The Effect of Political Violence On Religiosity (Zussman, 2014)

Journal of Economic Behavior & Organization 104 (2014) 64–83
Contents lists available at ScienceDirect
Journal of Economic Behavior & Organization

journal homepage: www.elsevier.com/locate/jebo
The effect of political violence on religiosity夽

Asaf Zussman ∗
Department of Economics, The Hebrew University of Jerusalem, Israel
a r t i c l e i n f o a b s t r a c t
Article history: This paper studies how politically motivated violence associated with the Israeli-Palestinian
Received 18 April 2013 conflict and the wider Arab-Israeli conflict affects religiosity among Jews and Muslims in
Received in revised form 4 September 2013 Israel. In order to explore this relationship I link data from the Israeli Social Surveys to
Accepted 12 October 2013
information on Israeli conflict-related fatalities by date and location of survey interviews.
Available online 1 November 2013
The analysis, which covers the period 2002–2010, yields robust evidence that violence
makes both Jewish and Muslim Israelis self-identify as more religious.
JEL classification: © 2013 Elsevier B.V. All rights reserved.
Z12
Keywords:
Political violence
Religiosity
Israel
1. Introduction
Religion is a key aspect of individual and group identity which has shaped human relations since the dawn of civilization.
A large social science literature demonstrates that religion and religiosity are linked to attitudes, behaviors and outcomes.1
A separate literature studies the effect of the economic and political environment on religion and religiosity. This literature
has tended to focus on long-term forces, such as modernization and economic development, that might lead to a decline in
religiosity and even to the disappearance of religion as a major force in human relations.2
There has been relatively little rigorous research, however, on the causal effect of shorter term fluctuations in the
environment on religiosity (and through it on various outcomes of interest). Studies which credibly estimate such effects
include Gruber and Hungerman (2008) who show that the repeal of United States state laws that prohibit retail activity
on Sunday lead to a fall in religious attendance,3 Clingingsmith et al. (2009) who find that performing the Hajj pilgrimage
to Mecca affects the attitudes of Pakistani pilgrims (e.g. increasing belief in peace and in equality and harmony among
adherents of different religions), Chen (2007, 2010) who finds that the Indonesian financial crisis in the late 1990s increased
religious intensity among Muslims in that country, and Gould and Klor (2012) who find that post-9/11 hate crimes directed
夽 I thank the editor, Daniel Chen, an anonymous referee, Moses Shayo, Noam Zussman and participants of the NBER conference on Economics, Religion,
and Culture for their useful comments and advice. I am grateful to Nurit Dobrin from the Israeli Central Bureau of Statistics for her help with the data and
to Revital Bar for excellent research assistance.
∗ Correspondence to: Department of Economics, The Hebrew University of Jerusalem, Mount Scopus, Jerusalem 91905, Israel. Tel.: +972 25883124.
E-mail address: azussman@mscc.huji.ac.il
1
Classic treatments of this subject are contained in Smith (1776) and Weber (1920). For more recent contributions see, for example, Akerlof and Kranton
(2010), Barro and McCleary (2003), Guiso et al. (2003), Lehrer (2011) and Sen (2006).
2
For a discussion of this literature see Iannaccone (1998).
3
Gruber and Hungerman (2008) go on to show that the laws’ repeal led to an increase in drinking and drug use. Gerber et al. (2008), Cohen-Zada and
Sander (2011), and Lee (2013) use the same source of exogenous shock to study causal effects on, respectively, political participation, happiness, and
educational attainment.
0167-2681/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jebo.2013.10.006
A. Zussman / Journal of Economic Behavior & Organization 104 (2014) 64–83 65
at Muslims had an adverse effect on the assimilation of Muslim immigrants in the United States. This paper contributes
to this literature by analyzing how variation in the intensity of the Israeli-Palestinian conflict and the wider Arab-Israeli
conflict affects religiosity among Jews and Muslims in Israel.
The case of Israel is particularly interesting for several reasons. First, compared to other countries with a similar level of
economic development, in Israel religion plays an important role in shaping culture and politics. Second, for Jews, who make
up around eighty percent of the country’s population, religion and ethnicity practically overlap; for Muslims, who are the
largest religious minority, there is a very tight relationship between religion and ethnicity: virtