variables are almost always the same. However, when the two
pieces of information differ, we believe the ‘ancestry’ variable is likely to be more
accurate, since some individuals are born in foreign countries when their parents
lived abroad for a brief period of time. More importantly, our definition on ancestry
allows us to categorise native-born individuals as having a Muslim background as well.
This is important in order to estimate the size of the Muslim population in each state
and also to provide a more accurate measure of whether an immigrant who is
married to a native has in fact married someone of the same background or not.

2 The sample size of Muslims in the General Social Survey is too small to analyse (about 10 self-identified
Muslims per year).

3 The ACS is the largest representative survey that was conducted after 2000. Many existing studies
examine inequality and other labour issues over time using the Census for years up to and including 2000,
and the ACS for the post-2000 period. For example, see Beaudry et al. (2010, p. 1005) and Acemoglu and
Autor (2011, p. 1050).

4 The only exceptions are that we did not include Lebanon and Malaysia, since we believe most Lebanese
immigrants in the US are Christian and it was not clear that Malaysia is mostly Muslim. However, our results
are robust to changing the definition to include or exclude these countries.

© 2014 Royal Economic Society.

2068 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



If we used country of birth to determine whether a person is likely to be Muslim, any
immigrant who is married to a native would be considered to have married outside of
his/her group. Since this is one of our main measures of assimilation, we based our
definition of being a Muslim on the country of ancestry.

Appendix Table A1 lists the 20 countries of ancestry used in our definition of being
a Muslim. The most common countries are Pakistan and Iran, followed by Turkey,
Egypt and Bangladesh. As discussed below, there is considerable variation across the
different countries of origin in terms of their assimilation outcomes and the extent to
which each group experienced a backlash in the wake of the 9/11 attacks.

Using the Census and ACS data, we constructed several standard measures of
assimilation for each Muslim immigrant in the sample. Our first measure is whether
the person is ‘intra-married’ in the sense of being married to someone whose ancestry
is coded as one of the Muslim countries described above.5 This measure is widely used
to capture the strength of one’s ethnic identity and level of assimilation (Bisin and
Verdier, 2000; Bisin et al., 2004, 2008). There are two alternatives to being ‘intra-
married’: being single or being ‘inter-married’ in terms of marrying someone whose
ancestry is not one of the Muslim countries described above. Since someone who is
single may eventually be ‘intra-married’, as opposed to someone who is currently ‘inter-
married’ (unless they divorce and remarry), we will also examine the rate of ‘inter-
marriage’ as a relevant assimilation outcome.

Summary statistics for these marriage outcomes are presented in Table 1. For
example, 43.4% of the sample is intra-married in the 2000 Census, while 13.5% are
inter-married. Among the 69% that have ever been married, 63% are married within
the Muslim community according to our measure, while 37% are married outside of
the community. So, although intra-marriage is the most common type of family
formation within this community, there does appear to be a significant level of
integration into the rest of society as measured by the rate of inter-marriage.6

However, as seen in Appendix Table A1, the rate of intra-marriage varies consid-
erably across the different countries of origin. For example, individuals from Pakistan
and Bangladesh have very strong ethnic identities (their rate of intra-marriage is 53%
and 57% respectively), while individuals with a background from Egypt, Iran and
Turkey are much more assimilated (the rate of intra-marriage is 46%, 36% and 38%
respectively).

Our other measures of assimilation include the respondent’s fertility, labour force
participation and proficiency in English. Fertility is measured by the respondent’s
number of children, or alternatively, by the number of children below the age of 5. A
higher level of fertility is consistent with a stronger ethnic identity. This can be seen in
Appendix Table A1 which shows that immigrants from Pakistan and Bangladesh,
which have a high rate of intra-marriage compared to other countries (e.g. Egypt, Iran
and Turkey) also tend to have much higher rates of fertility. (Egypt is the exception,
with a comparable level of fertility to Bangladesh.)

5 The spouse is considered a Muslim if he/she meets the ancestry criterion, regardless of whether the
spouse was born in the US or not and regardless of whether the spouse meets the age requirements to be
included in our empirical analysis (between 20 and 40 years of age).

6 Table 1 reveals little about the aggregate trends in assimilation outcomes, due to changes over time in
the size and demographic composition of the population of Muslim immigrants.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2069



Female labour force participation is also a useful measure of assimilation, since
families with a higher sense of ethnic identity are more likely to adhere to traditional
family roles whereby women are less likely to participate in the labour market. Again,
evidence for this notion is seen in Appendix Table A1, which shows that women from
countries with a higher intra-marriage rate (Bangladesh and Pakistan) have much
higher rates of being in the ‘home sector’ (47% and 52% respectively) than women
from countries that have much lower intra-marriage rates (31% from Turkey and 24%
from Iran). Being in the ‘home sector’ is defined as not being in the labour force and
not being enrolled at school. It is worth noting that there are no differences in the rate
of being in the ‘home sector’ for men across these countries of origin, which is

Table 1

Descriptive Statistics of Immigrants from Muslim Countries and Number of Hate Crimes
Against Muslims

1990 2000 2007–10

Mean SD Mean SD Mean SD

Assimilation outcomes
Intra-married 0.370 0.48 0.434 0.50 0.451 0.50
Inter-married 0.186 0.39 0.135 0.34 0.131 0.34
Ever married 0.681 0.47 0.690 0.46 0.688 0.46
Number of children 0.896 1.25 0.992 1.31 1.002 1.30
Number of children less than age 5 0.400 0.68 0.425 0.71 0.442 0.71
Home sector (not in labour force or school) 0.171 0.38 0.237 0.43 0.220 0.41
Speak English at home 0.098 0.30 0.075 0.26 0.098 0.30
Speak English well 0.672 0.47 0.649 0.48 0.647 0.48

Demographic characteristics
Male 0.626 0.48 0.570 0.50 0.514 0.50
Years in the US 8.849 5.94 10.790 7.81 11.280 8.30
Age 30.550 5.64 31.145 5.84 31.106 5.85
High school dropout 0.071 0.26 0.074 0.26 0.063 0.24
High school graduate 0.194 0.40 0.217 0.41 0.191 0.39
Some college 0.284 0.45 0.258 0.44 0.250 0.43
College graduate 0.451 0.50 0.451 0.50 0.495 0.50
Home sector rate for women within state 0.218 0.02 0.222 0.02 0.192 0.02
Male unemployment rate within state 0.058 0.01 0.055 0.01 0.087 0.03
Male mean log wage within state 10.264 0.11 10.567 0.11 10.823 0.12
Percentage of males within state (Muslims) 0.604 0.03 0.560 0.03 0.512 0.05
Percentage Muslim within state 0.005 0.00 0.008 0.00 0.009 0.00

Hate crimes against Muslims
Muslim hate crimes (since last period) 14.549 14.58 78.966 62.49
Muslim hate crimes per capita 0.004 0.00 0.028 0.03
Muslim hate crimes per capita (excl. 2002–8) 0.014 0.01
Hate crimes against blacks per capita 0.052 0.06 0.037 0.05
Log hate crimes against homosexuals 6.084 1.42 5.884 1.25
Log hate crimes against jews 5.997 1.76 5.519 1.69

Number of observations 13,467 20,139 17,558

Notes. Hate crime data come from the FBI Uniform Crime Reporting (UCR) Program. All other variables are
computed from the United States Census (1990 and 2000) and the American Community Surveys (2007–10).
The sample is restricted to immigrants from Muslim countries, as detailed in Appendix Table A1.

© 2014 Royal Economic Society.

2070 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



consistent with the idea that the differences across groups represent different rates of
assimilation choices rather than opportunities.

Similar patterns are also seen with our remaining measures of assimilation which
capture the respondent’s proficiency in English. Our first measure is whether the
respondent speaks English at home (only 7.5% in 2000), and the second measure is
whether the respondent speaks English well (65% in 2000). A respondent was
classified as speaking well if he/she was coded as speaking ‘only English’ or
speaking ‘very well’. Again, we see a consistent pattern whereby immigrants from a
more assimilated background (Iran and Turkey) have lower intra-marriage rates,
higher female labour force participation, lower fertility and a higher chance of
speaking English at home or speaking very well (11% and 76% respectively for Iran,
11% and 63% for Turkey). In contrast, immigrants with less assimilated outcomes
(Bangladesh and Pakistan) tend to have higher intra-marriage rates and fertility,
with lower rates of female labour participation, speaking English at home, and
speaking English well (4.5% and 47% for Bangladesh on the last two outcomes,
and 6.6% and 69% for Pakistan). The consistency of these independently-measured
and defined assimilation outcomes across groups supports the idea that they
individually and jointly capture an accurate picture of the assimilation rate of each
group, and therefore, justifies their use as meaningful outcomes in our empirical
analysis.

The US Census and ACS surveys also contain a rich set of demographic information
such as: gender, age, education level and, notably, the state of residence for each
respondent. This geographic information is particularly important for our identifica-
tion strategy since we do not want to rely on aggregate time trends to identify the causal
effect of anti-Muslim hate crimes on assimilation behaviour. Rather, we control for
aggregate time trends and exploit the geographic variation in hate crimes across states
to explain the changes in the assimilation outcomes across locations. To do this, we
need variation in the location of Muslims across the US, which is shown in Appendix
Table A2.

2.2. Data on Hate Crimes Against Muslims

Our goal is to estimate the effect of the anti-Muslim backlash on the assimilation
outcomes of Muslim immigrants. However, it is not straightforward to measure the
extent of the backlash in an accurate and consistent way across states. To do this, we
use information on the reported number of ‘hate crimes’ against Muslims as a proxy
for the extent of the backlash felt across states by immigrants from Muslim countries.
Since 1992, the FBI Uniform Crime Reporting Program (UCR) has been collecting
data on crimes motivated by racial, religious, ethnicity/national-origin, sexual-
orientation and disability bias. A hate crime is not defined as a separate type of crime.
Rather, hate crimes are typical types of crime like assault or burglary but are defined as
a hate crime if the offender’s motive is based on a bias against the victim’s gender, race,
etc. That is, a crime can be classified as a hate crime only if the police obtain additional
information about the motives of the perpetrator and, using their judgment,
determine that a biased motivation exists.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2071



Hate crimes are reported to the FBI for eleven categories of offences, but the most
common type by far is ‘intimidation’. In Appendix Table A3, we present the number
of hate crimes against Muslims in 2001 for all eleven categories. The largest categories
are ‘intimidation’ with 53.4% of the total, ‘destruction/vandalism’ with 24.2%, the two
types of assault (simple and aggravated) with 16.2% and arson with 3.3%. According
to the UCR Hate Crime Report, 2001, ‘hate crimes touch not only the individual
victim, but they also affect the entire group associated with the particular bias
motivation’. For this reason, the reported number of hate crimes against Muslims is
likely to be a good proxy for how Muslims are accepted by the wider public in their
locality. To the extent that this measure provides an inaccurate portrayal of the
relationship between the Muslim community and the rest of society, our results should
be biased towards zero.

Since more populous states will tend to have larger numbers of hate crimes, we
divide the number of hate crimes against Muslims in each state by the number of
Muslims (individuals with an ancestry from a Muslim country as defined above)
found in the 2000 Census in each state. We normalise by the state-level Muslim
population from the 2000 Census, and not the earlier or later years, for a few
reasons. The main drawback of normalising by the 1990 Census is that this census
does not include the large wave of immigration during the 1990s (Table 1 shows
that share of the state population represented by Muslim immigrants almost
doubled from 1990 to 2000). There are two advantages of using the 2000 Census
over the ACS Surveys from 2007–10. First, the 2000 Census is a much larger sample
than the ACS samples for 2007–10. More importantly, it is possible that the
distribution of Muslims across states in 2007–10 is endogenous to the extent of the
local backlash in the wake of the 9/11 attacks, so that estimates of the Muslim
population for each state after 9/11 should not be considered exogenous. For these
reasons, we normalise the number of hate crimes against Muslims by the 2000
state-level estimate of the Muslim population size, which we refer to as ‘hate crimes
per capita’.

Appendix Table A2 presents the number of hate crimes against Muslims per capita
for each state in the post-9/11 period. The Table displays considerable variation in
hate crimes against Muslims across states. For example, hate crimes per capita were
much higher in Michigan, Ohio, Massachusetts, and New Jersey relative to Virginia,
Maryland, Texas, Pennsylvania, California, Illinois, Florida and New York. Therefore, a
high intensity of hate crimes against Muslims was not limited to the largest states or to
any particular region of the country.

For this study, it is important to note that there was a dramatic increase in hate
crimes against Muslims in the wake of the 9/11 attacks. With only four months
remaining in the calendar year after the incident, the reported total number of hate
crimes against Muslims increased 1,600% from 2000 to 2001. This sudden jump is
displayed in Figure 1, which also shows that the number of hate crimes against Muslims
decreased after the surge following the 9/11 attacks but settled down to a yearly mean
of 139.5 incidents after 2001 compared to 23.3 prior to 2001. It is worth noting,
however, that even the peak year of anti-Muslim hate crimes does not approach the
levels of attacks against other groups (Blacks, Jews, Gays and Lesbians), as seen in

© 2014 Royal Economic Society.

2072 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



Figure 2. Even on a per capita basis, Table 1 shows that hate crime against blacks after
2001 is considerably more prevalent than hate crime against Muslims.7

Although the number of hate crimes against Muslims never approached levels
experienced by other groups, it is clear that the 9/11 attacks caused an unexpected
surge in anti-Muslim sentiment, which led to a permanent shift in sentiment thereafter.
The goal of this article is to examine whether the change in sentiment towards Muslims
strengthened their ethnic identify. To do this, our strategy is to examine whether

0

100

200

300

400

500

N
um

b
er

 o
f 

H
at

e 
C

ri
m

es

1990 1995 2000 2005 2010

year

Fig. 1. Hate Crimes against Muslims in the US
Source. FBI Uniform Crime Reporting (UCR) Program.

0

1,000

2,000

3,000

4,000

N
um

b
er

 o
f 

H
at

e 
C

ri
m

es

1990 1995 2000 2005 2010
year

Anti-Muslim Anti-Black
Anti-Jew Anti-Gay

Fig. 2. Hate Crimes in the US
Source. FBI Uniform Crime Reporting (UCR) Program.

7 The Census data do not provide information on the size of the Jewish or homosexual community.
Hence, we cannot normalise hate crimes for those groups by their population sizes.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2073



changes in the assimilation patterns of Muslims across states since 2001 are associated
with the size of the local backlash. Figure 3 presents a first look at whether changes in
the state-level intra-marriage rate between 2000–10 are correlated with the state-level
changes in Muslim hate crimes per capita. For both men and women, there is a
statistically significant positive relationship, suggesting that the assimilation rate of
Muslim immigrants was indeed slower in places which experienced a more intensive
backlash. This preliminary analysis is explored extensively throughout the remainder
of the article but Figure 4 conducts a simple placebo analysis by showing that the post-
9/11 backlash was not correlated with higher rates of intra-marriage for Muslim
immigrants over the age of 40 – most of whom had already made their marriage
decision before the 9/11 attacks. It is also important to note that the size of the
backlash was not higher in places that were already experiencing slower rates of
assimilation. Figure 5 shows that the change in the state-level intra-marriage rate
between 1990 and 2000 was not significantly correlated with the backlash felt by
Muslim immigrants in the wake of the attacks in 2001. For women, the relationship
even appears to be negative. In the next Section, we describe our empirical strategy to
establish these results more extensively, while providing support for a causal
interpretation of the results.

3. Empirical Strategy

Our empirical strategy is designed to identify the causal effect of hate crimes against
Muslims on the assimilation outcomes of Muslim immigrants. Our unit of
observation is the individual, and we model his/her outcome as a function of his/

–0.2

0

0.2

0.4

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males
2

00
0–

1
0 

C
h

an
ge

 i
n 

In
tr

a-
m

ar
ri

ag
e

2000–10 Change in Muslim Hate Crimes Per Capita

Means by State for Sample under Age 41

Fig. 3. State Analysis of Intra-marriage, 2000–10
Notes. Female slope = 1.177 (t-stat = 3.56); Male slope = 0.764 (t-stat = 2.31). Regressions are
weighted by state sample size, represented by the size of each circle.

© 2014 Royal Economic Society.

2074 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



her personal characteristics, state of residence, survey year and the level of hate
crimes per capita in the individual’s state of residence. Specifically, we estimate the
following model:

–0.2

0

0.2

0.4

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males

1
99

0–
2

00
0 

C
h

an
ge

 i
n 

In
tr

a-
m

ar
ri

ag
e

2000–10 Change in Muslim Hate Crimes Per Capita

Examining Pre-existing Trends for Sample under Age 41

Fig. 5. State Analysis of Intra-marriage, 1990–2000
Notes. Female slope = �0.797 (t-stat = �1.60); Male slope = 0.345 (t-stat = 1.05). Regressions are
weighted by state sample size, represented by the size of each circle.

–0.2

0

0.2

0.4

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males
2

00
0–

1
0 

C
h

an
ge

 i
n 

In
tr

a-
m

ar
ri

ag
e

2000–10 Change in Muslim Hate Crimes Per Capita

Means by State

Fig. 4. Intra-marriage for Those Older than 40, 2000–10
Notes. Female slope = 0.408 (t-stat = 1.04); Male slope = �0.0022 (t-stat = �0.01). Regressions
are weighted by state sample size, represented by the size of each circle.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2075



outcomeist ¼ a � Muslim Hate Crimes PCst þ b � xist þ ct þ ls þ eist (1)
where outcomeist represents the assimilation outcome of individual i who lives in state s
in year t; Muslim Hate Crimes PCst is the number of hate crimes against Muslims per capita
in state s in year t; ct is a fixed-effect for each survey year t; ls is a fixed-effect unique to
state s; and xist is a vector of individual and state characteristics which include the
individual’s gender, age, years of schooling and years living in the US. In some
specifications, the xist vector will also include characteristics which vary at the state-year
level (computed from the same Census and ACS Surveys) such as the female ‘home
sector’ rate (not enrolled in school or in the labour force) for native women in state s,
the unemployment rate of native men in the state and the mean log wage for native
men in the state. Unobserved determinants of the individual’s assimilation outcome
are captured by the error term, eist.

As indicated earlier, Muslim Hate Crimes PCst is the total number of hate crimes in
state s at year t, divided by the total number of individuals reporting an ancestry from a
Muslim country in state s in the year 2000. When t equals the year 2000, the total
number of hate crimes in that state refers to the sum of the hate crimes in that state
from 1992 (the first year that the FBI started collecting hate crime data) to the year
2000. When t refers to a year after 2001 (2007–10), the total number of hate crimes in
that state refers to the sum of hate crimes against Muslims between 2001 and 2008. We
use the sum of hate crimes across years in order to obtain a more precise measure of
the level of anti-Muslim sentiment, since the number of hate crimes is quite low,
particularly in the pre-2001 period. In this manner, our measure of hate crimes
per capita is designed to capture the extent to which Muslims felt an increase in hate
crimes after 2001 within each state.

The goal of the proposed econometric specification is to identify a, which represents
the causal effect of anti-Muslim hate crimes on an individual’s assimilation outcome.
There are several reasons why hate crimes against Muslims may affect their rate of
assimilation. It is likely that hate crimes, which may be accompanied by increased
discrimination, harassment, racial profiling and hostile rhetoric against Muslims,
create a barrier between the Muslim community and the rest of society. By increasing
the isolation of the Muslim community, hate crimes raise the costs of assimilation and
perhaps increase the return to investing in social capital within their own community,
thus leading to lower rates of integration. As mentioned above, these forces have been
highlighted in the existing literature (Abdo, 2006; Barrett, 2008; Bakalian and
Bozorgmehr, 2009). However, hate crimes also increase the cost of being publicly
identified as a Muslim and, therefore, might lead some Muslims to shed their identity
and blend into society at large. Given these two conflicting predictions, the goal of our
analysis is to determine which hypothesis is supported by the data.

By including fixed-effects for each state and survey year, we are essentially examining
whether changes over time in hate crimes against Muslims within a state are correlated
with the changes over time in the assimilation rate of Muslims within that state, after
controlling for the national trend and a rich set of personal and state-level
characteristics. Our identifying assumption in (1), therefore, is that hate crimes
against Muslims at the state level are not correlated with omitted variables that affect
the local assimilation rate and that changes in Muslim hate crimes at the state level are

© 2014 Royal Economic Society.

2076 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



not caused by changes in the assimilation rate within the state (i.e. no reverse
causality).

We address these issues in several ways. First, as depicted in Figure 3, we show that
changes in the number of hate crimes per capita against Muslims are positively
correlated with concurrent changes in intra-marriage for younger Muslim immigrants
but not with the intra-marriage rates of older Muslim immigrants (Figure 4) who are
likely to be characterised by the same unobservable proclivities for assimilation, but
were too old to change their marriage behaviour after 9/11. In addition, we conduct
extensive robustness checks across assimilation outcomes and by including/excluding
a vast array of control variables which are highly significant determinants of
assimilation tendencies. Regarding reverse-causality, we show that hate crimes after
9/11 were not more prevalent in states that were already becoming less-assimilated
(Figure 5 and Tables 2 and 3) and that the results are similar using the immediate
backlash against Muslims, which could not be a response to assimilation trends over
the decade after 9/11, as the treatment variable. Finally, we examine whether there is a
systematic relationship between the local level of anti-Muslim hate crimes and
observable characteristics of the local Muslim population which are likely to affect the
local trends in assimilation.

Table 2 analyses which state-level characteristics as of 2000, just prior to the 9/11
attacks, are associated with a larger backlash after the attacks. The Table distinguishes
between two broad groups of state characteristics: those associated with the state’s non-
Muslim population (upper panel) and those characterising the local Muslim
population (bottom panel). Each coefficient comes from a simple regression of the
state level backlash after 9/11 (Muslim Hate Crimes PCst for t = 2010) on the state mean
in the year 2000 of the variable indicated above each estimate. The Table shows that
the extent of the backlash was correlated with certain characteristics of the local non-
Muslim population. In particular, hate crimes against Muslims were larger in states that
tended to be more white, native-born (less immigrants including those from Muslim
countries), and where hate crimes were already abundant against blacks and Muslims
before the 9/11 attacks.

The bottom panel of Table 2 shows very little connection between the backlash after
9/11 and the characteristics of the Muslim state population in 2000. The backlash is
not associated with most of the country-of-origin shares, and most importantly, is
completely uncorrelated with the assimilation and education characteristics of the
local Muslim community just prior to 9/11.

Overall, Table 2 demonstrates that the backlash was not entirely random – it was
indeed larger in areas that had a history of hate crimes against Muslims and other
groups (blacks) and tended to occur in areas where immigrants (not just Muslim
immigrants) were smaller in concentration. But, the extent of the backlash was not
correlated with the local assimilation rates or characteristics of Muslims themselves.
This lends support to our identifying assumption that the local backlash was an event
external to the behaviour and character of the local Muslim community and, therefore,
was not correlated with omitted factors affecting their assimilation rates after 9/11.

To provide further support for this assumption, Table 3 examines whether the
backlash after 9/11 was associated with the pre-existing trends (between 1990–2000) in
the characteristics of the Muslim and non-Muslim communities before the attacks.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2077



T
a
b
le

2

T
h
e
R
el
a
ti
on
sh
ip

be
tw
ee
n
A
n
ti
-M

u
sl
im

H
a
te
C
ri
m
es

by
S
ta
te
a
ft
er

9
/1
1
(2
0
0
1
–
8
)
a
n
d
S
ta
te
C
h
a
ra
ct
er
is
ti
cs

Ju
st
P
ri
or

to
9
/1
1

D
e
p
e
n
d
e
n
t
va
ri
a
b
le
:
a
n
ti
-M

u
sl
im

h
a
te

cr
im

e
s
p
er

ca
p
it
a
b
y
S
ta
te

fr
o
m

2
0
0
1
–
8

C
h
a
ra
ct
er
is
ti
cs

of
th
e
st
a
te
p
op
u
la
ti
on

in
ye
a
r
2
0
0
0

S
ta
te

le
ve
l
h
a
te

cr
im

e
s
in

ye
ar

2
0
0
0
a
g
a
in
st
:

M
u
sl
im

h
at
e

cr
im

e
s
P
C

B
la
ck
s
h
a
te

cr
im

e
s
P
C

G
ay

h
a
te

cr
im

e
s
P
C

Je
w
is
h
h
a
te

cr
im

e
s
P
C

4
.0
3
3
*
*
*

0
.1
9
8
*
*
*

�1
.4
8
9

�0
.9
9
1

(0
.7
6
8
)

(0
.0
5
9
)

(3
.6
2
6
)

(1
.5
7
2
)

S
ta
te

le
ve
l
d
e
m
o
g
ra
p
h
ic
s
a
n
d
e
co

n
o
m
ic

co
n
d
it
io
n
s
fo
r
2
0
–
4
0
ye
a
r
o
ld
s
in

ye
a
r
2
0
0
0

%
N
at
iv
e

%
B
la
ck

%
W
h
it
e

%
M
u
sl
im

M
a
le

u
n
e
m
p
lo
ym

e
n
t

ra
te

M
a
le

m
e
a
n

w
a
g
e

%
m
a
le

H
S
d
ro
p
o
u
ts

%
m
a
le

co
ll
e
g
e
g
ra
d
s

0
.1
1
9
*
*
*

�0
.0
8
1

0
.1
4
3
*
*
*

�3
.2
0
0
*
*
*

�0
.7
7
4
*

�0
.0
3
7

�0
.2
6
0

�0
.1
3
4

(0
.0
3
4
)

(0
.0
6
8
)

(0
.0
3
9
)

(0
.9
0
8
)

(0
.3
9
0
)

(0
.0
3
9
)

(0
.1
8
4
)

(0
.1
2
7
)

C
h
a
ra
ct
er
is
ti
cs

of
M
u
sl
im

im
m
ig
ra
n
ts
by

st
a
te
in

ye
a
r
2
0
0
0

S
ta
te

le
ve
l
a
ss
im

il
a
ti
o
n
a
n
d
e
d
u
ca
ti
o
n
ch

a
ra
ct
e
ri
st
ic
s
o
f
M
u
sl
im

im
m
ig
ra
n
ts

in
ye
a
r
2
0
0
0

In
tr
a
-m

a
rr
ie
d

N
u
m
b
e
r
o
f

ch
il
d
re
n

N
u
m
b
e
r
o
f

ch
il
d
re
n
<
5
ye
ar
s

S
p
e
a
k
E
n
g
li
sh

a
t
h
o
m
e

S
p
e
a
k
E
n
g
li
sh

w
e
ll

%
H
S

d
ro
p
o
u
ts

%
co

ll
e
g
e

g
ra
d

�0
.0
9
0

0
.0
2
3

0
.0
3
6

0
.1
7
9

�0
.0
0
8

0
.0
7
8

0
.0
0
7

(0
.0
6
3
)

(0
.0
3
1
)

(0
.0
6
8

(0
.1
4
0
)

(0
.0
8
4
)

(0
.1
9
0
)

(0
.0
7
7
)

S
ta
te

le
ve
l
p
e
rc
e
n
ta
g
e
o
f
M
u
sl
im

im
m
ig
ra
n
ts

b
y
co

u
n
tr
y
o
f
o
ri
g
in

in
ye
ar

2
0
0
0

E
g
yp
t

M
o
rr
o
co

Ir
a
n

Ir
a
q

Jo
rd
a
n

S
yr
ia

T
u
rk
e
y

0
.0
4
2

0
.1
0
4

�0
.0
5
8

0
.5
2
4
*
*
*

0
.6
9
2
*

0
.1
4
1

0
.0
3
5

(0
.0
8
1
)

(0
.1
9
2
)

(0
.0
3
8
)

(0
.0
5
9
)

(0
.3
4
6
)

(0
.1
0
1
)

(0
.1
6
8
)

P
a
le
st
in
e

A
fg
h
a
n
is
ta
n

B
a
n
g
le
d
e
sh

P
a
k
is
ta
n

In
d
o
n
e
si
a

O
th
e
r
A
ra
b

0
.1
1
0

�0
.3
0
6
*

�0
.1
3
9

�0
.1
0
5

0
.0
6
4

0
.1
9
7

(0
.1
7
1
)

(0
.1
5
9
)

(0
.0
8
6
)

(0
.0
9
0
)

(0
.2
2
4
)

(0
.1
1
7
)

N
ot
es
.
R
o
b
u
st
st
a
n
d
ar
d
e
rr
o
rs

a
p
p
e
ar

in
p
a
re
n
th
e
se
s.
E
ac
h
co

e
ffi
ci
e
n
t
co

m
e
s
fr
o
m

a
se
p
a
ra
te

re
g
re
ss
io
n
w
h
e
re

th
e
n
u
m
b
e
r
o
f
a
n
ti
-M

u
sl
im

h
a
te

cr
im

e
s
p
er
ca
p
it
a
b
y

st
a
te

is
re
g
re
ss
e
d
o
n
th
e
st
a
te
-l
e
ve
l
va
ri
a
b
le

in
d
ic
a
te
d
in

th
e
T
a
b
le

a
b
o
ve

e
a
ch

co
e
ffi
e
n
t.
N
u
m
b
e
r
o
f
o
b
se
rv
a
ti
o
n
s
in

e
a
ch

re
g
re
ss
io
n
is
5
0
.
R
e
g
re
ss
io
n
s
a
re

w
e
ig
h
te
d
b
y

th
e
st
a
te

sa
m
p
le

si
ze

o
f
M
u
sl
im

im
m
ig
ra
n
ts

in
2
0
0
0
.
S
ig
n
ifi
ca
n
ce

le
ve
ls
a
re

in
d
ic
a
te
d
b
y
*
,
*
*
,
o
r
*
*
*
w
h
ic
h
re
p
re
se
n
t
1
0
%
,
5
%

a
n
d
1
%

le
ve
ls
,
re
sp
e
ct
iv
e
ly
.
H
at
e

cr
im

e
s
p
er

ca
p
it
a
fo
r
g
a
ys

a
n
d
Je
w
s
a
re

cr
e
a
te
d
b
y
d
iv
id
in
g
b
y
th
e
st
a
te

p
o
p
u
la
ti
o
n
si
ze

in
2
0
0
0
.

© 2014 Royal Economic Society.

2078 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



T
a
b
le

3

R
el
a
ti
on
sh
ip

be
tw
ee
n
th
e
C
h
a
n
ge

in
A
n
ti
-M

u
sl
im

H
a
te
C
ri
m
es

fr
om

2
0
0
0
–
1
0
a
n
d
th
e
C
h
a
n
ge

in
S
ta
te
C
h
a
ra
ct
er
is
ti
cs

fr
om

1
9
9
0
–
2
0
0
0

D
e
p
e
n
d
e
n
t
va
ri
a
b
le
:
ch

a
n
g
e
in

a
n
ti
-M

u
sl
im

h
at
e
cr
im

e
s
p
er

ca
p
it
a
b
y
st
a
te

fr
o
m

2
0
0
0
–
1
0

C
h
a
n
ge

in
ch
a
ra
ct
er
is
ti
cs

of
th
e
st
a
te
p
op
u
la
ti
on

1
9
9
0
–
2
0
0
0

S
ta
te

le
ve
l
d
e
m
o
g
ra
p
h
ic
s
a
n
d
e
co

n
o
m
ic

co
n
d
it
io
n
s
fo
r
2
0
–
4
0
ye
a
r
o
ld
s

%
N
a
ti
ve

%
B
la
ck

%
W
h
it
e

%
M
u
sl
im

M
a
le

u
n
e
m
p
lo
ym

e
n
t

ra
te

M
a
le

m
e
an

w
a
g
e

%
m
a
le

H
S

d
ro
p
o
u
ts

%
m
a
le

co
ll
e
g
e
g
ra
d
s

0
.4
3
6
*

0
.1
3
3

0
.3
3
9
*
*

�7
.1
1
1
*
*

�1
.3
8
2
*
*
*

0
.4
2
8
*

�0
.4
3
2

0
.6
2
3

(0
.2
4
5
)

(0
.4
6
2
)

(0
.1
4
5
)

(3
.1
6
0
)

(0
.4
7
8
)

(0
.2
5
4
)

(0
.3
6
7
)

(0
.4
7
9
)

C
h
a
n
ge

in
ch
a
ra
ct
er
is
ti
cs

of
M
u
sl
im

im
m
ig
ra
n
ts
by

st
a
te
1
9
9
0
–
2
0
0
0

S
ta
te

le
ve
l
a
ss
im

il
a
ti
o
n
a
n
d
e
d
u
ca
ti
o
n
ch

a
ra
ct
e
ri
st
ic
s
o
f
M
u
sl
im

im
m
ig
ra
n
ts

In
tr
a
-m

a
rr
ie
d

N
u
m
b
e
r
o
f
ch

il
d
re
n

N
u
m
b
e
r
o
f

ch
il
d
re
n
<
5
ye
a
rs

S
p
e
a
k
E
n
g
li
sh

a
t

h
o
m
e

S
p
e
a
k
E
n
g
li
sh

w
e
ll

%
H
S

d
ro
p
o
u
t

%
co

ll
e
g
e

g
ra
d

�0
.0
0
7

0
.0
0
1

0
.0
1
3

�0
.0
9
5

�0
.0
5
6

0
.0
2
9

�0
.0
8
9

(0
.0
7
0
)

(0
.0
3
0
)

(0
.0
3
4
)

(0
.1
4
7
)

(0
.0
5
4
)

(0
.2
6
6
)

(0
.0
8
1
)

S
ta
te

le
ve
l
p
e
rc
e
n
ta
g
e
o
f
M
u
sl
im

im
m
ig
ra
n
ts

b
y
co

u
n
tr
y
o
f
o
ri
g
in

E
g
yp
t

M
o
rr
o
co

Ir
a
n

Ir
a
q

Jo
rd
a
n

S
yr
ia

T
u
rk
e
y

0
.2
2
8
*

0
.1
7
1

0
.1
0
5

0
.4
8
1

0
.5
4
9

�0
.1
4
1

0
.1
9
2

(0
.1
3
4
)

(0
.2
1
5
)

(0
.1
6
9
)

(0
.3
6
3
)

(0
.3
5
9
)

(0
.1
3
5
)

(0
.2
3
4
)

P
a
le
st
in
e

A
fg
h
a
n
is
ta
n

B
a
n
g
le
d
e
sh

P
a
k
is
ta
n

In
d
o
n
e
si
a

O
th
e
r
A
ra
b

�0
.1
1
7

�0
.2
0
9

�0
.1
5
0

�0
.0
6
9

0
.2
8
2

�0
.2
2
2

(0
.2
6
6
)

(0
.2
9
1
)

(0
.1
7
9
)

(0
.0
8
9
)

(0
.2
6
1
)

(0
.2
0
3
)

N
ot
es
.
R
o
b
u
st
st
a
n
d
ar
d
e
rr
o
rs

a
p
p
e
ar

in
p
a
re
n
th
e
se
s.
E
a
ch

co
e
ffi
ci
e
n
t
co

m
e
s
fr
o
m

a
se
p
a
ra
te

re
g
re
ss
io
n
w
h
e
re

th
e
ch

a
n
g
e
in

th
e
n
u
m
b
e
r
o
f
a
n
ti
-M

u
sl
im

h
a
te

cr
im

e
s

p
er
ca
p
it
a
b
y
st
a
te

fr
o
m

2
0
0
0
–
1
0
(a
s
d
e
fi
n
e
d
in

th
e
te
x
t,
w
h
e
re

2
0
0
0
re
fe
rs

to
th
e
p
e
ri
o
d
1
9
9
2
–
2
0
0
0
a
n
d
th
e
2
0
1
0
p
e
ri
o
d
re
fe
rs

to
2
0
0
1
–
8
)
is
re
g
re
ss
e
d
o
n
th
e
1
9
9
0
–

2
0
0
0
ch

a
n
g
e
in

st
a
te
-l
e
ve
l
va
ri
a
b
le

in
d
ic
a
te
d
in

th
e
T
a
b
le

a
b
o
ve

e
a
ch

co
e
ffi
e
n
t.
N
u
m
b
e
r
o
f
o
b
se
rv
a
ti
o
n
s
in

e
a
ch

re
g
re
ss
io
n
is
9
9
,
a
s
th
e
re
g
re
ss
io
n
w
a
s
p
e
rf
o
rm

e
d
w
it
h

a
p
a
n
e
l
d
a
ta

se
t
o
f
st
a
te
s
w
it
h
st
a
te

fi
x
e
d
-e
ff
e
ct
s
a
n
d
ye
a
r
e
ff
e
ct
s.
R
e
g
re
ss
io
n
s
a
re

w
e
ig
h
te
d
b
y
th
e
st
a
te

sa
m
p
le

si
ze

o
f
M
u
sl
im

im
m
ig
ra
n
ts
in

2
0
0
0
.
S
ig
n
ifi
ca
n
ce

le
ve
ls

a
re

in
d
ic
a
te
d
b
y
*
,
*
*
,
o
r
*
*
*
w
h
ic
h
re
p
re
se
n
t
1
0
%
,
5
%
,
a
n
d
1
%

le
ve
ls
,
re
sp
e
ct
iv
e
ly
.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2079



Table 3 does not include measures for the pre-existing trends in hate crimes against
various groups, since the FBI did not collect these data prior to 1992. Table 3 shows
that the change in Muslim hate crimes per capita between 2000–10 was positively
associated with certain pre-existing changes in the non-Muslim community between
1990–2000 (states that were becoming relatively more native-born, more white, less
Muslim and better economic conditions). However, the backlash after the attacks is
completely uncorrelated with the pre-existing assimilation trends of the local Muslim
community. This result provides additional support for the assumption that the
backlash was not associated with contemporaneous factors which affected Muslim
assimilation after 9/11.

This assumption is explored further in Table 4 which examines whether the change
in local hate crimes against Muslims after 9/11 is correlated with other factors which
may be concurrently affecting the assimilation rate of Muslims in the post-9/11 period.
This Table does not include the assimilation outcomes of Muslims, since these are the
main outcomes of interest which are examined in the remaining sections of the article.
Once again, Table 4 shows that the extent of the local backlash after 9/11 was not
correlated with the trend in hate crimes against other groups, the demographic and
schooling characteristics of the state population, and the education and country-of-
origin shares of the local Muslim population.

Although Table 4 shows that the local backlash was not correlated with a larger share
of Muslim immigrants from a particular country, Table 5 examines whether there is an
underlying, systematic pattern whereby immigrants with a particular tendency to
assimilate became more concentrated in areas with a larger local backlash.8 As
discussed previously, Muslim immigrants from certain countries display much stronger
tendencies to be assimilated versus others (Turkey and Iran versus Bangladesh and
Pakistan). In order to see if hate crimes after 9/11 are associated with higher
concentrations of immigrations from the group of countries that tend to be less
assimilated, we assign to each Muslim immigrant in the 2000 and 2010 samples the
mean of the assimilation outcome for that person’s country of ancestry in the year
2000, and regress that variable on the state level of hate crimes against Muslims in both
periods. This analysis appears in Table 5.

The upper two panels of Table 5 use data from the 2000 and 2010 periods; each
coefficient comes from a regression of the mean of the assimilation outcome for the
individual’s country of ancestry in the year 2000 on a state fixed-effect, year fixed-
effects and Muslim Hate Crimes PCst, as defined above. The bottom two panels use data
only on Muslim immigrants that arrived after 2001, so state-fixed effects cannot be
included since only the 2010 period is used.

The 10 coefficients in the upper panel of Table 5 do not show any systematic
evidence that immigrants from countries with lower assimilation rates tended to
concentrate in areas with more hate crimes. Only one of the ten coefficients is
significant at the 10% level (which is likely to occur even if the true parameter is zero in
all ten regressions). The sign of this coefficient indicates that hate crimes are

8 The existing literature has emphasised that immigrant assimilation is heavily determined by the country
of origin and parental identity (Dustmann, 1996; Casey and Dustmann, 2010).

© 2014 Royal Economic Society.

2080 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



T
a
b
le

4

R
el
a
ti
on
sh
ip

be
tw
ee
n
th
e
C
h
a
n
ge

in
A
n
ti
-M

u
sl
im

H
a
te
C
ri
m
es

fr
om

2
0
0
0
–
1
0
a
n
d
th
e
C
h
a
n
ge

in
S
ta
te
C
h
a
ra
ct
er
is
ti
cs

fr
om

2
0
0
0
–
1
0

D
e
p
e
n
d
e
n
t
va
ri
a
b
le
:
th
e
ch

a
n
g
e
in

a
n
ti
-M

u
sl
im

h
a
te

cr
im

e
s
p
er

ca
p
it
a
b
y
st
a
te

fr
o
m

2
0
0
0
–
1
0

C
h
a
n
ge

in
ch
a
ra
ct
er
is
ti
cs

of
th
e
st
a
te
p
op
u
la
ti
on

2
0
0
0
–
1
0

S
ta
te

le
ve
l
h
a
te

cr
im

e
s
a
g
a
in
st
:

B
la
ck
s
h
a
te

cr
im

e
s
P
C

G
a
y
h
a
te

cr
im

e
s

P
C

Je
w
is
h
h
a
te

cr
im

e
s

P
C

�0
.0
7
8

4
.2
3
9

3
.9
0
9

(0
.1
5
0
)

(5
.4
2
6
)

(3
.5
6
5
)

S
ta
te

le
ve
l
d
e
m
o
g
ra
p
h
ic
s
a
n
d
e
co

n
o
m
ic

co
n
d
it
io
n
s
fo
r
2
0
–
4
0
ye
ar

o
ld
s

%
N
a
ti
ve

%
B
la
ck

%
W
h
it
e

%
M
u
sl
im

M
a
le

u
n
e
m
p
lo
ym

e
n
t

ra
te

M
a
le

m
e
an

w
a
g
e

%
m
a
le

H
S

d
ro
p
o
u
ts

%
m
a
le

co
ll
e
g
e
g
ra
d
s

�0
.3
2
8
*
*

0
.1
5
0

�0
.2
0
1

�8
.2
5
6

0
.4
2
9

�0
.2
0
5

0
.1
8
4

�0
.0
7
8

(0
.1
5
1
)

(0
.2
9
5
)

(0
.1
8
5
)

(9
.2
8
1
)

(0
.5
7
6
)

(0
.1
8
1
)

(0
.7
9
8
)

(0
.4
1
8
)

C
h
a
n
ge

in
ch
a
ra
ct
er
is
ti
cs

of
M
u
sl
im

im
m
ig
ra
n
ts
by

st
a
te
2
0
0
0
–
1
0

S
ta
te

le
ve
l
e
d
u
ca
ti
o
n
ch

a
ra
ct
e
ri
st
ic
s
o
f
M
u
sl
im

im
m
ig
ra
n
ts

P
e
rc
e
n
t
H
S
d
ro
p
o
u
t

P
e
rc
e
n
t
co

ll
e
g
e

g
ra
d

0
.1
6
2

�0
.0
8
0

(0
.1
8
4
)

(0
.0
9
3
)

S
ta
te

le
ve
l
p
e
rc
e
n
ta
g
e
o
f
M
u
sl
im

im
m
ig
ra
n
ts

b
y
co

u
n
tr
y
o
f
o
ri
g
in

E
g
yp
t

M
o
rr
o
co

Ir
a
n

Ir
a
q

Jo
rd
a
n

S
yr
ia

T
u
rk
e
y

�0
.0
8
8

�0
.1
6
7

�0
.1
1
2

�0
.0
0
8

�0
.5
4
5

�0
.1
7
5

0
.0
5
8

(0
.1
5
2
)

(0
.5
6
1
)

(0
.1
1
4
)

(0
.3
9
1
)

(0
.4
1
5
)

(0
.2
5
6
)

(0
.1
6
7
)

P
a
le
st
in
e

A
fg
h
a
n
is
ta
n

B
a
n
g
le
d
e
sh

P
a
k
is
ta
n

In
d
o
n
e
si
a

O
th
e
r
A
ra
b

0
.0
6
0

0
.1
3
7

�0
.2
4
4

0
.0
4
6

�0
.1
7
4

0
.2
6
5
*

(0
.3
6
0
)

(0
.5
0
4
)

(0
.2
5
5
)

(0
.1
1
8
)

(0
.2
8
1
)

(0
.1
4
8
)

N
ot
es
.
R
o
b
u
st
st
a
n
d
a
rd

e
rr
o
rs

a
p
p
e
ar

in
p
a
re
n
th
e
se
s.
E
a
ch

co
e
ffi
ci
e
n
t
co

m
e
s
fr
o
m

a
se
p
a
ra
te

re
g
re
ss
io
n
w
h
e
re

th
e
ch

a
n
g
e
in

th
e
n
u
m
b
e
r
o
f
a
n
ti
-M

u
sl
im

h
at
e
cr
im

e
s

p
er
ca
p
it
a
b
y
st
a
te

fr
o
m

2
0
0
0
–
1
0
(a
s
d
e
fi
n
e
d
in

th
e
te
x
t,
w
h
e
re

2
0
0
0
re
fe
rs

to
th
e
p
e
ri
o
d
1
9
9
2
–
2
0
0
0
a
n
d
th
e
2
0
1
0
p
e
ri
o
d
re
fe
rs

to
2
0
0
1
–
8
)
is
re
g
re
ss
e
d
o
n
th
e
2
0
0
0
–
1
0

ch
a
n
g
e
in

st
a
te
-l
e
ve
l
va
ri
a
b
le

in
d
ic
a
te
d
in

th
e
T
a
b
le

a
b
o
ve

e
a
ch

co
e
ffi
e
n
t.
N
u
m
b
e
r
o
f
o
b
se
rv
a
ti
o
n
s
in

e
a
ch

re
g
re
ss
io
n
is
1
0
0
,
a
s
th
e
re
g
re
ss
io
n
w
a
s
p
e
rf
o
rm

e
d
w
it
h
a

p
a
n
e
l
d
a
ta

se
t
o
f
st
a
te
s
w
it
h
st
a
te

fi
x
e
d
-e
ff
e
ct
s
a
n
d
ye
a
r
e
ff
e
ct
s.
R
e
g
re
ss
io
n
s
a
re

w
e
ig
h
te
d
b
y
th
e
st
a
te

sa
m
p
le

si
ze

o
f
M
u
sl
im

im
m
ig
ra
n
ts
in

2
0
0
0
.
S
ig
n
ifi
ca
n
ce

le
ve
ls
a
re

in
d
ic
a
te
d
b
y
*
,
*
*
,
o
r
*
*
*
w
h
ic
h

re
p
re
se
n
t
1
0
%
,
5
%
,
a
n
d
1
%

le
ve
ls
,
re
sp
e
ct
iv
e
ly
.
H
at
e
cr
im

e
s
p
er

ca
p
it
a
fo
r
g
a
ys

a
n
d
Je
w
s
a
re

cr
e
a
te
d
b
y
d
iv
id
in
g
b
y
th
e
st
a
te

p
o
p
u
la
ti
o
n
si
ze

in
2
0
0
0
.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2081



T
a
b
le

5

R
el
a
ti
on
sh
ip

be
tw
ee
n
M
u
sl
im

H
a
te
C
ri
m
es

a
n
d
P
er
so
n
a
l
C
h
a
ra
ct
er
is
ti
cs

of
Im

m
ig
ra
n
t
M
u
sl
im
s

C
o
u
n
tr
y
o
f
o
ri
g
in

m
e
a
n
in

ye
a
r
2
0
0
0
a
ss
ig
n
e
d
to

e
a
ch

re
sp
o
n
d
e
n
t
in

e
a
ch

p
e
ri
o
d
(2
0
0
0
–
1
0
)

In
tr
a
-m

a
rr
ie
d

N
u
m
b
e
r
o
f
ch

il
d
re
n

N
u
m
b
e
r
o
f
ch

il
d
re
n
le
ss

th
a
n
a
g
e
5

E
n
g
li
sh

a
t
h
o
m
e

S
p
e
a
k
E
n
g
li
sh

w
e
ll

(1
)

(2
)

(3
)

(4
)

(5
)

2
0
0
0
–
1
0
a
n
a
ly
si
s
w
it
h
a
ll
im
m
ig
ra
n
ts
a
n
d
st
a
te
fi
xe
d
-e
ff
ec
ts

M
a
le
s

M
u
sl
im

h
at
e
cr
im

e
s
P
C

0
.0
5
1

0
.3
0
8

0
.0
9
6

�0
.0
1
9

�0
.0
5
0

(0
.0
5
7
)

(0
.1
9
2
)

(0
.0
8
9
)

(0
.0
3
1
)

(0
.0
7
7
)

S
a
m
p
le

si
ze

2
0
,4
2
2

2
0
,4
2
2

2
0
,4
2
2

2
0
,4
2
2

2
0
,4
2
2

F
e
m
al
e
s

M
u
sl
im

h
at
e
cr
im

e
s
P
C

�0
.1
2
5

0
.2
1
8

0
.0
6
9

0
.0
0
7

0
.1
3
6
*

(0
.1
0
1
)

(0
.3
2
8
)

(0
.0
7
0
)

(0
.0
3
4
)

(0
.0
6
8
)

S
a
m
p
le

si
ze

1
7
,1
0
8

1
7
,1
0
8

1
7
,1
0
8

1
7
,1
0
8

1
7
,1
0
8

2
0
0
7
–
1
0
a
n
a
ly
si
s
w
it
h
im
m
ig
ra
n
ts
a
rr
iv
in
g
a
ft
er

2
0
0
1

M
a
le
s

M
u
sl
im

h
at
e
cr
im

e
s
P
C

�0
.1
3
1

0
.5
3
8
*

0
.1
9
2

0
.0
0
7

�0
.0
5
3

(0
.1
3
0
)

(0
.3
0
8
)

(0
.1
1
8
)

(0
.0
7
0
)

(0
.2
0
2
)

S
a
m
p
le

si
ze

3
,0
1
9

3
,0
1
9

3
,0
1
9

3
,0
1
9

3
,0
1
9

F
e
m
al
e
s

M
u
sl
im

h
at
e
cr
im

e
s
P
C

�0
.2
4
1

0
.8
8
5

0
.2
7
3
*

0
.0
1
1

0
.1
8
8

(0
.2
0
9
)

(0
.8
9
7
)

(0
.1
5
9
)

(0
.0
5
4
)

(0
.1
9
5
)

S
a
m
p
le

si
ze

3
,0
1
4

3
,0
1
4

3
,0
1
4

3
,0
1
4

3
,0
1
4

N
ot
es
.
S
ta
n
d
ar
d
e
rr
o
rs
,
w
h
ic
h
a
re

cl
u
st
e
re
d
b
y
st
a
te
,
a
p
p
e
ar

in
p
a
re
n
th
e
se
s.
S
ig
n
ifi
ca
n
ce

le
ve
ls
a
re

in
d
ic
a
te
d
b
y
*
,
*
*
o
r
*
*
*
w
h
ic
h
re
p
re
se
n
t
1
0
%
,
5
%

a
n
d
1
%

le
ve
ls

re
sp
e
ct
iv
e
ly
.
In

th
e
u
p
p
e
r
p
a
n
e
l,
w
h
ic
h
u
se
s
d
a
ta

fr
o
m

2
0
0
0
–
1
0
,
th
e
O
L
S
re
g
re
ss
si
o
n
s
in
cl
u
d
e
a
s
co

n
tr
o
l
va
ri
a
b
le
s:
st
a
te

fi
x
e
d
-e
ff
e
ct
s,
ye
ar

fi
x
e
d
-e
ff
e
ct
s,
a
g
e
,
a
g
e
-

sq
u
a
re
d
,
a
n
d
a
g
e
-c
u
b
e
d
.
U
si
n
g
d
a
ta

fr
o
m

2
0
0
7
–
1
0
in

th
e
b
o
tt
o
m

p
a
n
e
l,
th
e
O
L
S
re
g
re
ss
si
o
n
s
in
cl
u
d
e
a
s
co

n
tr
o
l
va
ri
a
b
le
s:
ye
a
r
fi
x
e
d
-e
ff
e
ct
s,
a
g
e
,
a
g
e
-s
q
u
a
re
d
,
a
n
d

a
g
e
-c
u
b
e
d
.

© 2014 Royal Economic Society.

2082 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



associated with a higher rate of assimilation for Muslim women (more likely to speak
English well) – which is the opposite direction of our main findings.

The bottom panel performs a similar analysis focusing only on those immigrants that
arrived after the 9/11 attacks. The results show that new immigrants did not settle
down across states in a way that produced a correlation between hate crimes and
country of origin. Therefore, there is no evidence that new immigrants, with
observable tendencies to assimilate less, settled down in states with a higher intensity
of hate crimes against Muslims.

To summarise, the backlash against Muslims after the 9/11 attacks was larger in
states which tended to be less familiar with immigrants (i.e. a higher share of native
born residents) and already had a history of abundant hate crimes against minorities
(Blacks and Muslims). However, we present several arguments in favour of our
identifying assumption that local hate crimes against Muslims after 9/11 were not
correlated with unobserved factors which affected the assimilation rate of Muslims in
the aftermath of the attacks. First, we showed that hate crimes against Muslims after 9/
11 were not associated with the pre-existing levels or trends in the demographic and
assimilation characteristics of Muslims before the attacks (Figures 3–17, Tables 2–4).
Second, Table 5 demonstrates that local hate crimes were not higher in states that
were increasingly populated by the type of Muslim immigrants (according to their
country of origin) which tend to be less assimilated. Further support for our
identifying assumption is provided later when we show the robustness of the results
across different assimilation outcomes, and also to the inclusion or exclusion of a vast
array of personal and state-level control variables. In addition, we perform a placebo
analysis on other groups (older Muslims and members of other ethnic groups) which
should not be affected by the anti-Muslim backlash after 2001. Finally, it is important
to note that the 9/11 attacks were a total surprise, which resulted in a sudden and
unexpected surge in anti-Muslim hate crimes. The extent of this backlash was
correlated with pre-existing factors which made the non-Muslim community more
prone to commit hate crimes but was uncorrelated with the trend or level of any
characteristic or behaviour that is associated with the assimilation rate of the local
Muslim community.

Taken as a whole, these results suggest that it is safe to regard the backlash against
Muslims in the wake of the 9/11 attacks as an exogenous surge in anti-Muslim
sentiment, with variation across states that can be exploited to examine whether the
backlash affected the assimilation rate of local Muslim immigrants.

4. The Effect of Anti-Muslim Hate Crimes on Intra-marriage Rates

We now analyse the effect of anti-Muslim hate crimes on our main assimilation
outcome variable: being ‘intra-married’ in the sense of marrying someone with a
similar background from a Muslim country. Table 6 presents the main results for
immigrant Muslim men, while Table 7 presents a similar analysis for immigrant
Muslim women. In the first column of both Tables, the ‘intra-marriage’ status of each
individual is regressed only on Muslim hate crimes per capita and dummy variables for
each sample year, with the reported coefficients representing the marginal effects
evaluated at the means from a probit analysis. For both men and women, the intra-

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2083



marriage rate increases with the number of hate crimes against Muslims per capita, but
is significant only for men. In the second column of each Table, ‘basic controls’ for
personal characteristics (age, number of years in the US, education, country of origin
fixed-effects) are added to the specification. The coefficients are virtually unchanged,

–0.4

–0.2

0

0.2

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males

1
99

0–
2

00
0 

C
h

an
ge

 i
n 

In
te

r-
m

ar
ri

ag
e

2000–10 Change in Muslim Hate Crimes Per Capita

Examining Pre-existing Trends

Fig. 7. State Analysis of Inter-marriage, 1990–2000
Notes. Female slope = 0.300 (t-stat = 0.98); Male slope = 0.421 (t-stat = 1.20). Regressions are
weighted by state sample size, represented by the size of each circle.

–0.4

–0.2

0

0.2

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males
2

00
0–

1
0 

C
h

an
ge

 i
n 

In
te

r-
m

ar
ri

ag
e

2000–10 Change in Muslim Hate Crimes Per Capita

Means by State

Fig. 6. State Analysis of Inter-marriage, 2000–10
Notes. Female slope = �0.352 (t-stat = �1.35); Male slope = �0.562 (t-stat = �1.98). Regressions
are weighted by state sample size, represented by the size of each circle.

© 2014 Royal Economic Society.

2084 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



although both are now very significant since the standard error for women became
much smaller. State fixed-effects are added to the specification in the third column of
Tables 6 and 7. Again, the coefficients for men and women remain statistically
significant.

–1

–0.5

0

0.5

1

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males

1
99

0–
2

00
0 

C
h

an
ge

 i
n 

N
u

m
be

r 
o

f 
C

h
il

dr
en

2000–10 Change in Muslim Hate Crimes Per Capita

Examining Pre-existing Trends

Fig. 9. State Analysis of Number of Children, 1990–2000
Notes. Female slope = �1.065 (t-stat = �0.90); Male slope = 0.882 (t-stat = 1.40). Regressions are
weighted by state sample size, represented by the size of each circle.

–0.5

0

0.5

1

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males
2

00
0–

1
0 

C
h

an
ge

 i
n 

N
u

m
be

r 
o

f 
C

h
il

dr
en

2000–10 Change in Muslim Hate Crimes Per Capita

Means by State

Fig. 8. State Analysis of Number of Children, 2000–10
Notes. Female slope = 2.226 (t-stat = 2.42); Male slope = 1.669 (t-stat = 2.67). Regressions are
weighted by state sample size, represented by the size of each circle.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2085



In the next column of both Tables, we examine whether the strong results for hate
crimes against Muslims are simply capturing a more widespread atmosphere of
animosity to minorities in general. To do this, we add the number of hate crimes
against blacks per capita (normalising by the state-level size of the black population

–0.5

0

0.5

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males

1
99

0–
2

00
0 

C
h

an
ge

 i
n 

C
h

il
dr

en
 <

 5
 Y

rs

2000–10 Change in Muslim Hate Crimes Per Capita

Examining Pre-existing Trends

Fig. 11. State Analysis – Children Less Than Age 5, 1990–2000
Notes. Female slope = �0.390 (t-stat = �0.68); Male slope = 0.400 (t-stat = 0.81). Regressions are
weighted by state sample size, represented by the size of each circle.

–0.2

0

0.2

0.4

0.6

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males
2

00
0–

1
0 

C
h

an
ge

 i
n 

C
h

il
dr

en
 <

 5
 Y

rs

2000–10 Change in Muslim Hate Crimes Per Capita

Means by State

Fig. 10. State Analysis – Children Less Than Age 5, 2000–10
Notes. Female slope = 1.634 (t-stat = 2.95); Male slope = 0.725 (t-stat = 1.75). Regressions are
weighted by state sample size, represented by the size of each circle.

© 2014 Royal Economic Society.

2086 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



according to the 2000 Census), and the log number of hate crimes against Jews and
homosexuals (we cannot normalise by the size of the Jewish and homosexual
populations since this information does not exist). Although hate crimes against these
groups are much more prevalent than anti-Muslim hate crimes, the estimated effects of

–0.4

–0.2

0

0.2

0.4

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males

1
99

0–
2

00
0 

C
h

an
ge

 i
n 

H
o

m
e 

S
ec

to
r

2000–10 Change in Muslim Hate Crimes Per Capita

Examining Pre-existing Trends

Fig. 13. State Analysis of Being at Home, 1990–2000
Notes. Female slope = �0.791 (t-stat = �1.68); Male slope = �0.155 (t-stat = �1.05). Regressions
are weighted by state sample size, represented by the size of each circle.

–0.4

–0.2

0

0.2

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males
2

00
0–

1
0 

C
h

an
ge

 i
n 

H
o

m
e 

S
ec

to
r

2000–10 Change in Muslim Hate Crimes Per Capita

Means by State

Fig. 12. State Analysis of Being at Home, 2000–10
Notes. Female slope = 0.855 (t-stat = 2.68); Male slope = 0.168 (t-stat = 0.93). Regressions are
weighted by state sample size, represented by the size of each circle.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2087



anti-Muslim hate crimes on the rate of intra-marriage become even larger after
controlling for other types of hate crimes. These findings suggest that the intra-
marriage rates of Muslim immigrants are responding very specifically to the level of
animosity towards Muslims, regardless of the overall level of bigotry to other minorities
which typically suffer much higher rates of hate crime.

–0.3

–0.2

–0.1

0

0.1

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males

1
99

0–
2

00
0 

C
h

an
ge

 i
n 

E
ng

li
sh

 a
t 

H
om

e

2000–10 Change in Muslim Hate Crimes Per Capita

Examining Pre-existing Trends

Fig. 15. State Analysis of English at Home, 1990–2000
Notes. Female slope = �0.159 (t-stat = �0.68); Male slope = 0.048 (t-stat = 0.19). Regressions are
weighted by state sample size, represented by the size of each circle.

–0.4

–0.2

0

0.2

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males
2

00
0–

1
0 

C
h

an
ge

 i
n 

E
ng

li
sh

 a
t 

H
om

e

2000–10 Change in Muslim Hate Crimes Per Capita

Means by State

Fig. 14. State Analysis of English at Home, 2000–10
Notes. Female slope = �0.434 (t-stat = �2.15); Male slope = �0.578 (t-stat = �2.58). Regressions
are weighted by state sample size, represented by the size of each circle.

© 2014 Royal Economic Society.

2088 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



Since our main variable of interest, anti-Muslim hate crimes, varies at the state
level for each time period, we now test whether the coefficient on this variable is
robust to the inclusion of other variables which vary at the state-year level. These

–0.2

0

0.2

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males

1
99

0–
2

00
0 

C
h

an
ge

 i
n 

S
pe

ak
 E

ng
li

sh
 W

el
l

2000–10 Change in Muslim Hate Crimes Per Capita

Examining Pre-existing Trends

Fig. 17. State Analysis of Speak English Well, 1990–2000
Notes. Female slope = �0.361 (t-stat = �0.65); Male slope = �0.645 (t-stat = �1.85). Regressions
are weighted by state sample size, represented by the size of each circle.

–0.4

–0.2

0

0.2

0.4

0 0.05 0.1 0.15 0 0.05 0.1 0.15

Females Males
2

00
0–

1
0 

C
h

an
ge

 i
n 

S
pe

ak
 E

ng
li

sh
 W

el
l

2000–10 Change in Muslim Hate Crimes Per Capita

Means by State

Fig. 16. State Analysis of Speak English Well, 2000–10
Notes. Female slope = �0.609 (t-stat = �1.51); Male slope = �0.782 (t-stat = �2.65). Regressions
are weighted by state sample size, represented by the size of each circle.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2089



variables include measures for the state-level marriage market conditions for
Muslims (percentage of Muslims in the state and the percentage of males among
Muslims in the state) and for state-year economic conditions (percentage of native-
born females in the home sector, the unemployment rate of native-born men and
the mean log wage for native-born men). In principle, these variables could be
endogenous to the level of hate crime against Muslims, particularly the marriage
market variables. However, in column 5 of Tables 6 and 7, the estimated effect of
anti-Muslim hate crimes on the intra-marriage rate of men and women remains
statistically significant and virtually of the same magnitude after including these
additional state-year controls. It is worth noting that the gender composition of the
local Muslim community is significant and displays the expected sign – a higher
percentage of men versus women in the state decreases the likelihood that a Muslim
male will intra-marry while increasing the chances for Muslim females to do the
same. The fact that the intra-marriage rate is responding in expected ways to factors
which should be relevant justifies the use of this measure as an accurate and
meaningful assimilation outcome.

In column 6 of Tables 6 and 7, we use an alternative measure of anti-Muslim hate
crimes which considers only the hate crimes that occurred before the 2001 attacks and
those that occurred right after the attacks (only 2001 and 2002). In other words, we
ignore anti-Muslim hate crimes that occurred after 2002, since it could be the case that
hate crimes after 2002 were responding endogenously to changes in the local Muslim
community’s behaviour and characteristics, while the sudden burst of anti-Muslim
activity after the 9/11 attacks was more of an unexpected, exogenous shock. Column 6
shows that the results are still very significant after discarding the hate crime
information after 2002.

So far, the results show that anti-Muslim hate crimes led Muslim immigrant men and
women to a less-assimilated marriage outcome and that these findings are robust to
specifications which include very few controls or a vast array of personal and state-level
characteristics. The robustness of the results to the inclusion of virtually any control
variable provides support for the identifying assumption that anti-Muslim hate crimes
are not correlated with an omitted variable which affects assimilation outcomes. The
estimated coefficients are not only statistically significant and robust across different
specifications but also substantial in magnitude. The average number of hate crimes per
capita after 9/11 is equal to 0.028, which implies an increase in the intra-marriage rate
of 2% for men and 4.3% for women, according to the coefficients in column 5. These
numbers constitute a 6.4% and a 7.3% increase in the intra-marriage rate of Muslim
men and women respectively, relative to their means in the 2000 Census (0.31 for men
and 0.60 for women).

We now address whether these findings are due to Muslim immigrants simply
getting married at younger ages, or whether they are choosing to marry within their
community at the expense of marrying someone outside of their ethnic group. The
last two columns of Tables 6 and 7 show very clearly that anti-Muslim hate crimes
are significantly lowering the rate of marrying outside of the group (inter-marriage)
for men and women, while increasing the chances of being married only for
women. In particular, for an individual experiencing the mean rate of Muslim hate
crimes after 9/11, the estimated coefficients imply a decrease of 8.3% and 13.1% in

© 2014 Royal Economic Society.

2090 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



T
a
b
le

6

M
a
rr
ia
ge

P
a
tt
er
n
s
of

Im
m
ig
ra
n
t
M
u
sl
im

M
en

In
tr
a
-m

a
rr
ie
d

In
te
r-
m
a
rr
ie
d

E
ve
r
m
a
rr
ie
d

(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

(9
)

M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.6
8
4
*
*

0
.6
4
2
*
*

0
.6
8
8
*
*
*

0
.8
6
7
*
*
*

0
.6
9
7
*
*
*

�0
.4
7
2
*
*

�0
.0
7
5

(0
.2
9
6
)

(0
.2
5
7
)

(0
.2
1
4
)

(0
.2
6
4
)

(0
.2
4
0
)

(0
.2
3
2
)

(0
.3
2
5
)

M
u
sl
im

h
a
te

cr
im

e
s
P
C
,
2
0
0
1
–
2

1
.7
5
6
*
*
*

(0
.5
6
3
)

L
e
a
d
M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.3
0
0

(0
.2
6
6
)

A
g
e

0
.2
9
5
*
*
*

0
.2
9
9
*
*
*

0
.2
9
9
*
*
*

0
.2
9
9
*
*
*

0
.2
9
8
*
*
*

0
.1
3
3
*
*

0
.2
2
9
*
*
*

0
.2
4
1
*
*
*

(0
.0
6
5
)

(0
.0
6
6
)

(0
.0
6
6
)

(0
.0
6
7
)

(0
.0
6
7
)

(0
.0
6
1
)

(0
.0
4
4
)

(0
.0
4
8
)

A
g
e
-s
q
u
a
re
d

�0
.0
0
7
*
*
*

�0
.0
0
7
*
*
*

�0
.0
0
7
*
*
*

�0
.0
0
7
*
*
*

�0
.0
0
7
*
*
*

�0
.0
0
2

�0
.0
0
6
*
*
*

�0
.0
0
4
*
*
*

(0
.0
0
2
)

(0
.0
0
2
)

(0
.0
0
2
)

(0
.0
0
2
)

(0
.0
0
2
)

(0
.0
0
2
)

(0
.0
0
1
)

(0
.0
0
2
)

A
g
e
cu

b
e
d

0
.0
0
0
*
*

0
.0
0
0
*
*

0
.0
0
0
*
*

0
.0
0
0
*
*

0
.0
0
0
*
*

0
.0
0
0

0
.0
0
0
*
*
*

0
.0
0
0

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

Y
e
a
rs

in
U
S

0
.0
0
1

0
.0
0
0

0
.0
0
0

0
.0
0
0

0
.0
0
0

0
.0
0
4

0
.0
0
6
*
*
*

0
.0
0
9
*
*
*

(0
.0
0
2
)

(0
.0
0
2
)

(0
.0
0
2
)

(0
.0
0
2
)

(0
.0
0
2
)

(0
.0
0
3
)

(0
.0
0
1
)

(0
.0
0
3
)

Y
e
a
rs

in
U
S
sq
u
a
re
d

�0
.0
0
0
*
*
*

�0
.0
0
0
*
*
*

�0
.0
0
0
*
*
*

�0
.0
0
0
*
*
*

�0
.0
0
0
*
*
*

�0
.0
0
0
*
*
*

�0
.0
0
0
*
*

�0
.0
0
0
*
*
*

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

H
ig
h
sc
h
o
o
l
g
ra
d
u
a
te

�0
.0
2
2
*

�0
.0
2
2

�0
.0
2
2

�0
.0
2
3

�0
.0
2
3

�0
.0
1
7

0
.0
6
2
*
*
*

�0
.0
0
3

(0
.0
1
3
)

(0
.0
1
4
)

(0
.0
1
4
)

(0
.0
1
4
)

(0
.0
1
4
)

(0
.0
1
6
)

(0
.0
1
7
)

(0
.0
2
4
)

S
o
m
e
co

ll
e
g
e

�0
.0
5
5
*
*
*

�0
.0
5
5
*
*
*

�0
.0
5
5
*
*
*

�0
.0
5
6
*
*
*

�0
.0
5
6
*
*
*

�0
.0
6
0
*
*
*

0
.1
1
3
*
*
*

�0
.0
2
1

(0
.0
1
1
)

(0
.0
1
0
)

(0
.0
0
9
)

(0
.0
1
0
)

(0
.0
1
0
)

(0
.0
1
6
)

(0
.0
1
9
)

(0
.0
2
2
)

C
o
ll
e
g
e
g
ra
d
u
a
te

0
.0
2
2
*
*

0
.0
2
0
*

0
.0
2
0
*

0
.0
1
8

0
.0
1
8
*

0
.0
1
3

0
.0
6
3
*
*
*

�0
.0
4
0

(0
.0
1
1
)

(0
.0
1
1
)

(0
.0
1
1
)

(0
.0
1
1
)

(0
.0
1
1
)

(0
.0
1
4
)

(0
.0
1
2
)

(0
.0
2
6
)

B
la
ck

h
a
te

cr
im

e
s
P
C

0
.2
0
6

�0
.0
4
6

�0
.0
0
1

�0
.3
6
3
*
*

�0
.2
4
6

(0
.2
2
5
)

(0
.1
9
0
)

(0
.1
7
2
)

(0
.1
4
9
)

(0
.2
2
4
)

L
o
g
h
o
m
o
se
x
u
a
l
h
at
e
cr
im

e
s

�0
.0
4
9
*
*

�0
.0
2
2

�0
.0
2
1

0
.0
1
2

0
.0
1
5

(0
.0
2
0
)

(0
.0
1
8
)

(0
.0
1
8
)

(0
.0
1
4
)

(0
.0
1
7
)

L
o
g
Je
w
is
h
h
a
te

cr
im

e
s

0
.0
2
3

0
.0
0
7

0
.0
0
9

�0
.0
2
4

�0
.0
3
6

(0
.0
1
9
)

(0
.0
1
7
)

(0
.0
1
7
)

(0
.0
2
3
)

(0
.0
2
8
)

S
ta
te

%
m
a
le

a
m
o
n
g
M
u
sl
im

s
�0

.7
3
9
*
*
*

�0
.7
4
8
*
*
*

0
.1
4
5
*

�0
.5
0
4
*
*
*

(0
.1
1
3
)

(0
.1
1
4
)

(0
.0
8
4
)

(0
.1
1
2
)

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2091



T
a
b
le

6

(C
on
ti
n
u
ed
)

In
tr
a
-m

a
rr
ie
d

In
te
r-
m
a
rr
ie
d

E
ve
r
m
a
rr
ie
d

(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

(9
)

S
ta
te

%
M
u
sl
im

�1
9
.2
0
6
*
*
*

�1
7
.9
5
0
*
*
*

1
.1
2
4

�2
6
.8
4
7
*
*
*

(5
.5
9
8
)

(6
.0
0
0
)

(6
.1
2
3
)

(8
.2
2
9
)

S
ta
te

fe
m
a
le

%
h
o
m
e
(n

a
ti
ve
-b
o
rn
)

�0
.9
0
9

�1
.0
3
7

�0
.6
2
9

�1
.2
8
6

(0
.7
1
8
)

(0
.6
9
5
)

(0
.4
2
1
)

(0
.8
7
4
)

S
ta
te

M
a
le

U
n
e
m
p
.
R
a
te

(N
a
ti
ve
-b
o
rn
)

0
.8
8
9
*
*

0
.8
4
0
*
*

0
.1
9
4

0
.4
7
3

(0
.4
2
5
)

(0
.3
9
8
)

(0
.4
4
7
)

(0
.4
3
8
)

S
ta
te

m
a
le

m
e
an

w
a
g
e
(n

a
ti
ve
-b
o
rn
)

0
.0
3
8

�0
.0
4
0

0
.0
9
0

�0
.1
5
1

(0
.1
7
4
)

(0
.1
6
9
)

(0
.1
6
9
)

(0
.1
9
4
)

Y
e
a
rs

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

1
9
9
0
–
2
0
0
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

Y
e
a
r
fi
x
e
d
e
ff
e
ct
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

B
a
si
c
co

n
tr
o
ls

N
o

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

C
o
u
n
tr
y
o
f
o
ri
g
in

fi
x
e
d
e
ff
e
ct
s

N
o

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

S
ta
te

fi
x
e
d
e
ff
e
ct
s

N
o

N
o

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

O
th
e
r
st
a
te
-y
e
a
r
co

n
tr
o
ls

N
o

N
o

N
o

N
o

Y
e
s

Y
e
s

N
o

Y
e
s

Y
e
s

O
b
se
rv
a
ti
o
n
s

1
9
,9
4
0

1
9
,9
4
0

1
9
,9
4
0

1
9
,9
4
0

1
9
,9
4
0

1
9
,9
4
0

1
9
,3
9
3

1
9
,9
1
8

2
0
,4
8
5

N
ot
es
.
T
h
e
re
p
o
rt
e
d
co

e
ffi
ci
e
n
ts
a
re

th
e
m
a
rg
in
a
l
e
ff
e
ct
s
fr
o
m

a
p
ro
b
it
,
co

m
p
u
te
d
a
t
th
e
m
e
a
n
s
o
f
th
e
in
d
e
p
e
n
d
e
n
t
va
ri
a
b
le
s.
S
ta
n
d
ar
d
e
rr
o
rs
,
w
h
ic
h
a
re

cl
u
st
e
re
d
b
y

st
a
te
,
a
p
p
e
a
r
in

p
a
re
n
th
e
se
s.
S
ig
n
ifi
ca
n
ce

le
ve
ls
a
re

in
d
ic
a
te
d
b
y
*
,
*
*
,
o
r
*
*
*
w
h
ic
h
re
p
re
se
n
t
1
0
%
,
5
%
,
a
n
d
1
%

le
ve
ls
,
re
sp
e
ct
iv
e
ly
.
T
h
e
‘b
a
si
c
co

n
tr
o
ls
’
in
cl
u
d
e
th
e

p
e
rs
o
n
a
l
ch

a
ra
ct
e
ri
st
ic
s
a
d
d
e
d
to

th
e
sp
e
ci
fi
ca
ti
o
n
in

co
lu
m
n
(2
)
re
la
ti
ve

to
co

lu
m
n
(1
),
w
h
il
e
th
e
co

u
n
tr
ie
s
in
cl
u
d
e
d
in

th
e
‘c
o
u
n
tr
y
o
f
o
ri
g
in
’
fi
x
e
d
-e
ff
e
ct
s
a
re

li
st
e
d

in
A
p
p
e
n
d
ix

T
a
b
le

A
1
.

© 2014 Royal Economic Society.

2092 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



T
a
b
le

7

M
a
rr
ia
ge

P
a
tt
er
n
s
of

Im
m
ig
ra
n
t
M
u
sl
im

W
om

en

In
tr
a
-m

a
rr
ie
d

In
te
r-
m
a
rr
ie
d

E
ve
r
m
a
rr
ie
d

(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

(9
)

M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.5
3
2

0
.5
1
1
*

1
.2
3
5
*
*
*

1
.3
8
5
*
*
*

1
.5
4
3
*
*
*

�0
.4
6
2
*
*
*

0
.5
6
7
*
*

(0
.5
0
7
)

(0
.2
8
2
)

(0
.2
6
4
)

(0
.2
6
3
)

(0
.2
8
2
)

(0
.1
7
3
)

(0
.2
8
8
)

M
u
sl
im

h
a
te

cr
im

e
s
P
C
,
2
0
0
1
–
2

3
.3
6
3
*
*
*

(0
.8
4
2
)

L
e
a
d
m
u
sl
im

h
a
te

cr
im

e
s
P
C

�0
.8
3
8
*
*

(0
.4
1
8
)

A
g
e

0
.4
8
2
*
*
*

0
.4
8
5
*
*
*

0
.4
8
8
*
*
*

0
.4
8
5
*
*
*

0
.4
8
6
*
*
*

0
.3
5
5
*
*
*

0
.0
9
8
*
*
*

0
.2
1
7
*
*
*

(0
.0
8
8
)

(0
.0
8
8
)

(0
.0
8
9
)

(0
.0
8
8
)

(0
.0
8
8
)

(0
.0
6
1
)

(0
.0
3
6
)

(0
.0
5
4
)

A
g
e
-s
q
u
a
re
d

�0
.0
1
2
*
*
*

�0
.0
1
2
*
*
*

�0
.0
1
2
*
*
*

�0
.0
1
2
*
*
*

�0
.0
1
2
*
*
*

�0
.0
0
8
*
*
*

�0
.0
0
2
*
*

�0
.0
0
4
*
*

(0
.0
0
3
)

(0
.0
0
3
)

(0
.0
0
3
)

(0
.0
0
3
)

(0
.0
0
3
)

(0
.0
0
2
)

(0
.0
0
1
)

(0
.0
0
2
)

A
g
e
cu

b
e
d

0
.0
0
0
*
*
*

0
.0
0
0
*
*
*

0
.0
0
0
*
*
*

0
.0
0
0
*
*
*

0
.0
0
0
*
*
*

0
.0
0
0
*
*

0
.0
0
0

0
.0
0
0

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

Y
e
a
rs

in
U
S

�0
.0
1
3
*
*
*

�0
.0
1
4
*
*
*

�0
.0
1
4
*
*
*

�0
.0
1
4
*
*
*

�0
.0
1
4
*
*
*

�0
.0
0
9
*
*
*

0
.0
0
3
*
*
*

�0
.0
0
9
*
*
*

(0
.0
0
1
)

(0
.0
0
2
)

(0
.0
0
2
)

(0
.0
0
2
)

(0
.0
0
2
)

(0
.0
0
2
)

(0
.0
0
1
)

(0
.0
0
1
)

Y
e
a
rs

in
U
S
sq
u
a
re
d

�0
.0
0
0
*

�0
.0
0
0

�0
.0
0
0

�0
.0
0
0

�0
.0
0
0

�0
.0
0
0
*
*
*

0
.0
0
0

0
.0
0
0
*
*

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

(0
.0
0
0
)

H
ig
h
sc
h
o
o
l
g
ra
d
u
a
te

�0
.0
3
7
*
*

�0
.0
3
8
*
*

�0
.0
3
8
*
*

�0
.0
3
8
*
*

�0
.0
3
8
*
*

�0
.0
2
6
*

0
.0
4
7
*
*
*

�0
.0
1
0

(0
.0
1
7
)

(0
.0
1
8
)

(0
.0
1
8
)

(0
.0
1
8
)

(0
.0
1
8
)

(0
.0
1
5
)

(0
.0
1
8
)

(0
.0
1
4
)

S
o
m
e
co

ll
e
g
e

�0
.1
6
0
*
*
*

�0
.1
5
8
*
*
*

�0
.1
5
8
*
*
*

�0
.1
5
8
*
*
*

�0
.1
5
8
*
*
*

�0
.1
5
0
*
*
*

0
.0
7
5
*
*
*

�0
.1
0
7
*
*
*

(0
.0
1
7
)

(0
.0
1
6
)

(0
.0
1
7
)

(0
.0
1
7
)

(0
.0
1
7
)

(0
.0
1
5
)

(0
.0
1
6
)

(0
.0
2
0
)

C
o
ll
e
g
e
g
ra
d
u
a
te

�0
.1
9
3
*
*
*

�0
.1
8
9
*
*
*

�0
.1
8
9
*
*
*

�0
.1
9
0
*
*
*

�0
.1
9
0
*
*
*

�0
.1
7
2
*
*
*

0
.0
7
4
*
*
*

�0
.1
4
8
*
*
*

(0
.0
1
3
)

(0
.0
1
2
)

(0
.0
1
2
)

(0
.0
1
3
)

(0
.0
1
3
)

(0
.0
1
6
)

(0
.0
1
1
)

(0
.0
1
7
)

B
la
ck

h
a
te

cr
im

e
s
P
C

0
.5
4
2
*

0
.4
6
5

0
.4
9
5
*

�0
.2
4
2
*
*

0
.2
7
1

(0
.2
8
5
)

(0
.3
1
6
)

(0
.2
8
7
)

(0
.1
2
3
)

(0
.2
2
3
)

L
o
g
h
o
m
o
se
x
u
a
l
h
a
te

cr
im

e
s

�0
.0
1
5

�0
.0
2
5

�0
.0
1
9

0
.0
1
1

0
.0
2
5
*

(0
.0
2
4
)

(0
.0
2
2
)

(0
.0
2
2
)

(0
.0
1
2
)

(0
.0
1
3
)

L
o
g
Je
w
is
h
h
a
te

cr
im

e
s

0
.0
0
7

0
.0
1
5

0
.0
1
9

�0
.0
1
1

�0
.0
4
3
*
*

(0
.0
2
3
)

(0
.0
2
4
)

(0
.0
2
6
)

(0
.0
1
3
)

(0
.0
1
7
)

S
ta
te

%
m
a
le

a
m
o
n
g
M
u
sl
im

s
0
.3
8
1
*
*
*

0
.3
5
0
*
*

�0
.1
0
6
*

0
.0
6
2

(0
.1
4
1
)

(0
.1
3
8
)

(0
.0
6
1
)

(0
.1
1
5
)

S
ta
te

%
M
u
sl
im

0
.5
9
2

2
.9
5
8

�1
3
.2
4
3
*
*
*

�1
2
.3
3
2
*
*

(9
.6
0
4
)

(9
.7
2
0
)

(3
.8
1
0
)

(5
.7
6
0
)

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2093



T
a
b
le

7

(C
on
ti
n
u
ed
)

In
tr
a
-m

a
rr
ie
d

In
te
r-
m
a
rr
ie
d

E
ve
r
m
a
rr
ie
d

(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

(9
)

S
ta
te

fe
m
a
le

%
h
o
m
e
(n

a
ti
ve
-b
o
rn
)

�1
.8
3
6
*
*

�2
.0
6
3
*
*

1
.0
4
5
*
*

0
.2
5
7

(0
.8
4
9
)

(0
.8
2
8
)

(0
.4
3
4
)

(0
.5
6
2
)

S
ta
te

m
a
le

u
n
e
m
p
.
ra
te

(n
a
ti
ve
-b
o
rn
)

0
.5
3
7

0
.4
4
9

�0
.3
1
0

�0
.6
1
7

(0
.6
5
8
)

(0
.6
0
1
)

(0
.2
4
9
)

(0
.4
3
8
)

S
ta
te

m
a
le

m
e
a
n
w
a
g
e
(n

a
ti
ve
-b
o
rn
)

0
.0
1
6

�0
.1
8
8

�0
.2
0
4
*
*

�0
.4
2
8
*
*
*

(0
.2
4
2
)

(0
.2
2
7
)

(0
.1
0
0
)

(0
.1
5
6
)

Y
e
a
rs

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

1
9
9
0
–
2
0
0
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

Y
e
a
r
fi
x
e
d
e
ff
e
ct
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

B
a
si
c
co

n
tr
o
ls

N
o

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

C
o
u
n
tr
y
o
f
o
ri
g
in

fi
x
e
d
e
ff
e
ct
s

N
o

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

S
ta
te

fi
x
e
d
e
ff
e
ct
s

N
o

N
o

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

O
th
e
r
st
a
te
-y
e
a
r
co

n
tr
o
ls

N
o

N
o

N
o

N
o

Y
e
s

Y
e
s

N
o

Y
e
s

Y
e
s

O
b
se
rv
a
ti
o
n
s

1
6
,7
4
9

1
6
,7
4
9

1
6
,7
4
9

1
6
,7
4
9

1
6
,7
4
9

1
6
,7
4
9

1
3
,3
9
3

1
6
,7
4
1

1
7
,1
5
6

N
ot
es
.
T
h
e
re
p
o
rt
e
d
co

e
ffi
ci
e
n
ts
a
re

th
e
m
a
rg
in
a
l
e
ff
e
ct
s
fr
o
m

a
p
ro
b
it
,
co

m
p
u
te
d
a
t
th
e
m
e
a
n
s
o
f
th
e
in
d
e
p
e
n
d
e
n
t
va
ri
a
b
le
s.
S
ta
n
d
ar
d
e
rr
o
rs
,
w
h
ic
h
a
re

cl
u
st
e
re
d
b
y

st
a
te
,
a
p
p
e
a
r
in

p
a
re
n
th
e
se
s.
S
ig
n
ifi
ca
n
ce

le
ve
ls
a
re

in
d
ic
a
te
d
b
y
*
,
*
*
,
o
r
*
*
*
w
h
ic
h
re
p
re
se
n
t
1
0
%
,
5
%
,
a
n
d
1
%

le
ve
ls
,
re
sp
e
ct
iv
e
ly
.
T
h
e
‘b
a
si
c
co

n
tr
o
ls
’
in
cl
u
d
e
th
e

p
e
rs
o
n
a
l
ch

a
ra
ct
e
ri
st
ic
s
a
d
d
e
d
to

th
e
sp
e
ci
fi
ca
ti
o
n
in

co
lu
m
n
(2
)
re
la
ti
ve

to
co

lu
m
n
(1
),
w
h
il
e
th
e
co

u
n
tr
ie
s
in
cl
u
d
e
d
in

th
e
‘c
o
u
n
tr
y
o
f
o
ri
g
in
’
fi
x
e
d
-e
ff
e
ct
s
a
re

li
st
e
d

in
A
p
p
e
n
d
ix

T
a
b
le

A
1
.

© 2014 Royal Economic Society.

2094 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



the inter-marriage rate of men and women, relative to their respective means in
2000. Therefore, the increasing rate of intra-marriage for men is coming at the
expense of a declining rate of inter-marriage, while the increasing rate of intra-
marriage for women is coming from both the declining rate of inter-marriage and
the increasing probability of being married. For both men and women, the results
point to a striking phenomenon whereby immigrants from Muslim countries are
increasing their ethnic identity in response to the backlash in the wake of the 9/11
attacks.

We now examine whether our main findings stem from a correlation between the
state-level backlash against Muslims and state-level trends in unobserved factors which
affect their assimilation rate. First, we perform a similar analysis with data from the pre-
2001 period, and test whether the state-level backlash against Muslims after the 2001
attacks was correlated with changes in the intra-marriage rate before the attacks
(between 1990 and 2000). This placebo analysis is presented in column 7 of Tables 6
and 7 where we estimate the effect of anti-Muslim hate crimes between 2000 and 2010
(‘Lead Muslim Hate Crimes PC’) on the intra-marriage outcomes between 1990 and
2000. A comparison between columns 3 and 7 of both Tables reveals that hate crimes
against Muslims during the 2000–10 period are very significant and positive
determinants of intra-marriage during the same period but are insignificant (for
men) or significantly negative (for women) for intra-marriage during the pre-9/11
period between 1990 and 2000. Figures 3–7 show similar results for the state-level
means in the intra-marriage and inter-marriage rates, but contain no other controls in
the regression.

A further placebo analysis is presented in Table 8 for all Muslim immigrants and for
only those that are married. A simple comparison of columns 1 and 7 shows that our
main results tend to be stronger for men when the sample is restricted to married
individuals but smaller in magnitude (still significant) for women because hate crimes
had a much larger effect on the marriage rate for women versus men. In the second
column for each group in Table 8, we exploit the idea that older Muslims are less able
to respond to the sudden backlash after 9/11 in terms of intra-marriage, since they
most likely already made their marriage decisions prior to the attacks. However, they
are likely to be affected by unobserved determinants shaping the local trends in
assimilation. Therefore, a second placebo strategy is to perform a similar analysis on
Muslim immigrants who were most likely too old to respond to the 9/11 backlash in
the marital outcome. In particular, the sample in the second column of Table 8 is
restricted to Muslim immigrants who were older than 40 in the survey year. That is, the
youngest people in this sample were 31 at the time of the attacks (those that are 41 in
the 2010 ACS Survey).

Consistent with Figure 4, the results in Table 8 indicate that the intra-marriage rate
for older Muslim immigrants was indeed unaffected by the state-level backlash against
Muslims after 9/11. The differential results for those that were young enough to be
affected by the backlash versus those that were older, provides further evidence that
the changes in the local assimilation rate was a response to the backlash and not due
to unobserved factors which should affect young and old alike. The remaining
columns in Table 8 pool both samples together, and show that the differential effect is
significant. In addition, using the older group as a control group within each state,

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2095



T
a
b
le

8

In
tr
a
-m
a
rr
ia
ge

fo
r
Im

m
ig
ra
n
t
M
u
sl
im
s
w
it
h
In
te
ra
ct
io
n
s
w
it
h
A
ge
,
2
0
0
0
–
1
0

(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

(9
)

(1
0
)

(1
1
)

(1
2
)

A
ll
m
e
n

M
a
rr
ie
d
m
e
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.6
8
8
*
*
*

�0
.0
5
5

0
.8
6
4
*
*
*

2
.2
3
9
*
*
*

1
.1
5
6
*
*
*

�0
.1
8
6

0
.8
2
4
*
*
*

2
.3
6
8
*
*
*

(0
.2
1
4
)

(0
.2
7
3
)

(0
.1
6
3
)

(0
.4
9
3
)

(0
.3
4
5
)

(0
.2
9
4
)

(0
.1
6
9
)

(0
.5
0
1
)

M
u
sl
im

h
a
te

cr
im

e
s

P
C

9
a
b
o
ve

a
g
e

4
0

�0
.9
8
6
*
*
*

�0
.9
8
0
*
*
*

�0
.8
2
3
*
*
*

�0
.8
3
4
*
*
*

(0
.3
0
7
)

(0
.2
9
7
)

(0
.2
6
3
)

(0
.2
5
3
)

M
u
sl
im

h
a
te

cr
im

e
s
P
C

9
a
g
e

�0
.0
4
5
*
*
*

�0
.0
4
5
*
*
*

�0
.0
4
7
*
*
*

�0
.0
4
7
*
*
*

(0
.0
1
3
)

(0
.0
1
3
)

(0
.0
1
4
)

(0
.0
1
3
)

S
a
m
p
le

si
ze

1
9
,9
4
0

2
1
,4
2
7

4
1
,3
4
2

4
1
,3
4
2

4
1
,3
4
2

4
1
,3
4
2

1
1
,9
1
1

2
0
,2
9
9

3
2
,1
8
5

3
2
,1
8
5

3
2
,1
8
5

3
2
,1
8
5

A
ll
w
o
m
e
n

M
a
rr
ie
d
w
o
m
e
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

1
.2
3
5
*
*
*

0
.1
2
4

0
.8
2
3
*
*
*

1
.4
5
9
*
*
*

0
.8
0
6
*
*
*

0
.2
4
7

0
.7
0
0
*
*
*

1
.3
7
7
*
*
*

(0
.2
6
4
)

(0
.4
5
4
)

(0
.2
3
5
)

(0
.3
7
7
)

(0
.2
5
1
)

(0
.4
3
4
)

(0
.2
5
1
)

(0
.3
9
0
)

M
u
sl
im

h
a
te

cr
im

e
s

P
C

9
a
b
o
ve

a
g
e

4
0

�0
.4
0
1

�0
.4
0
8

�0
.3
4
5
*

�0
.3
6
1
*
*

(0
.2
6
5
)

(0
.2
6
8
)

(0
.1
7
9
)

(0
.1
8
0
)

M
u
sl
im

h
a
te

cr
im

e
s
P
C

9
a
g
e

�0
.0
2
1
*
*

�0
.0
2
1
*
*

�0
.0
2
1
*
*
*

�0
.0
2
2
*
*
*

(0
.0
1
0
)

(0
.0
1
0
)

(0
.0
0
7
)

(0
.0
0
7
)

S
a
m
p
le

si
ze

1
6
,7
4
9

1
3
,9
3
2

3
0
,6
1
5

3
0
,6
1
5

3
0
,6
1
5

3
0
,6
1
5

1
3
,0
7
8

1
3
,3
5
4

2
6
,3
6
6

2
6
,3
6
6

2
6
,3
6
6

2
6
,3
6
6

A
g
e
ra
n
g
e
s

2
0
–
4
0

4
1
–
7
0

2
0
–
7
0

2
0
–
7
0

2
0
–
7
0

2
0
–
7
0

2
0
–
4
0

4
1
–
7
0

2
0
–
7
0

2
0
–
7
0

2
0
–
7
0

2
0
–
7
0

Y
e
a
r
fi
x
e
d
e
ff
e
ct
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

S
ta
te

fi
x
e
d
e
ff
e
ct
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

S
ta
te

sp
e
ci
fi
c
ti
m
e

tr
e
n
d
s

N
o

N
o

N
o

Y
e
s

N
o

Y
e
s

N
o

N
o

N
o

Y
e
s

N
o

Y
e
s

N
ot
es
.
T
h
e
re
p
o
rt
e
d
co

e
ffi
ci
e
n
ts
a
re

th
e
m
a
rg
in
a
l
e
ff
e
ct
s
fr
o
m

a
p
ro
b
it
,
co

m
p
u
te
d
a
t
th
e
m
e
an

s
o
f
th
e
in
d
e
p
e
n
d
e
n
t
va
ri
a
b
le
s.
S
ta
n
d
ar
d
e
rr
o
rs
,
w
h
ic
h
a
re

cl
u
st
e
re
d
b
y

st
a
te
,
a
p
p
e
ar

in
p
a
re
n
th
e
se
s.
S
ig
n
ifi
ca
n
ce

le
ve
ls
a
re

in
d
ic
a
te
d
b
y
*
,
*
*
,
o
r
*
*
*
w
h
ic
h
re
p
re
se
n
t
1
0
%
,
5
%
,
a
n
d
1
%

le
ve
ls
,
re
sp
e
ct
iv
e
ly
.
E
a
ch

re
g
re
ss
io
n
in
cl
u
d
e
s
th
e
se
t
o
f

co
n
tr
o
lv
a
ri
a
b
le
s
in

th
e
co

re
sp
e
ci
fi
ca
ti
o
n
d
e
sc
ri
b
e
d
in

co
lu
m
n
(3
)
o
f
T
a
b
le

6
,i
n
a
d
d
it
io
n
to

th
o
se

in
d
ic
a
te
d
a
t
th
e
b
o
tt
o
m

o
f
e
a
ch

co
lu
m
n
.S
ta
te
-s
p
e
ci
fi
c
ti
m
e
tr
e
n
d
s
a
re

d
e
fi
n
e
d
a
s
st
a
te
fi
x
e
d
-e
ff
e
ct
s
in
te
ra
ct
e
d
w
it
h
p
e
ri
o
d
,w

h
e
re
b
y
th
e
fi
rs
t
p
e
ri
o
d
is
fo
r
2
0
0
0
a
n
d
th
e
se
co

n
d
p
e
ri
o
d
is
fo
r
th
e
p
o
st
-9
/
1
1
p
e
ri
o
d
(2
0
0
6
–
1
0
).
T
h
e
sa
m
p
le
s
w
h
ic
h

in
cl
u
d
e
th
o
se

a
b
o
ve

th
e
a
g
e
o
f
4
0
a
re

re
st
ri
ct
e
d
to

th
o
se

th
a
t
a
re

n
o
t
w
id
o
w
s.

© 2014 Royal Economic Society.

2096 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



T
a
b
le

9

In
tr
a
-m
a
rr
ia
ge

fo
r
M
u
sl
im

Im
m
ig
ra
n
ts
by

S
u
bg
ro
u
p
s,
2
0
0
0
–
1
0

A
ll

im
m
ig
ra
n
ts

A
t
le
as
t

n
in
e

ye
a
rs

in
U
S

A
rr
iv
e
d

b
e
fo
re

a
g
e

2
0

A
rr
iv
e
d

a
ft
e
r

a
g
e

2
0

N
o
t
a

co
ll
e
g
e

g
ra
d

C
o
ll
e
g
e

g
ra
d

A
ll

im
m
ig
ra
n
ts

A
t
le
as
t

n
in
e

ye
ar
s
in

U
S

A
rr
iv
e
d

b
e
fo
re

a
g
e
2
0

A
rr
iv
e
d

a
ft
e
r
a
g
e
2
0

N
o
t
a

co
ll
e
g
e

g
ra
d

C
o
ll
e
g
e

g
ra
d

(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

(9
)

(1
0
)

(1
1
)

(1
2
)

M
e
n

W
o
m
e
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.8
6
7
*
*
*

0
.6
0
1
*

0
.5
5
3
*

0
.8
9
4
*

1
.0
2
3
*
*
*

0
.5
1
5

1
.3
8
5
*
*
*

0
.8
9
7
*

1
.8
9
8
*
*
*

0
.8
2
7
*
*
*

1
.6
5
0
*
*
*

0
.8
5
1
*
*

(0
.2
6
4
)

(0
.3
1
9
)

(0
.2
8
6
)

(0
.4
6
1
)

(0
.2
1
8
)

(0
.5
4
0
)

(0
.2
6
3
)

(0
.4
7
8
)

(0
.4
3
3
)

(0
.3
0
5
)

(0
.3
8
1
)

(0
.3
8
5
)

S
a
m
p
le

si
ze

1
9
,9
4
0

1
1
,3
9
6

9
,6
6
2

1
0
,2
4
3

1
0
,2
2
7

9
,6
7
9

1
6
,7
4
9

8
,9
8
9

7
,6
8
0

9
,0
4
4

9
,0
3
6

7
,7
0
1

M
a
rr
ie
d
m
e
n

M
a
rr
ie
d
w
o
m
e
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

1
.3
0
3
*
*
*

0
.9
7
4
*
*

0
.8
0
8

1
.4
0
4
*

1
.6
1
4
*
*
*

0
.9
6
2

0
.8
3
5
*
*
*

0
.6
7
9
*

1
.3
4
7
*
*

0
.5
2
6
*

0
.9
2
3
*
*

0
.5
9
2
*
*

(0
.4
3
4
)

(0
.4
2
3
)

(0
.5
2
1
)

(0
.7
2
7
)

(0
.5
5
2
)

(0
.6
1
5
)

(0
.2
5
6
)

(0
.4
0
3
)

(0
.5
5
1
)

(0
.2
7
3
)

(0
.3
9
4
)

(0
.2
3
4
)

S
a
m
p
le

si
ze

1
1
,9
1
1

7
,4
9
5

4
,7
3
7

7
,1
6
2

5
,6
3
6

6
,2
5
8

1
3
,0
7
8

6
,9
0
9

4
,9
0
3

8
,1
4
0

7
,1
1
6

5
,9
4
4

N
ot
es
.
T
h
e
re
p
o
rt
e
d
co

e
ffi
ci
e
n
ts
a
re

th
e
m
a
rg
in
a
l
e
ff
e
ct
s
fr
o
m

a
p
ro
b
it
,
co

m
p
u
te
d
a
t
th
e
m
e
an

s
o
f
th
e
in
d
e
p
e
n
d
e
n
t
va
ri
a
b
le
s.
S
ta
n
d
a
rd

e
rr
o
rs

a
p
p
e
a
r
in

p
a
re
n
th
e
se
s,

w
h
ic
h
a
re

cl
u
st
e
re
d
b
y
st
a
te
.
S
ig
n
ifi
ca
n
ce

le
ve
ls
a
re

in
d
ic
a
te
d
b
y
*
,
*
*
,
o
r
*
*
*
w
h
ic
h
re
p
re
se
n
t
1
0
%
,
5
%
,
a
n
d
1
%

le
ve
ls
,
re
sp
e
ct
iv
e
ly
.
E
a
ch

re
g
re
ss
io
n
in
cl
u
d
e
s
th
e
se
t
o
f

co
n
tr
o
l
va
ri
a
b
le
s
in

th
e
co

re
sp
e
ci
fi
ca
ti
o
n
d
e
sc
ri
b
e
d
in

co
lu
m
n
(3
)
o
f
T
a
b
le

6
.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2097



Table 8 shows that these patterns are significant after pooling both age groups
together and controlling for state-specific time trends. This later specification is
essentially a triple differences specification and the results demonstrate that the local
backlash caused younger individuals to increase their intra-marriage rate relative to
older Muslim immigrants in the same state. The last two columns show similar results
but instead of using a somewhat arbitrary threshold age for being ‘too old’ to be
affected, we simply interact the treatment variable (hate crimes) with the person’s age.
Again, the results show a significant positive effect of hate crimes on intra-marriage
(even after controlling for state-specific time trends) but the effect weakens
dramatically with age.

Along with Figures 3–7, Tables 6–8 show that the local backlash against Muslims
in the wake of the 9/11 attacks was not correlated with any pre-existing tendency
for local Muslims to become less assimilated over time and that hate crimes led
younger Muslim immigrants to a less-assimilated outcome relative to older Muslims
in the same state, even after controlling for a broad array of control variables and
for the state-specific trends of all unobserved factors affecting the marriage
decisions of both groups. These findings support a causal interpretation of our
main results in Tables 6 and 7.

Table 9 examines whether the effect of the backlash on the intra-marriage rate
varies across different segments of the Muslim immigrant population. Of notable
interest is the robustness of the main results to the sample that includes only those
immigrants who have been in the US for at least nine years – meaning that
everyone in this sample was in the US before the 9/11 attacks in 2001. This finding
once again demonstrates that our main results are not due to the placement of new
immigrants across locations according to the local upsurge in hate crimes. The
results are similar for those that arrived before or after the age of 20 and for
different education groups (with or without a college degree) – although the effects
appear to be stronger for the less educated immigrants (who tend to be more
assimilated) than those with more education. Overall, the evidence indicates that
the backlash against Muslims in the wake of the 9/11 attacks led all segments of the
Muslim population to a less assimilated marriage outcome. The next Section
examines whether these findings are robust to using other assimilation outcomes,
which would lend further support to the causal interpretation of our intra-marriage
results.

5. The Effect of Anti-Muslim Hate Crimes on other Assimilation Outcomes

We now perform a similar analysis with other assimilation outcomes measured by
the individual’s fertility, labour market participation, and English proficiency. For
each outcome, we first present results from a core specification, and then test
the robustness of the results to the inclusion of additional control variables. We
also examine whether our findings are possibly due to pre-existing state-specific
trends.

Table 10 examines whether anti-Muslim hate crimes affected the fertility level of
Muslim immigrants. Fertility is measured by the total number of children in the
household, or alternatively, by the number of children less than five years of age. The

© 2014 Royal Economic Society.

2098 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



T
a
b
le

1
0

E
ff
ec
t
of

H
a
te
C
ri
m
es

on
th
e
N
u
m
be
r
of

C
h
il
d
re
n
fo
r
Im

m
ig
ra
n
t
M
u
sl
im
s

M
e
n

W
o
m
e
n

(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

(9
)

(1
0
)

N
u
m
b
e
r
o
f
ch

il
d
re
n

N
u
m
b
e
r
o
f
ch

il
d
re
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

1
.3
9
9
*
*

1
.6
4
7
*
*

1
.4
8
2
*
*

1
.3
2
2
*
*

1
.5
6
0
*
*
*

2
.1
3
3
*
*
*

(0
.5
4
6
)

(0
.6
4
1
)

(0
.6
0
3
)

(0
.6
4
3
)

(0
.5
6
7
)

(0
.7
1
5
)

M
u
sl
im

h
a
te

cr
im

e
s
P
C
,
2
0
0
1
–
2

3
.3
9
7
*
*

4
.4
5
5
*
*

(1
.2
6
9
)

(2
.1
0
5
)

L
e
a
d
M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.4
3
8

�0
.4
9
2

(0
.4
9
2
)

(0
.5
9
6
)

N
u
m
b
e
r
o
f
ch

il
d
re
n
u
n
d
e
r
a
g
e
5

N
u
m
b
e
r
o
f
ch

il
d
re
n
u
n
d
e
r
a
g
e
5

M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.6
7
8
*

0
.7
2
2
*

0
.8
2
1
*
*

1
.1
5
7
*

1
.0
7
5

1
.4
4
1
*
*

(0
.3
9
1
)

(0
.4
1
0
)

(0
.4
0
6
)

(0
.6
1
6
)

(0
.6
8
2
)

(0
.6
6
7
)

M
u
sl
im

h
a
te

cr
im

e
s
P
C
,
2
0
0
1
–
2

2
.3
8
7
*
*

2
.5
9
5

(0
.9
0
8
)

(1
.8
9
9
)

L
e
a
d
M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.2
1
9

�0
.2
0
1

(0
.4
4
6
)

(0
.5
7
9
)

S
a
m
p
le

si
ze

2
0
,4
8
5

2
0
,4
8
5

2
0
,4
8
5

2
0
,4
8
5

1
9
,8
9
1

1
7
,1
6
2

1
7
,1
6
2

1
7
,1
6
2

1
7
,1
6
2

1
3
,6
6
4

Y
e
a
rs

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

1
9
9
0
–
2
0
0
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

1
9
9
0
–
2
0
0
0

Y
e
a
r
a
n
d
st
a
te

F
E

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

O
th
e
r
h
a
te

cr
im

e
s

N
o

Y
e
s

Y
e
s

Y
e
s

N
o

N
o

Y
e
s

Y
e
s

Y
e
s

N
o

O
th
e
r
st
a
te
-y
e
a
r
co

n
tr
o
ls

N
o

N
o

Y
e
s

Y
e
s

N
o

N
o

N
o

Y
e
s

Y
e
s

N
o

N
ot
es
.
T
h
e
re
p
o
rt
e
d
co

e
ffi
ci
e
n
ts
a
re

fr
o
m

O
L
S
re
g
re
ss
io
n
s.
S
ta
n
d
ar
d
e
rr
o
rs
,
w
h
ic
h
a
re

cl
u
st
e
re
d
b
y
st
a
te
,
a
p
p
e
ar

in
p
a
re
n
th
e
se
s.
S
ig
n
ifi
ca
n
ce

le
ve
ls
a
re

in
d
ic
a
te
d
b
y

*
,
*
*
,
o
r
*
*
*
w
h
ic
h
re
p
re
se
n
t
1
0
%
,
5
%
,
a
n
d
1
%

le
ve
ls
,
re
sp
e
ct
iv
e
ly
.
E
ac
h
re
g
re
ss
io
n
in
cl
u
d
e
s
th
e
se
t
o
f
co

n
tr
o
l
va
ri
a
b
le
s
in

th
e
co

re
sp
e
ci
fi
ca
ti
o
n
d
e
sc
ri
b
e
d
in

co
lu
m
n

(3
)
o
f
T
a
b
le

6
,
in

a
d
d
it
io
n
to

th
o
se

in
d
ic
a
te
d
a
t
th
e
b
o
tt
o
m

o
f
e
a
ch

co
lu
m
n
.
T
h
e
‘o
th
e
r
st
a
te
-y
e
a
r’

co
n
tr
o
ls
a
re

th
o
se

a
d
d
e
d
to

th
e
sp
e
ci
fi
ca
ti
o
n
in

co
lu
m
n
(5
)
o
f

T
a
b
le

6
:
p
e
rc
e
n
ta
g
e
m
a
le

a
m
o
n
g
o
f
th
o
se

o
f
M
u
sl
im

o
ri
g
in
,
p
e
rc
e
n
ta
g
e
w
it
h
M
u
sl
im

o
ri
g
in
,
p
e
rc
e
n
ta
g
e
o
f
fe
m
a
le
s
in

th
e
‘h
o
m
e
’
se
ct
o
r,
m
a
le

u
n
e
m
p
lo
ym

e
n
t
ra
te

a
m
o
n
g
n
at
iv
e
s,
m
a
le

m
e
an

w
a
g
e
a
m
o
n
g
n
a
ti
ve
s.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2099



advantage of the latter measure is that we know that children under the age of five in
the 2007–10 period were conceived after 2001 and, therefore, were the product of
decisions taken in the aftermath of the attacks.

The first column for each group in Table 10 presents the core specification which
includes individual controls and fixed-effects for each year and state.9 Using either
measure of fertility, the results show that anti-Muslim hate crimes led men and women
to increase their fertility. The results are a bit stronger when other hate crimes (against
blacks, Jews and homosexuals) and other state-year controls (related to the marriage
market and economic conditions) are added to the specification in the second and
third columns for both men and women. The estimated coefficients imply that the
number of children in a Muslim household increased by 5.7% and by 4.5% for males
and females, respectively, as a consequence of the mean level of hate crimes against
Muslims in the aftermath of 9/11. If we measure the anti-Muslim backlash by the
unexpected surge in hate crimes occurring right after the 9/11 attacks (i.e. anti-
Muslim hate crimes in 2001–2 only), the results are still significant (columns 4 and 9 of
Table 10).

The last column for both groups in Table 10 uses the sample prior to the 9/11
attacks (1990–2000) and shows very clearly that the surge in hate crimes against
Muslims after 9/11 was not correlated with pre-existing state-level trends in
fertility rates. These findings are illustrated very simply (without any additional
control variables) in Figures 8–11, which show that hate crimes against Muslims
from 2000–10 are strongly correlated with the changes in fertility during the
2000–10 period but uncorrelated with fertility in the previous period from 1990–
2000.

Since the previous Section showed that hate crimes against Muslims affected
marriage patterns, it is natural to wonder whether the fertility results are coming
indirectly from changes in the propensity to get married. Table 11 investigates this
issue and also presents results for different segments of the Muslim immigrant
population. In general, the results are stronger for married men versus all men but the
reverse pattern holds for women. These findings are consistent with our previous
results that anti-Muslim hate crimes affected the marriage rate for women but not for
men, so the indirect channel of affecting fertility through the marriage rate should be
larger for women versus men. Furthermore, Table 11 displays similar results, although
not always significant, across a broad spectrum of the Muslim immigrant population
according to their education, when they arrived in the US, and whether they were in
the US at the time of the 9/11 attacks.

Overall, the unexpected increase in hate crimes against Muslims after the 9/11
attacks appears to have made the Muslim community more insular in terms of
increasing the rate of intra-marriage, while at the same reinforcing traditional customs
by increasing their fertility and subsequent demographic strength.10 Once again, these

9 The core specification is detailed in column 3 of Tables 6 and 7.
10 These results are consistent with an increasing preference by US Muslim immigrants for having children

that religiously identify with Islam. Bisin et al. (2004) label this the ‘relative intolerance’ parameter. Their
dynamic analysis shows that small changes in this parameter have substantial and long lasting effects on the
religious composition of the population.

© 2014 Royal Economic Society.

2100 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



T
a
b
le

1
1

E
ff
ec
t
of

H
a
te
C
ri
m
es

on
th
e
N
u
m
be
r
of

C
h
il
d
re
n
fo
r
M
u
sl
im

Im
m
ig
ra
n
ts
by

S
u
bg
ro
u
p
s,
2
0
0
0
–
1
0

A
ll
im

m
ig
ra
n
ts

A
t
le
a
st

n
in
e
ye
a
rs

in
U
S

A
rr
iv
e
d

b
e
fo
re

a
g
e
2
0

A
rr
iv
e
d

a
ft
e
r

a
g
e
2
0

N
o
t
a

co
ll
e
g
e

g
ra
d

C
o
ll
e
g
e

g
ra
d

A
ll

im
m
ig
ra
n
ts

A
t
le
a
st

n
in
e
ye
ar
s

in
U
S

A
rr
iv
e
d

b
e
fo
re

a
g
e
2
0

A
rr
iv
e
d
a
ft
e
r

a
g
e
2
0

N
o
t
a

co
ll
e
g
e
g
ra
d

C
o
ll
e
g
e

g
ra
d

(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

(9
)

(1
0
)

(1
1
)

(1
2
)

D
e
p
e
n
d
e
n
t
va
ri
a
b
le
:
n
u
m
b
e
r
o
f
ch

il
d
re
n

M
e
n

W
o
m
e
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

1
.6
4
7
*
*

1
.2
9
8

0
.7
2
2

2
.0
3
5
*
*

1
.1
4
7

1
.9
5
8
*
*

1
.5
6
0
*
*
*

3
.7
7
9
*
*
*

3
.1
8
0
*
*
*

0
.5
3
3

1
.2
0
7

2
.1
5
7
*
*

(0
.6
4
1
)

(0
.8
7
3
)

(0
.6
3
7
)

(0
.8
1
7
)

(0
.9
7
1
)

(0
.8
3
7
)

(0
.5
6
7
)

(1
.0
7
5
)

(0
.9
1
1
)

(0
.8
1
2
)

(0
.7
7
3
)

(0
.8
4
7
)

S
a
m
p
le

si
ze

2
0
,4
8
5

1
1
,7
1
4

9
,9
3
7

1
0
,5
4
8

1
0
,5
8
1

9
,9
0
4

1
7
,1
6
2

9
,1
7
8

7
,8
9
4

9
,2
6
8

9
,3
1
8

7
,8
4
4

M
a
rr
ie
d
m
e
n

M
a
rr
ie
d
w
o
m
e
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

2
.4
1
5
*
*

2
.2
1
6
*
*

1
.3
5
4

2
.8
8
0
*
*

2
.0
9
1

2
.6
4
7
*
*

0
.9
8
3
*

3
.9
0
1
*
*
*

2
.9
4
0
*
*

�0
.3
7
9

0
.8
3
5

1
.5
3
5
*

(0
.9
9
0
)

(1
.0
7
6
)

(1
.0
1
4
)

(1
.3
0
1
)

(1
.9
1
5
)

(1
.0
6
5
)

(0
.4
9
7
)

(1
.1
8
6
)

(1
.2
9
6
)

(0
.8
6
7
)

(0
.8
5
9
)

(0
.8
8
5
)

S
a
m
p
le

si
ze

1
2
,4
5
6

7
,8
1
6

4
,9
8
9

7
,4
6
7

5
,9
7
4

6
,4
8
2

1
3
,4
9
1

7
,0
9
5

5
,1
1
4

8
,3
7
7

7
,4
0
4

6
,0
8
7

D
e
p
e
n
d
e
n
t
va
ri
a
b
le
:
n
u
m
b
e
r
o
f
ch

il
d
re
n
u
n
d
e
r
a
g
e
5

M
e
n

W
o
m
e
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.7
2
2
*

0
.8
8
6

0
.4
9
1

0
.8
1
2
*

0
.3
6
2

0
.9
9
8

1
.0
7
5

1
.6
9
5
*
*

1
.7
1
1
*
*

0
.9
6
5

0
.6
7
5

1
.7
1
4
*
*

(0
.4
1
0
)

(0
.7
3
7
)

(0
.5
7
7
)

(0
.4
6
6
)

(0
.4
6
1
)

(0
.6
6
5
)

(0
.6
8
2
)

(0
.8
0
4
)

(0
.8
2
1
)

(0
.6
2
0
)

(0
.9
8
0
)

(0
.6
5
6
)

S
a
m
p
le

si
ze

2
0
,4
8
5

1
1
,7
1
4

9
,9
3
7

1
0
,5
4
8

1
0
,5
8
1

9
,9
0
4

1
7
,1
6
2

9
,1
7
8

7
,8
9
4

9
,2
6
8

9
,3
1
8

7
,8
4
4

M
a
rr
ie
d
m
e
n

M
a
rr
ie
d
w
o
m
e
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

1
.1
5
0
*

1
.4
3
2

0
.7
8
1

1
.3
5
9
*
*

0
.8
1
5

1
.5
2
7
*

0
.7
4
6

1
.5
9
8
*
*

1
.1
5
3

0
.6
7
3

0
.2
7
8

1
.3
8
4
*
*

(0
.5
7
5
)

(0
.9
7
3
)

(0
.9
3
7
)

(0
.5
9
7
)

(0
.8
8
2
)

(0
.8
9
5
)

(0
.6
9
1
)

(0
.7
9
2
)

(0
.9
8
0
)

(0
.6
6
4
)

(1
.0
7
2
)

(0
.6
7
7
)

S
a
m
p
le

si
ze

1
2
,4
5
6

7
,8
1
6

4
,9
8
9

7
,4
6
7

5
,9
7
4

6
,4
8
2

1
3
,4
9
1

7
,0
9
5

5
,1
1
4

8
,3
7
7

7
,4
0
4

6
,0
8
7

N
ot
es
.
T
h
e
re
p
o
rt
e
d
co

e
ffi
ci
e
n
ts
a
re

fr
o
m

O
L
S
re
g
re
ss
io
n
s.
S
ta
n
d
ar
d
e
rr
o
rs
,
w
h
ic
h
a
re

cl
u
st
e
re
d
b
y
ye
a
r,
a
p
p
e
ar

in
p
a
re
n
th
e
se
s.
S
ig
n
ifi
ca
n
ce

le
ve
ls
a
re

in
d
ic
a
te
d
b
y

*
,
*
*
,
o
r
*
*
*
w
h
ic
h
re
p
re
se
n
t
1
0
%
,
5
%
,
a
n
d
1
%

le
ve
ls
,
re
sp
e
ct
iv
e
ly
.
E
a
ch

re
g
re
ss
io
n
in
cl
u
d
e
s
th
e
se
t
o
f
co

n
tr
o
l
va
ri
a
b
le
s
in

th
e
co

re
sp
e
ci
fi
ca
ti
o
n
d
e
sc
ri
b
e
d
in

co
lu
m
n

(3
)
o
f
T
a
b
le

6
.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2101



findings are robust to the inclusion or exclusion of a broad array of control variables,
and the surge in hate crimes against Muslims did not occur in states which were already
undergoing an increase in fertility.

Our next assimilation outcome is measured by being in the ‘home sector’, which is
defined as a person who is not enroled in school and not participating in the labour
market. Table 12 shows that hate crimes against Muslims increased the probability
that females stayed at home, with no effect on immigrant men from Muslim
countries. The magnitude of the coefficient implies that the mean level of hate
crimes in the wake of 9/11 increased the probability that Muslim women stayed at
home by 7.6% relative to its mean in 2000. Furthermore, although the state-level of
hate crimes against Muslims is positively related to the change in the ‘home sector’
for females in the aftermath of the 9/11 attacks, the state-level surge in anti-Muslim
activity after 2001 was not positively correlated with pre-existing trends in the labour
market behaviour of females (the correlation is actually negative in column 5 of
Table 12). These results are robust to the inclusion of several control variables (see
also Figures 12 and 13 for no controls) and are found across many segments of the
female Muslim population (Table 13). The finding that anti-Muslim hate crimes led

Table 12

Effect of Hate Crimes on Being in the Home Sector for Muslim Immigrants

(1) (2) (3) (4) (5)

Men
Muslim hate crimes PC 0.149 0.131 0.031

(0.168) (0.166) (0.178)
Muslim hate crimes PC, 2001–2 0.120

(0.511)
Lead Muslim hate crimes PC �0.288*

(0.155)
Sample size 20,462 20,462 20,462 20,462 19,714

Women
Muslim hate crimes PC 0.757** 0.987** 1.152***

(0.360) (0.405) (0.372)
Muslim hate crimes PC, 2001–2 3.261***

(0.812)
Lead Muslim hate crimes PC �1.120***

(0.389)

Sample size 17,154 17,154 17,154 17,154 13,657
Years 2000–10 2000–10 2000–10 2000–10 1990–2000
Year and state FE Yes Yes Yes Yes Yes
Other hate crimes No Yes Yes Yes No
Other state-year controls No No Yes Yes No

Notes. The reported coefficients are the marginal effects from a probit, computed at the means of the
independent variables. Standard errors, which are clustered by state, appear in parentheses. Significance
levels are indicated by *, **, or *** which represent 10%, 5%, and 1% levels, respectively. Each regression
includes the set of control variables in the core specification described in column (3) of Table 6, in addition
to those indicated at the bottom of each column. The ‘other state-year’ controls are those added to the
specification in column (5) of Table 6: percentage male among of those of Muslim origin, percentage with
Muslim origin, percentage of females in the ‘home’ sector, male unemployment rate among natives, male
mean wage among natives.

© 2014 Royal Economic Society.

2102 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



T
a
b
le

1
3

B
ei
n
g
in

th
e
H
om

e
S
ec
to
r
fo
r
M
u
sl
im

Im
m
ig
ra
n
ts
by

S
u
bg
ro
u
p
s,
2
0
0
0
–
1
0

A
ll
im

m
ig
ra
n
ts

A
t
le
a
st

n
in
e
ye
ar
s

in
U
S

A
rr
iv
e
d

b
e
fo
re

a
g
e
2
0

A
rr
iv
e
d

a
ft
e
r

a
g
e
2
0

N
o
t
a

co
ll
e
g
e

g
ra
d

C
o
ll
e
g
e

g
ra
d

A
ll

im
m
ig
ra
n
ts

A
t
le
a
st

n
in
e
ye
a
rs

in
U
S

A
rr
iv
e
d

b
e
fo
re

a
g
e
2
0

A
rr
iv
e
d

a
ft
e
r

a
g
e
2
0

N
o
t
a

co
ll
e
g
e

g
ra
d

C
o
ll
e
g
e

g
ra
d

(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

(9
)

(1
0
)

(1
1
)

(1
2
)

M
e
n

W
o
m
e
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.1
3
1

0
.1
4
2

�0
.0
8
4

0
.2
1
9

0
.2
1
4

0
.0
2
7

0
.9
8
7
*
*

0
.3
4
0

1
.0
8
7
*
*

0
.7
0
9

1
.7
0
5
*
*
*

0
.0
4
7

(0
.1
6
6
)

(0
.2
1
1
)

(0
.2
0
6
)

(0
.2
6
7
)

(0
.2
4
1
)

(0
.2
4
3
)

(0
.4
0
5
)

(0
.6
2
6
)

(0
.5
3
5
)

(0
.4
4
7
)

(0
.6
5
5
)

(0
.4
7
0
)

S
a
m
p
le

si
ze

2
0
,4
6
2

1
1
,6
3
5

9
,7
5
2

1
0
,5
1
5

1
0
,5
3
2

9
,7
8
0

1
7
,1
5
4

9
,1
5
1

7
,8
7
3

9
,2
6
5

9
,3
0
9

7
,8
3
6

M
a
rr
ie
d
m
e
n

M
a
rr
ie
d
w
o
m
e
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.3
7
9
*
*

0
.2
7
1
*

0
.0
0
1

0
.5
1
5
*

0
.6
8
4
*
*
*

�0
.0
4
1

0
.6
5
5
*

0
.2
1
5

0
.4
5
7

0
.6
2
4

1
.3
0
7
*
*

�0
.2
2
7

(0
.1
5
9
)

(0
.1
6
1
)

(0
.2
2
3
)

(0
.2
6
5
)

(0
.1
9
3
)

(0
.2
4
7
)

(0
.3
9
7
)

(0
.6
4
6
)

(0
.6
6
4
)

(0
.4
5
1
)

(0
.6
4
6
)

(0
.5
8
8
)

S
a
m
p
le

si
ze

1
2
,3
8
6

7
,6
3
5

4
,8
1
2

7
,3
5
2

5
,8
2
7

6
,4
0
6

1
3
,4
8
4

7
,0
7
5

5
,0
9
8

8
,3
7
5

7
,3
9
5

6
,0
7
2

N
ot
es
.
T
h
e
re
p
o
rt
e
d
co

e
ffi
ci
e
n
ts
a
re

th
e
m
a
rg
in
a
l
e
ff
e
ct
s
fr
o
m

a
p
ro
b
it
,
co

m
p
u
te
d
a
t
th
e
m
e
a
n
s
o
f
th
e
in
d
e
p
e
n
d
e
n
t
va
ri
a
b
le
s.
S
ta
n
d
ar
d
e
rr
o
rs
,
w
h
ic
h
a
re

cl
u
st
e
re
d
b
y

st
a
te
,
a
p
p
e
ar

in
p
a
re
n
th
e
se
s.
S
ig
n
ifi
ca
n
ce

le
ve
ls
a
re

in
d
ic
a
te
d
b
y
*
,
*
*
,
o
r
*
*
*
w
h
ic
h
re
p
re
se
n
t
1
0
%
,
5
%
,
a
n
d
1
%

le
ve
ls
,
re
sp
e
ct
iv
e
ly
.
E
ac
h
re
g
re
ss
io
n
in
cl
u
d
e
s
th
e
se
t

o
f
co

n
tr
o
l
va
ri
a
b
le
s
in

th
e
co

re
sp
e
ci
fi
ca
ti
o
n
d
e
sc
ri
b
e
d
in

co
lu
m
n
(3
)
o
f
T
a
b
le

6
.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2103



T
a
b
le

1
4

E
ff
ec
t
of

H
a
te
C
ri
m
es

on
E
n
gl
is
h
P
ro
fi
ci
en
cy

fo
r
Im

m
ig
ra
n
t
M
u
sl
im
s

(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

(9
)

(1
0
)

M
e
n

W
o
m
e
n

S
p
e
a
k
E
n
g
li
sh

a
t
h
o
m
e

S
p
e
a
k
E
n
g
li
sh

a
t
h
o
m
e

M
u
sl
im

h
a
te

cr
im

e
s
P
C

�0
.5
2
6
*

�0
.4
8
4

�0
.3
5
6

�0
.2
4
4
*

�0
.3
6
1
*
*
*

�0
.4
0
7
*
*
*

(0
.2
9
0
)

(0
.3
0
1
)

(0
.2
3
9
)

(0
.1
2
5
)

(0
.1
3
9
)

(0
.1
4
8
)

M
u
sl
im

h
a
te

cr
im

e
s
P
C
,
2
0
0
1
–
2

�1
.0
1
7
*
*

�0
.8
1
0
*
*
*

(0
.4
8
6
)

(0
.2
7
4
)

L
e
a
d
M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.1
1
1

�0
.1
8
9

(0
.1
7
5
)

(0
.1
4
7
)

S
a
m
p
le

si
ze

2
0
,4
6
1

2
0
,4
6
1

2
0
,4
6
1

2
0
,4
6
1

1
9
,8
5
6

1
7
,1
0
3

1
7
,1
0
3

1
7
,1
0
3

1
7
,1
0
3

1
3
,4
9
2

S
p
e
a
k
E
n
g
li
sh

w
e
ll

S
p
e
a
k
E
n
g
li
sh

w
e
ll

M
u
sl
im

h
a
te

cr
im

e
s
P
C

�0
.7
3
0
*
*

�0
.7
1
3
*

�0
.9
3
3
*
*

�0
.4
0
6

�0
.6
2
2
*
*

�0
.6
5
7
*

(0
.3
5
9
)

(0
.3
8
5
)

(0
.4
0
2
)

(0
.3
0
9
)

(0
.3
1
5
)

(0
.3
7
3
)

M
u
sl
im

h
a
te

cr
im

e
s
P
C
,
2
0
0
1
–
2

�1
.4
2
7
*

�1
.1
7
7

(0
.7
9
0
)

(0
.9
1
0
)

L
e
a
d
M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.1
5
3

�0
.0
9
6

(0
.3
1
7
)

(0
.4
6
5
)

S
a
m
p
le

si
ze

2
0
,4
8
5

2
0
,4
8
5

2
0
,4
8
5

2
0
,4
8
5

1
9
,8
8
3

1
7
,1
5
4

1
7
,1
5
4

1
7
,1
5
4

1
7
,1
5
4

1
3
,6
4
4

Y
e
a
rs

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

1
9
9
0
–
2
0
0
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

2
0
0
0
–
1
0

1
9
9
0
–
2
0
0
0

Y
e
a
r
a
n
d
st
a
te

F
E

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

Y
e
s

O
th
e
r
h
at
e
cr
im

e
s

N
o

Y
e
s

Y
e
s

Y
e
s

N
o

N
o

Y
e
s

Y
e
s

Y
e
s

N
o

O
th
e
r
st
a
te
-y
e
a
r
co

n
tr
o
ls

N
o

N
o

Y
e
s

Y
e
s

N
o

N
o

N
o

Y
e
s

Y
e
s

N
o

N
ot
es
.
T
h
e
re
p
o
rt
e
d
co

e
ffi
ci
e
n
ts
a
re

th
e
m
a
rg
in
a
l
e
ff
e
ct
s
fr
o
m

a
p
ro
b
it
,
co

m
p
u
te
d
a
t
th
e
m
e
a
n
s
o
f
th
e
in
d
e
p
e
n
d
e
n
t
va
ri
a
b
le
s.
S
ta
n
d
ar
d
e
rr
o
rs
,
w
h
ic
h
a
re

cl
u
st
e
re
d
b
y

st
a
te
,
a
p
p
e
ar

in
p
a
re
n
th
e
se
s.
S
ig
n
ifi
ca
n
ce

le
ve
ls
a
re

in
d
ic
a
te
d
b
y
*
,
*
*
,
o
r
*
*
*
w
h
ic
h
re
p
re
se
n
t
1
0
%
,
5
%
,
a
n
d
1
%

le
ve
ls
,
re
sp
e
ct
iv
e
ly
.
E
a
ch

re
g
re
ss
io
n
in
cl
u
d
e
s
th
e
se
t

o
f
co

n
tr
o
l
va
ri
a
b
le
s
in

th
e
co

re
sp
e
ci
fi
ca
ti
o
n
d
e
sc
ri
b
e
d
in

co
lu
m
n
(3
)
o
f
T
a
b
le

6
,
in

a
d
d
it
io
n
to

th
o
se

in
d
ic
a
te
d
a
t
th
e
b
o
tt
o
m

o
f
e
a
ch

co
lu
m
n
.
T
h
e
‘o
th
e
r
st
a
te
-y
e
a
r’

co
n
tr
o
ls
a
re

th
o
se

a
d
d
e
d
to

th
e
sp
e
ci
fi
ca
ti
o
n
in

co
lu
m
n
(5
)
o
f
T
a
b
le

6
:
p
e
rc
e
n
ta
g
e
m
a
le

a
m
o
n
g
o
f
th
o
se

o
f
M
u
sl
im

o
ri
g
in
,
p
e
rc
e
n
ta
g
e
w
it
h
M
u
sl
im

o
ri
g
in
,
p
e
rc
e
n
t

fe
m
a
le
s
in

th
e
‘h
o
m
e
’
se
ct
o
r,
m
a
le

u
n
e
m
p
lo
ym

e
n
t
ra
te

a
m
o
n
g
n
a
ti
ve
s,
m
a
le

m
e
a
n
w
a
g
e
a
m
o
n
g
n
a
ti
ve
s.

© 2014 Royal Economic Society.

2104 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



T
a
b
le

1
5

E
ff
ec
t
of

H
a
te
C
ri
m
es

on
th
e
E
n
gl
is
h
P
ro
fi
ci
en
cy

fo
r
M
u
sl
im

Im
m
ig
ra
n
ts
by

S
u
bg
ro
u
p
s,
2
0
0
0
–
1
0

A
ll

im
m
ig
ra
n
ts

A
t
le
a
st

n
in
e
ye
a
rs

in
U
S

A
rr
iv
e
d

b
e
fo
re

a
g
e
2
0

A
rr
iv
e
d

a
ft
e
r

a
g
e
2
0

N
o
t
a

co
ll
e
g
e

g
ra
d

C
o
ll
e
g
e

g
ra
d

A
ll

im
m
ig
ra
n
ts

A
t
le
a
st

n
in
e
ye
a
rs

in
U
S

A
rr
iv
e
d

b
e
fo
re

a
g
e
2
0

A
rr
iv
e
d

a
ft
e
r

a
g
e
2
0

N
o
t
a

co
ll
e
g
e

g
ra
d

C
o
ll
e
g
e

g
ra
d

(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

(9
)

(1
0
)

(1
1
)

(1
2
)

D
e
p
e
n
d
e
n
t
va
ri
a
b
le
:
S
p
e
a
k
E
n
g
li
sh

a
t
h
o
m
e

M
e
n

W
o
m
e
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

�0
.4
8
4

�0
.6
2
2

�0
.6
9
0
*
*

�0
.3
0
8

�0
.6
6
6
*
*

�0
.2
9
8

�0
.3
6
1
*
*
*

�0
.9
1
0
*
*
*

�0
.6
7
6
*
*

�0
.1
7
1

�0
.3
1
3
*

�0
.3
3
8

(0
.3
0
1
)

(0
.4
7
5
)

(0
.2
8
9
)

(0
.3
6
5
)

(0
.3
2
5
)

(0
.3
9
1
)

(0
.1
3
9
)

(0
.3
3
4
)

(0
.2
9
6
)

(0
.1
5
1
)

(0
.1
6
8
)

(0
.2
4
2
)

S
a
m
p
le

si
ze

2
0
,4
6
1

1
1
,6
7
9

9
,9
1
2

1
0
,5
1
9

1
0
,5
0
1

9
,8
9
0

1
7
,1
0
3

9
,1
3
6

7
,8
6
5

9
,1
7
8

9
,1
5
1

7
,7
6
9

M
a
rr
ie
d
m
e
n

M
a
rr
ie
d
w
o
m
e
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

�0
.5
8
6

�0
.9
8
7

�1
.1
9
7

�0
.1
9
9

�0
.5
5
4

�0
.5
2
6

�0
.4
2
3
*
*
*

�0
.9
0
8
*
*
*

�0
.8
8
9
*
*
*

�0
.1
7
4

�0
.4
3
3
*
*

�0
.4
2
9
*

(0
.4
5
2
)

(0
.6
3
8
)

(0
.7
5
7
)

(0
.3
9
9
)

(0
.5
0
0
)

(0
.5
4
0
)

(0
.1
4
4
)

(0
.2
9
2
)

(0
.2
6
2
)

(0
.1
6
0
)

(0
.1
9
5
)

(0
.2
3
2
)

S
a
m
p
le

si
ze

1
2
,4
2
9

7
,7
8
5

4
,9
6
3

7
,3
3
2

5
,9
2
1

6
,4
5
4

1
3
,4
3
6

7
,0
5
4

5
,0
9
5

8
,2
3
4

7
,2
6
8

5
,9
7
3

D
e
p
e
n
d
e
n
t
va
ri
a
b
le
:
S
p
e
a
k
E
n
g
li
sh

w
e
ll

M
e
n

W
o
m
e
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

�0
.7
1
3
*

�0
.8
8
4
*
*

�0
.9
3
1
*
*
*

�0
.2
1
2

�1
.0
1
8
*
*

�0
.3
5
1

�0
.6
2
2
*
*

�0
.7
7
8
*
*

�1
.3
1
1
*
*
*

�0
.0
4
5

�1
.1
1
8
*
*

�0
.0
9
9

(0
.3
8
5
)

(0
.3
8
8
)

(0
.2
2
1
)

(0
.6
1
8
)

(0
.4
7
3
)

(0
.3
9
4
)

(0
.3
1
5
)

(0
.3
9
5
)

(0
.3
4
9
)

(0
.4
1
0
)

(0
.5
3
8
)

(0
.6
4
5
)

S
a
m
p
le

si
ze

2
0
,4
8
5

1
1
,6
9
6

9
,9
1
4

1
0
,5
3
4

1
0
,5
7
6

9
,8
8
9

1
7
,1
5
4

9
,1
6
5

7
,8
8
4

9
,2
6
3

9
,3
1
2

7
,8
2
2

M
a
rr
ie
d
m
e
n

M
a
rr
ie
d
w
o
m
e
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

�0
.4
5
2

�0
.8
9
4

�0
.8
1
0
*

0
.0
3
0

�1
.1
9
5
*
*

0
.1
2
1

�0
.7
3
0
*
*

�1
.2
8
5
*
*

�1
.8
9
1
*
*
*

�0
.1
9
2

�1
.1
1
5
*
*

�0
.3
0
6

(0
.5
3
8
)

(0
.5
6
4
)

(0
.4
8
2
)

(0
.6
2
5
)

(0
.5
0
3
)

(0
.5
4
6
)

(0
.3
4
7
)

(0
.5
2
6
)

(0
.4
4
4
)

(0
.4
3
7
)

(0
.5
6
3
)

(0
.6
2
6
)

S
a
m
p
le

si
ze

1
2
,4
5
0

7
,7
8
9

4
,9
4
3

7
,4
6
0

5
,9
6
2

6
,4
7
3

1
3
,4
8
5

7
,0
8
5

5
,1
0
4

8
,3
7
2

7
,3
9
7

6
,0
6
2

N
ot
es
.
T
h
e
re
p
o
rt
e
d
co

e
ffi
ci
e
n
ts
a
re

th
e
m
a
rg
in
a
l
e
ff
e
ct
s
fr
o
m

a
p
ro
b
it
,
co

m
p
u
te
d
a
t
th
e
m
e
a
n
s
o
f
th
e
in
d
e
p
e
n
d
e
n
t
va
ri
a
b
le
s.
S
ta
n
d
ar
d
e
rr
o
rs
,
w
h
ic
h
a
re

cl
u
st
e
re
d
b
y

st
a
te
,
a
p
p
e
ar

in
p
a
re
n
th
e
se
s.
S
ig
n
ifi
ca
n
ce

le
ve
ls
a
re

in
d
ic
a
te
d
b
y
*
,
*
*
,
o
r
*
*
*
w
h
ic
h
re
p
re
se
n
t
1
0
%
,
5
%
,
a
n
d
1
%

le
ve
ls
,
re
sp
e
ct
iv
e
ly
.
E
ac
h
re
g
re
ss
io
n
in
cl
u
d
e
s
th
e
se
t

o
f
co

n
tr
o
l
va
ri
a
b
le
s
in

th
e
co

re
sp
e
ci
fi
ca
ti
o
n
d
e
sc
ri
b
e
d
in

co
lu
m
n
(3
)
o
f
T
a
b
le

6
.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2105



women, and not men, to be in the home sector, is consistent with a reinforcement of
traditional gender roles, thus providing additional support for the idea that the
backlash in the aftermath of 9/11 attacks intensified the ethnic identity of Muslim
immigrants.

Our last assimilation outcome is based on the individual’s level of English
proficiency. The two measures of English proficiency are whether the respondent
speaks English at home and whether the individual speaks English ‘well’. An
affirmative answer to either one is indicative of a more assimilated outcome. The
results for both measures are presented in Table 14, while Figures 14–17 present a
similar analysis without any additional controls. Comparing the first four columns
for each group with the fifth column, Table 14 shows that anti-Muslim hate
crimes decreased the use and proficiency of English for both men and women,
and these patterns were not consistent with pre-existing trends before 2001.
These findings are robust across both measures of English proficiency and to the
inclusion of many controls. Table 15 shows that the results are similar across
many segments of the Muslim immigrant community, including those that were
in the US prior to the 9/11 attacks (at least nine years in the US). Overall,
the findings for English proficiency are a bit weaker than previous outcomes in
terms of statistical significance but are consistent with previous findings that the
anti-Muslim backlash in the wake of the 9/11 attacks led to a more insular Muslim
community.

6. The Assimilation Outcomes of Other Immigrant Groups

The analysis in the previous Sections shows that anti-Muslim hate crimes after 9/11
created a more closed, less-assimilated Muslim community of immigrants in the US.
These findings are consistent across a variety of specifications, across men and
women, and for several different assimilation outcomes. Even when we control for
the level of hate crime activity against other often-persecuted groups, the analysis
shows that Muslim immigrants are responding particularly to the hate crimes against
their own group. As a further placebo analysis, this Section examines the assimilation
outcomes of other immigrant groups to see if they are responding to anti-Muslim
hate crimes in a similar way.

Table 16 presents estimates for the effect of anti-Muslim hate crimes on six
assimilation outcomes for the following immigrant groups: Hispanics, Chinese,
Japanese and Koreans. Our previous results for Muslim immigrants are also presented
using the same specification for comparison purposes.

Consistent with a causal interpretation of our previous results, Table 16 shows
that Muslim immigrants were the only group that became systematically less
assimilated in response to the surge in anti-Muslim sentiment after 2001. For men
and women, Muslim immigrants are responding in a significant way for each
assimilation outcome towards a more insular and traditional way of life. For the
other immigrant groups, most of the coefficients are not significant, and the ones
that are significant often point to a more assimilated outcome. The group with the

© 2014 Royal Economic Society.

2106 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



T
a
b
le

1
6

T
h
e
E
ff
ec
t
of

M
u
sl
im

H
a
te
C
ri
m
es

on
th
e
A
ss
im
il
a
ti
on

O
u
tc
om

es
of

V
a
ri
ou
s
Im

m
ig
ra
n
t
G
ro
u
p
s,
2
0
0
0
–
1
0

M
e
n

W
o
m
e
n

(1
)

(2
)

(3
)

(4
)

(5
)

(6
)

(7
)

(8
)

(9
)

(1
0
)

(1
1
)

(1
2
)

In
tr
a
-m

a
rr
ie
d

N
u
m
b
e
r

o
f
ch

il
d
re
n

N
u
m
b
e
r

o
f
ch

il
d
re
n

le
ss

th
a
n

a
g
e
5

H
o
m
e

se
ct
o
r

S
p
e
a
k

E
n
g
li
sh

a
t
h
o
m
e

S
p
e
a
k

E
n
g
li
sh

w
e
ll

In
tr
a
-m

a
rr
ie
d

N
u
m
b
e
r

o
f
ch

il
d
re
n

N
u
m
b
e
r

o
f
ch

il
d
re
n

le
ss

th
a
n

a
g
e
5

H
o
m
e

se
ct
o
r

S
p
e
a
k

E
n
g
li
sh

a
t
h
o
m
e

S
p
e
a
k

E
n
g
li
sh

w
e
ll

M
u
sl
im
s

M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.8
6
7
*
*
*

1
.6
4
7
*
*

0
.7
2
2
*

0
.1
3
1

�0
.4
8
4

�0
.7
1
3
*

1
.3
8
5
*
*
*

1
.5
6
0
*
*
*

1
.0
7
5

0
.9
8
7
*
*

�0
.3
6
1
*
*
*

�0
.6
2
2
*
*

(0
.2
6
4
)

(0
.6
4
1
)

(0
.4
1
0
)

(0
.1
6
6
)

(0
.3
0
1
)

(0
.3
8
5
)

(0
.2
6
3
)

(0
.5
6
7
)

(0
.6
8
2
)

(0
.4
0
5
)

(0
.1
3
9
)

(0
.3
1
5
)

S
a
m
p
le

si
ze

1
9
,9
4
0

2
0
,4
8
5

2
0
,4
8
5

2
0
,4
6
2

2
0
,4
6
1

2
0
,4
8
5

1
6
,7
4
9

1
7
,1
6
2

1
7
,1
6
2

1
7
,1
5
4

1
7
,1
0
3

1
7
,1
5
4

H
is
p
a
n
ic

M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.1
9
2

0
.6
7
9

0
.0
9
7

0
.5
2
2
*
*
*

0
.1
4
3

0
.2
3
8

�0
.0
1
3

0
.1
7
3

0
.0
1
4

0
.2
7
2
*

0
.1
1
1

0
.0
2
0

(0
.1
4
8
)

(0
.4
4
8
)

(0
.2
3
0
)

(0
.1
3
0
)

(0
.1
0
8
)

(0
.2
3
8
)

(0
.1
3
4
)

(0
.2
7
2
)

(0
.1
2
3
)

(0
.1
4
0
)

(0
.0
9
2
)

(0
.2
7
4
)

S
a
m
p
le

si
ze

3
0
8
,4
6
1

3
1
9
,2
3
4

3
1
9
,2
3
4

3
1
9
,2
3
4

3
1
9
,2
3
4

3
1
9
,2
3
4

2
6
0
,8
1
7

2
6
9
,8
6
6

2
6
9
,8
6
6

2
6
9
,8
6
6

2
6
9
,8
6
6

2
6
9
,8
6
6

C
h
in
es
e

M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.2
0
0

0
.9
7
6
*
*

0
.6
0
3
*
*
*

0
.2
8
4

�0
.1
2
4

�0
.0
8
1

�0
.2
7
8

0
.4
8
9

0
.3
7
1

0
.3
8
9
*
*

�0
.1
3
3

0
.1
0
6

(0
.2
7
8
)

(0
.3
9
0
)

(0
.2
2
4
)

(0
.1
8
4
)

(0
.1
2
9
)

(0
.3
4
0
)

(0
.3
1
0
)

(0
.6
5
1
)

(0
.2
4
6
)

(0
.1
8
9
)

(0
.1
6
7
)

(0
.4
4
7
)

S
a
m
p
le

si
ze

2
3
,5
9
9

2
4
,1
9
6

2
4
,1
9
6

2
4
,0
8
8

2
4
,1
6
9

2
4
,1
9
6

2
7
,4
6
8

2
8
,2
1
6

2
8
,2
1
6

2
8
,2
1
6

2
8
,1
3
6

2
8
,2
1
6

Ja
p
a
n
es
e

M
u
sl
im

h
a
te

cr
im

e
s
P
C

0
.6
1
6

0
.0
0
6

�0
.4
9
8

�0
.1
9
4

0
.2
9
7

�0
.6
9
7

�0
.1
9
1

�0
.3
9
2

�0
.9
7
0
*

�0
.5
8
2

�0
.0
6
4

0
.3
6
5

(0
.4
1
0
)

(0
.7
6
3
)

(0
.5
4
7
)

(0
.3
2
3
)

(0
.3
8
3
)

(0
.5
8
6
)

(0
.3
3
6
)

(0
.7
7
8
)

(0
.5
2
0
)

(0
.5
0
4
)

(0
.3
0
4
)

(0
.5
6
7
)

S
a
m
p
le

si
ze

4
,0
3
5

4
,1
4
7

4
,1
4
7

3
,8
6
3

4
,1
2
0

4
,1
2
1

6
,3
9
4

6
,6
7
8

6
,6
7
8

6
,6
7
2

6
,6
7
7

6
,6
7
0

K
or
ea
n

M
u
sl
im

h
a
te

cr
im

e
s
P
C

�0
.6
0
0

0
.4
6
1

�0
.0
4
6

�0
.0
6
9

0
.3
9
6
*
*

0
.9
6
5
*

0
.1
2
7

0
.7
7
9

0
.4
5
3
*

0
.5
0
1
*
*

0
.3
0
1

0
.2
2
5

(0
.3
8
5
)

(0
.5
0
4
)

(0
.3
6
1
)

(0
.2
1
3
)

(0
.1
7
1
)

(0
.5
4
0
)

(0
.3
4
4
)

(0
.4
9
4
)

(0
.2
6
0
)

(0
.2
4
6
)

(0
.2
2
3
)

(0
.4
3
7
)

S
a
m
p
le

si
ze

1
2
,0
4
1

1
2
,2
3
3

1
2
,2
3
3

1
2
,1
0
9

1
2
,2
3
3

1
2
,2
2
2

1
5
,6
0
9

1
5
,9
9
3

1
5
,9
9
3

1
5
,9
5
9

1
5
,9
9
3

1
5
,9
6
8

N
ot
es
.
T
h
e
re
p
o
rt
e
d
co

e
ffi
ci
e
n
ts
fr
o
m

re
g
re
ss
io
n
s
u
si
n
g
th
e
n
u
m
b
e
r
o
f
ch

il
d
re
n
a
re

fr
o
m

O
L
S
re
g
re
ss
io
n
s.
In

th
e
o
th
e
r
co

lu
m
n
s,
th
e
re
p
o
rt
e
d
co

e
ffi
ci
e
n
ts
a
re

th
e

m
a
rg
in
a
l
e
ff
e
ct
s
fr
o
m

a
p
ro
b
it
,
co

m
p
u
te
d

a
t
th
e
m
e
a
n
s
o
f
th
e
in
d
e
p
e
n
d
e
n
t
va
ri
a
b
le
s.

S
ta
n
d
ar
d

e
rr
o
rs
,
w
h
ic
h

a
re

cl
u
st
e
re
d

b
y
st
a
te
,
a
p
p
e
ar

in
p
a
re
n
th
e
se
s.

S
ig
n
ifi
ca
n
ce

le
ve
ls
a
re

in
d
ic
a
te
d
b
y
*
,
*
*
,
o
r
*
*
*
w
h
ic
h
re
p
re
se
n
t
1
0
%
,
5
%
,
a
n
d
1
%

le
ve
ls
,
re
sp
e
ct
iv
e
ly
.
E
a
ch

re
g
re
ss
io
n
in
cl
u
d
e
s
th
e
se
t
o
f
co

n
tr
o
l
va
ri
a
b
le
s
in

th
e

co
re

sp
e
ci
fi
ca
ti
o
n
d
e
sc
ri
b
e
d
in

co
lu
m
n
(4
)
o
f
T
a
b
le

6
.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2107



most significant coefficients is the Koreans but two of the four significant
coefficients are pointing to increased assimilation. While it is a priori conceivable
that the assimilation behaviour of other immigrants could respond to the overall
level of bigotry against their group or even other groups, the consistent pattern
pointing to less assimilation is found only for Muslim immigrants. By showing
that our findings for Muslim immigrants are not part of a general pattern found
for other groups, the results in this Section support a casual interpretation of
our previous findings showing less assimilation by Muslim immigrants in response to
the unexpected surge in anti-Muslim sentiment after the 2001 attacks.

7. Conclusions

Our analysis shows that the 9/11 attacks induced a backlash against the Muslim
community which, in turn, increased the ethnic identity and demographic
strength of the Muslim immigrant community in the US. The results are not due
to pre-existing trends in the assimilation outcomes of Muslim immigrants across
states and are found to be robust to the inclusion or exclusion of a wide array of
personal and state-level characteristics. Notably, we show that Muslim immigrants are
reacting specifically to hate crimes against Muslims, after controlling for the level of
hate crimes against other groups. Further support for a causal interpretation comes
from our findings that other groups (Hispanics, Chinese, Japanese, Koreans, as well
as Muslims who already made their marital decisions before the 9/11 attacks) did
not respond in a similar way to the anti-Muslim backlash. In this manner, our
analysis supports the idea that 9/11 may have a long-term political and socio-
economic impact, by creating a larger and more ethnically cohesive Muslim
community in this generation and also the next.11 Our findings are consistent with
the anecdotal evidence in the current literature which argues that the 9/11 attacks
led to a galvanised and more observant Muslim community.12

Overall, our analysis highlights a new type of ‘backlash’ that has been ignored
in the literature on political conflict. Existing models argue that extremist groups
commit terror attacks with the goal of provoking a backlash in order to radicalise
moderate supporters that reside in the same country as the perpetrators
(Rosendorff and Sandler, 2004, 2010; Siqueira and Sandler, 2006; Bueno de

11 While we show that hate crimes against Muslims lead to an increase in the demographic strength of the
US Muslim community, Adida et al. (2011) show that Muslim demographic strength leads to more
discrimination against Muslims. Hence, the combination of both results suggests the existence of a potential
feedback loop between Muslim demographic strength and discrimination against Muslims that reinforces
itself over time.

12 Although Westerners tend to link a heightened Islamic identification with radicalisation and
violence, there is no consistent and systematic evidence of such a link. For example, whereas Chen
(2006) uncovers a strong relationship between Islamic religious intensity and social violence during the
Indonesian financial crisis of 1997, Clingingsmith et al. (2009) find that participation in the Hajj
among Pakistanis causes an increase on the observance of Islamic practices, together with an increase in
more tolerant attitudes towards fellow Muslims and non-Muslims, and beliefs of harmony and peace
among different religions. Similarly, Manning and Roy (2010) report that second generation Muslim
immigrants to Britain tend to think of themselves as British almost as much as the white UK-born
population.

© 2014 Royal Economic Society.

2108 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



Mesquita and Dickson, 2007; Baliga and Sj€ostr€om, 2012).13 Despite differing
mechanisms and modelling assumptions, most of these models analyse how the
counter-terror response by the targeted country affects the local political support
for the terrorist group and their aims, and the subsequent recruitment of
members from the local population to commit terror attacks in the future.14

The predictions of these models are consistent with the idea that Al Qaeda
provoked the US to invade Afghanistan. Our findings, however, suggest that terror
groups may try to provoke a backlash against their own ethnic or religious group in
the targeted country, in order to halt the assimilation of Muslim adherents into
Western society.

Although our analysis uses data from the US, our results are likely to be highly
relevant for Europe for several reasons. First, the 9/11 attacks in the US produced a
backlash against Muslims in several countries throughout Europe (�Aslund and Rooth,
2005; Hanes and Machin, 2012; Sch€uller, 2012). Second, there were subsequent attacks
by Al Qaeda on major European cities after the 9/11 attacks (Madrid in 2004 and
London in 2005), which may have induced a further backlash. Third, Muslim
immigrants in Europe tend to be less assimilated than other immigrant groups, even
before the 9/11 attacks (Bisin et al., 2008). In general, the assimilation of Muslim
immigrants is a much larger public issue in Europe than in the US, most likely because
of the larger scale of the immigration wave and perhaps due to the lower education
levels of Muslims who migrated to Europe versus the US. Therefore, our findings shed
new light on our understanding of the increasing use of terror attacks on Western
countries, with the concurrent rise in social and political tensions surrounding the
assimilation of Muslim immigrants in Europe and the US.

13 In addition to a strategy of backlash, the related literature mentions a myriad of other goals behind a
terror campaign. See, for example, the analyses of Lapan and Sandler (1993), Kydd and Walter (2002, 2006),
Berman and Laitin (2005, 2008), Bloom (2005), Bueno de Mesquita (2005, 2013), Berrebi and Klor (2006),
Benmelech and Berrebi (2007), Rohner and Frey (2007), Abadie and Gardeazabal (2008), Gould and Klor
(2010), Becker and Rubinstein (2011) and Benmelech et al. (2012).

14 The empirical evidence on the effectiveness of provoking a military response to gain political
support at home is limited. Focusing on the Israeli-Palestinian conflict, Jaeger et al. (2012) show that
Palestinian fatalities cause the short-run radicalisation of the Palestinian population but the effect is
fleeting and disappears within ninety days. Jaeger et al. (2012) also show that more critical events of the
conflict, like the occurrence of the first Palestinian uprising or the signing of the Oslo accords, do have a
long lasting effect on Palestinians political attitudes, an effect that remains substantial even over 20 years
after the event. For other empirical studies on the effects of terrorism and political violence on the
political attitudes of the affected population, see Karol and Miguel (2007), Berrebi and Klor (2008),
Gardeazabal (2010), Gould and Klor (2010), Montalvo (2011), Shayo and Zussman (2011) and Jaeger
et al. (2015).

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2109



T
a
b
le

A
1

M
ea
n
C
h
a
ra
ct
er
is
ti
cs

a
n
d
O
u
tc
om

es
by

C
ou
n
tr
y
of

O
ri
gi
n
,
2
0
0
0
–
1
0

C
o
u
n
tr
y

o
f
a
n
ce
st
ry

N
u
m
b
e
r
o
f

o
b
se
rv
a
ti
o
n
s,

2
0
0
0
–
1
0

In
tr
a
-

m
a
rr
ie
d

In
te
r-

m
a
rr
ie
d

E
ve
r

m
a
rr
ie
d

N
u
m
b
e
r

o
f
ch

il
d
re
n

N
u
m
b
e
r

o
f
ch

il
d
re
n

b
e
lo
w
a
g
e
5

H
o
m
e

se
ct
o
r

m
e
n

H
o
m
e

se
ct
o
r

w
o
m
e
n

S
p
e
a
k

E
n
g
li
sh

a
t
h
o
m
e

S
p
e
a
k

E
n
g
li
sh

w
e
ll

Y
e
a
rs

in
U
S

M
a
le

C
o
ll
e
g
e

g
ra
d

M
u
sl
im

h
a
te

cr
im

e
s

P
C

A
fg
h
a
n

1
,5
1
6

0
.4
7
3

0
.0
6
1

0
.6
3
8

1
.1
3
7

0
.4
2
7

0
.0
9
6

0
.3
5
6

0
.0
5
7

0
.6
2
1

1
3
.8
2
2

0
.4
9
0

0
.2
4
2

0
.0
1
3

A
lg
e
ri
a
n

1
1
7

0
.3
6
5

0
.1
9
1

0
.6
9
2

0
.9
2
3

0
.4
7
9

0
.0
7
9

0
.2
9
6

0
.1
2
8

0
.5
6
4

8
.7
4
4

0
.5
3
8

0
.4
8
7

0
.0
2
7

A
ra
b

1
,6
9
7

0
.4
4
0

0
.1
4
3

0
.7
2
6

1
.2
4
0

0
.5
4
2

0
.0
9
5

0
.5
1
8

0
.0
6
7

0
.6
4
2

1
0
.7
0
2

0
.6
1
7

0
.3
7
7

0
.0
1
6

A
ra
b
ic

2
,4
4
3

0
.4
3
6

0
.1
4
6

0
.7
2
6

1
.3
2
5

0
.5
8
8

0
.1
1
4

0
.5
3
2

0
.0
6
7

0
.5
6
7

1
0
.6
2
0

0
.6
0
2

0
.3
0
0

0
.0
1
9

B
a
n
g
la
d
e
sh
i

2
,6
3
1

0
.5
7
0

0
.0
4
9

0
.7
4
2

0
.9
5
6

0
.3
9
6

0
.0
7
8

0
.4
6
6

0
.0
4
5

0
.4
7
0

8
.7
0
1

0
.5
0
6

0
.4
5
4

0
.0
1
1

B
e
n
g
al
i

4
4
1

0
.4
6
0

0
.1
2
7

0
.7
0
5

0
.9
8
0

0
.3
2
4

0
.0
5
5

0
.4
1
7

0
.0
6
8

0
.5
6
0

1
1
.1
2
9

0
.4
9
4

0
.4
3
5

0
.0
1
5

E
g
yp
ti
a
n

2
,9
0
9

0
.4
5
7

0
.1
4
4

0
.6
9
9

1
.0
0
6

0
.4
7
8

0
.0
8
3

0
.3
9
7

0
.1
1
6

0
.6
7
2

1
0
.7
7
3

0
.5
7
9

0
.6
2
2

0
.0
1
6

In
d
o
n
e
si
a
n

1
,5
3
3

0
.2
7
7

0
.2
4
0

0
.5
9
6

0
.6
6
3

0
.3
0
5

0
.0
5
5

0
.3
0
5

0
.1
4
6

0
.5
7
5

8
.9
5
4

0
.4
1
6

0
.4
9
1

0
.0
1
5

Ir
a
n
ia
n

6
,3
4
9

0
.3
6
3

0
.1
4
0

0
.6
0
6

0
.6
7
1

0
.2
8
6

0
.0
6
7

0
.2
4
1

0
.1
0
7

0
.7
6
1

1
4
.5
0
6

0
.4
9
9

0
.5
7
8

0
.0
1
1

Ir
a
q
i

1
,0
6
5

0
.4
3
3

0
.0
9
8

0
.6
2
8

1
.1
3
9

0
.4
6
2

0
.1
5
8

0
.4
3
7

0
.0
5
9

0
.4
8
1

9
.2
3
2

0
.5
5
3

0
.2
6
6

0
.0
3
0

Jo
rd
a
n
ia
n

1
,0
5
0

0
.5
2
2

0
.1
4
4

0
.7
6
8

1
.3
5
8

0
.6
2
5

0
.0
8
0

0
.5
2
7

0
.0
7
1

0
.7
0
6

1
1
.2
2
1

0
.5
8
1

0
.4
2
8

0
.0
1
5

K
a
sh
m
ir
i

1
9

0
.4
2
1

0
.1
5
8

0
.6
3
2

0
.7
3
7

0
.3
1
6

0
.0
0
0

0
.3
7
5

0
.2
6
3

0
.9
4
7

9
.5
7
9

0
.5
7
9

0
.7
3
7

0
.0
0
2

M
id
d
le

E
a
st
e
rn

9
4
4

0
.4
2
1

0
.1
5
6

0
.7
0
2

1
.0
7
0

0
.4
6
0

0
.0
6
9

0
.3
6
0

0
.1
3
1

0
.7
5
5

1
2
.9
8
9

0
.5
5
0

0
.5
4
4

0
.0
2
1

M
o
ro
cc
a
n

1
,2
7
6

0
.2
6
7

0
.2
7
6

0
.7
0
7

0
.6
7
2

0
.3
7
4

0
.0
7
4

0
.3
2
2

0
.1
1
4

0
.5
9
5

7
.9
3
7

0
.6
4
5

0
.3
5
4

0
.0
1
6

O
th
e
r
A
ra
b

4
7
9

0
.3
4
5

0
.1
6
6

0
.6
3
0

0
.8
1
4

0
.4
4
3

0
.0
8
2

0
.4
3
2

0
.1
1
3

0
.5
8
9

6
.0
2
3

0
.6
6
2

0
.4
7
8

0
.0
3
4

P
a
k
is
ta
n
i

7
,1
7
5

0
.5
2
7

0
.0
8
2

0
.7
2
7

1
.1
3
5

0
.4
7
4

0
.0
6
8

0
.5
1
9

0
.0
6
6

0
.6
8
7

1
0
.6
2
6

0
.5
3
8

0
.4
9
6

0
.0
1
4

P
a
le
st
in
ia
n

1
,2
7
3

0
.5
1
1

0
.1
7
8

0
.7
9
3

1
.6
0
5

0
.6
7
9

0
.0
7
9

0
.5
3
1

0
.0
6
5

0
.6
8
7

1
3
.1
8
2

0
.5
8
6

0
.3
6
7

0
.0
1
5

S
yr
ia
n

1
,0
7
4

0
.4
8
5

0
.1
5
2

0
.7
4
0

1
.2
5
3

0
.5
3
3

0
.0
8
9

0
.5
1
8

0
.0
8
8

0
.6
5
6

1
1
.7
1
0

0
.5
2
5

0
.4
2
4

0
.0
1
4

T
u
rk
is
h

3
,3
4
9

0
.3
7
9

0
.1
7
3

0
.6
5
5

0
.6
6
0

0
.3
1
0

0
.0
7
2

0
.3
1
1

0
.1
0
5

0
.6
3
1

8
.3
0
0

0
.5
6
6

0
.5
6
1

0
.0
1
7

Y
e
m
e
n
i

3
5
7

0
.5
2
9

0
.0
8
1

0
.8
4
0

1
.6
9
5

0
.7
0
0

0
.0
7
0

0
.7
7
1

0
.0
2
5

0
.3
9
8

1
0
.7
9
0

0
.5
9
7

0
.1
5
4

0
.0
3
0

N
ot
e.
T
h
e
sa
m
p
le

a
n
d
d
a
ta

so
u
rc
e
s
u
se
d
fo
r
th
e
se

ca
lc
u
la
ti
o
n
s
a
re

d
e
sc
ri
b
e
d
in

T
a
b
le

1
.

A
p
p
e
n
d
ix

A
.

© 2014 Royal Economic Society.

2110 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



Table A2

Number of Hate Crimes Against Muslims by State after the Year 2000

Number of
Muslims in

2000
Census

Hate
crimes,
2001–8

Hate crimes
in 2001–2

Hate crimes
per capita
(2001–8)

Hate crimes
per capita (2001–2)

Wyoming 9 3 2 0.333 0.222
Minnesota 293 54 23 0.184 0.078
Vermont 14 2 2 0.143 0.143
Nebraska 91 12 4 0.132 0.044
Maine 46 6 1 0.130 0.022
New Hampshire 79 8 4 0.101 0.051
Montana 20 2 0 0.100 0.000
Arizona 599 59 29 0.098 0.048
Michigan 2,014 186 68 0.092 0.034
Idaho 56 5 5 0.089 0.089
Tennessee 385 30 12 0.078 0.031
Utah 137 10 6 0.073 0.044
Delaware 97 7 2 0.072 0.021
Ohio 1,129 78 48 0.069 0.043
Colorado 449 30 18 0.067 0.040
Kentucky 214 14 7 0.065 0.033
Missouri 347 21 15 0.061 0.043
Washington 715 40 21 0.056 0.029
Massachusetts 1,204 64 34 0.053 0.028
Oregon 381 20 9 0.052 0.024
North Dakota 20 1 0 0.050 0.000
Alaska 40 2 2 0.050 0.050
Kansas 209 10 1 0.048 0.005
South Carolina 176 7 2 0.040 0.011
New Jersey 3,291 124 66 0.038 0.020
Nevada 286 10 5 0.035 0.017
West Virginia 94 3 2 0.032 0.021
Wisconsin 254 8 6 0.031 0.024
Oklahoma 244 7 3 0.029 0.012
Indiana 386 11 6 0.028 0.016
Connecticut 495 14 9 0.028 0.018
North Carolina 584 15 9 0.026 0.015
Virginia 2,215 51 27 0.023 0.012
District of Columbia 174 4 1 0.023 0.006
Maryland 1,397 29 18 0.021 0.013
Texas 2,955 60 38 0.020 0.013
New Mexico 128 2 2 0.016 0.016
Rhode Island 202 3 2 0.015 0.010
Pennsylvania 1,155 17 10 0.015 0.009
Arkansas 75 1 0 0.013 0.000
Illinois 2,179 28 12 0.013 0.006
California 13,756 173 87 0.013 0.006
Louisiana 281 3 2 0.011 0.007
Iowa 109 1 0 0.009 0.000
Florida 2,586 23 13 0.009 0.005
New York 7,302 55 0 0.008 0.000
Alabama 169 1 0 0.006 0.000
Georgia 862 4 3 0.005 0.003
South Dakota 37 0 0 0.000 0.000
Mississippi 89 0 0 0.000 0.000
Hawaii 67 0 0 0.000 0.000

Note. Data sources used for these calculations are described in Table 1.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2111



The Hebrew University of Jerusalem, IZA and CEPR
The Hebrew University of Jerusalem and CEPR

Submitted: 18 February 2014
Accepted: 12 September 2014

Additional Supporting Information may be found in the online version of this article:

Data S1.

References
Abadie, A. and Gardeazabal, J. (2008). ‘Terrorism and the world economy’, European Economic Review, vol. 52

(1), pp. 1–27.
Abdo, G. (2006). Mecca and Main Street: Muslim Life in America after 9/11, New York: Oxford University Press.
Acemoglu, D. and Autor, D. (2011). ‘Chapter 12 – skills, tasks and technologies: implications for employment

and earnings’, in (D. Card and O. Ashenfelter, eds.), Handbook of Labor Economics, Volume 4, Part B,
pp. 1043–171, Amsterdam: Elsevier.

Adida, C.L., Laitin, D.D. and Valfort, M.A. (2011). ‘One Muslim is enough! Evidence from a field experiment
in France’, Discussion Paper No. 6122, IZA.

�Aslund, O. and Rooth, D.O. (2005). ‘Shifts in attitudes and labour market discrimination: Swedish
experiences after 9-11’, Journal of Population Economics, vol. 18(4), pp. 603–29.

Bakalian, A.P. and Bozorgmehr, M. (2009). Backlash 9/11: Middle Eastern and Muslim Americans Respond,
Berkeley, CA: University of California Press.

Baliga, S. and Sj€ostr€om, T. (2012). ‘The strategy of manipulating conflict’, American Economic Review, vol. 102
(6), pp. 2897–922.

Barrett, P.M. (2008). American Islam: The Struggle for the Soul of a Religion, New York: Picador.
Beaudry, P., Doms, M. and Lewis, E. (2010). ‘Should the personal computer be considered a technological

revolution? Evidence from US metropolitan areas’, Journal of Political Economy, vol. 118(5), pp. 988–1036.
Becker, G.S. and Rubinstein, Y. (2011). ‘Fear and the response to terrorism: an economic analysis’,

Discussion Paper No. 1079, CEP.
Benmelech, E. and Berrebi, C. (2007). ‘Human capital and the productivity of suicide bombers’, Journal of

Economic Perspectives, vol. 21(3), pp. 223–38.

Table A3

Muslim Hate Crimes by Type of Crime in 2001

Number of reported offenses Percent

Intimidation 257 53.43
Destruction/vandalism 116 24.12
Simple assault 57 11.85
Aggravated assault 21 4.37
Arson 16 3.33
Burglary 4 0.83
All other larceny 4 0.83
Robbery 2 0.42
Weapon law violations 2 0.42
Shoplifting 1 0.21
Motor theft 1 0.21

Total 481 100

Note. Hate crime data come from the FBI Uniform Crime Reporting (UCR) Program.

© 2014 Royal Economic Society.

2112 T H E E C O N O M I C J O U R N A L [ N O V E M B E R



Benmelech, E., Berrebi, C. and Klor, E.F. (2012). ‘Economic conditions and the quality of suicide terrorism’,
Journal of Politics, vol. 74(1), pp. 113–28.

Berman, E. and Laitin, D.D. (2005). ‘Hard targets: theory and evidence on suicide attacks’, Working Paper
No. 11740, NBER.

Berman, E. and Laitin, D.D. (2008). ‘Religion, terrorism and public goods: testing the club model’, Journal of
Public Economics, vol. 92(10–11), pp. 1942–67.

Berrebi, C. and Klor, E.F. (2006). ‘On terrorism and electoral outcomes: theory and evidence from the
Israeli-Palestinian conflict’, Journal of Conflict Resolution, vol. 50(6), pp. 899–925.

Berrebi, C. and Klor, E.F. (2008). ‘Are voters sensitive to terrorism? Direct evidence from the Israeli
electorate’, American Political Science Review, vol. 102(3), pp. 279–301.

Bisin, A. and Verdier, T. (2000). ‘Beyond the melting pot: cultural transmission, marriage and the evolution
of ethnic and religious traits’, Quarterly Journal of Economics, vol. 115(4), pp. 955–88.

Bisin, A., Patacchini, E., Verdier, T. and Zenou, Y. (2008). ‘Are Muslim immigrants different in terms of
cultural integration?’, Journal of the European Economic Association, vol. 6(2–3), pp. 445–56.

Bisin, A., Topa, G. and Verdier, T. (2004). ‘Religious intermarriage and socialization in the United States’,
Journal of Political Economy, vol. 112(3), pp. 615–64.

Bloom, M. (2005). Dying to Kill: The Allure of Suicide Terror, New York: Columbia University Press.
Braakmann, N. (2010). ‘Islamistic terror and the labour market prospects of Arab men in England: does a

country’s direct involvement matter?’, Scottish Journal of Political Economy, vol. 57(4), pp. 430–54.
Bryan, J. (2005). ‘Constructing ‘the true Islam’ in hostile times: the impact of 9/11 on Arab Muslims in Jersey

City’, in (N. Foner, ed.), Wounded City: The Social Impact of 9/11, pp. 133–63, New York: Russell Sage.
Bueno de Mesquita, E. (2005). ‘Conciliation, counterterrorism, and patterns of terrorist violence’,

International Organization, vol. 59(1), pp. 145–76.
Bueno de Mesquita, E. (2013). ‘Rebel tactics’, Journal of Political Economy, vol. 121(2), pp. 323–57.
Bueno de Mesquita, E. and Dickson, E.S. (2007). ‘The propaganda of the deed: terrorism, counterterrorism,

and mobilization’, American Journal of Political Science, vol. 51(2), pp. 364–81.
Casey, T. and Dustmann, C. (2010). ‘Immigrants’ identity, economic outcomes and the transmission of

identity across generations’, ECONOMIC JOURNAL, vol. 120(542), pp. F31–51.
Chen, D.L. (2006). ‘Islamic resurgence and social violence during the Indonesian financial crisis’, in (M.

Gradstein and K.A. Konrad, eds.), Institutions and Norms in Economic Development, pp. 179–200, Cambridge,
MA: The MIT Press.

Clingingsmith, D., Khwaja, A.I. and Kremer, M. (2009). ‘Estimating the impact of the Hajj: religion and
tolerance in Islam’s global gathering’, Quarterly Journal of Economics, vol. 124(3), pp. 1133–70.

Dancygier, R. (2010). Immigration and Conflict in Europe, New York: Cambridge University Press.
D�avila, A. and Mora, M.T. (2005). ‘Changes in the earnings of Arab men in the US between 2000 and 2002’,

Journal of Population Economics, vol. 18(4), pp. 587–601.
Dustmann, C. (1996). ‘The social assimilation of immigrants’, Journal of Population Economics, vol. 9(1), pp.

37–54.
Gardeazabal, J. (2010). ‘Vote shares in Spanish general elections as a fractional response to the economy and

conflict’, Working Paper No. 33, Economics of Security.
Goldwasser, E. (2000). ‘Economic security and Muslim identity: a study of the immigrant community in

Durham, North Carolina’, in (Y.Y. Haddad and J.L. Esposito, eds.), Muslims on the Americanization Path?
pp. 301–15, New York: Oxford University Press.

Gould, E.D. and Klor, E.F. (2010). ‘Does terrorism work?’, Quarterly Journal of Economics, vol. 125(4), pp. 1459–
510.

Haddad, Y.Y. (2007). ‘The post-9/11 hijab as icon’, Sociology of Religion, vol. 68(3), pp. 253–67.
Haddad, Y.Y. and Esposito, J.L. (2000). Muslims on the Americanization Path? New York: Oxford University

Press.
Hanes, E. and Machin, S. (2012). ‘Hate crime in the wake of terror attacks: evidence from 7/7 and 9/11’,

Working Paper, University College London.
Hasan, A.G. (2000). American Muslims: The New Generation, New York: Continuum Books.
Jaeger, D.A., Klor, E.F., Miaari, S.H. and Paserman, M.D. (2012). ‘The struggle for Palestinians hearts and

minds: violence and public opinion in the second Intifada’, Journal of Public Economics, vol. 96(3–4), pp.
354–68.

Jaeger, D.A., Klor, E.F., Miaari, S.H. and Paserman, M.D. (2015). ‘Can militants use violence to win public
support? Evidence from the second Intifada’, Journal of Conflict Resolution, forthcoming.

Karol, D. and Miguel, E. (2007). ‘The electoral cost of war: Iraq casualties and the 2004 U.S. presidential
election’, Journal of Politics, vol. 69(3), pp. 633–48.

Kaushal, N., Kaestner, R. and Reimers, C. (2007). ‘Labor market effects of September 11th on Arab and
Muslim residents of the United States’, Journal of Human Resources, vol. 42(2), pp. 275–308.

Kulczycki, A. and Lobo, A.P. (2002). ‘Patterns, determinants, and implications of intermarriage among
Arabs’, Journal of Marriage and Family, vol. 64(1), pp. 202–10.

© 2014 Royal Economic Society.

2016] T E R R O R A N D A S S I M I L A T I O N O F M U S L I M S I N T H E W E S T 2113



Kydd, A. and Walter, B.F. (2002). ‘Sabotaging the peace: the politics of extremist violence’, International
Organization, vol. 56(2), pp. 263–96.

Kydd, A. and Walter, B.F. (2006). ‘The strategies of terrorism’, International Security, vol. 31(1), pp. 49–80.
Lapan, H.E. and Sandler, T. (1993). ‘Terrorism and signalling’, European Journal of Political Economy, vol. 9(3),

pp. 383–98.
Manning, A. and Roy, S. (2010). ‘Culture clash or culture club? National identity in Britain’, ECONOMIC

JOURNAL, vol. 120(542), pp. F72–100.
Montalvo, J.G. (2011). ‘Voting after the bombings: a natural experiment on the effect of terrorist attacks on

democratic elections’, Review of Economics and Statistics, vol. 93(4), pp. 1146–54.
Murshed, S.M. and Pavan, S. (2009). ‘Identity and Islamic radicalization in Western Europe’, Working Paper

No. 14, Economics of Security.
Peck, C.W. (2004). ‘Muslims in the American public square: shifting political winds & fallout from 9/11,

Afghanistan, and Iraq’, Zogby International, available at https://www.aclu.org/files/fbimappingfoia/
20111110/ACLURM001733.pdf (last accessed: 15 January 2015).

Pew Research Center (2011). ‘Muslim Americans, no signs of growth in alienation or support for extremism’,
August 2011.

Rabby, F. and Rogers, W.M.III (2009). ‘Post 9-11 U.S. Muslim labor market outcomes’, Discussion Paper No.
4411, IZA.

Rohner, D. and Frey, B.S. (2007). ‘Blood and ink! The common-interest-game between terrorists and the
media’, Public Choice, vol. 133(1–2), pp. 129–45.

Rosendorff, B.P. and Sandler, T. (2004). ‘Too much of a good thing? The proactive response dilemma’,
Journal of Conflict Resolution, vol. 48(4), pp. 657–71.

Rosendorff, B.P. and Sandler, T. (2010). ‘Suicide terrorism and the backlash effect’, Defence and Peace
Economics, vol. 21(5), pp. 443–57.

Ruggles, S., Sobek, M., Alexander, T., Fitch, C.A., Goeken, R., Hall, P.K., King, M. and Ronnander, C. (2008).
Integrated Public Use Microdata Series: Version 4.0 [Machine-readable database], Minneapolis, MN: Minnesota
Population Center [producer and distributor].

Schmidt, G. (2004). Islam in Urban America: Sunni Muslims in Chicago, Philadelphia, PA: Temple University
Press.

Sch€uller, S. (2012). ‘The effects of 9/11 on attitudes toward immigration and the moderating role of
education’, Discussion Paper No. 7052, IZA.

Shayo, M. and Zussman, A. (2011). ‘Judicial ingroup bias in the shadow of terrorism’, Quarterly Journal of
Economics, vol. 126(3), pp. 1447–84.

Siqueira, K. and Sandler, T. (2006). ‘Terrorist versus the government: strategic interaction, support, and
sponsorship’, Journal of Conflict Resolution, vol. 50(6), pp. 878–98.

Sniderman, P.M. and Hagendoorn, L. (2007). When Ways of Life Collide: Multiculturalism and its Discontents in
the Netherlands, Princeton, NJ: Princeton University Press.

UCR Hate Crime Report 2001. FBI, Washington, DC. http://www.fbi.gov/about-us/cjis/ucr/hate-crime/
2001 (last accessed: 18 September 2014).

© 2014 Royal Economic Society.

2114 T H E E C O N O M I C J O U R N A L [ N O V E M B E R 2016]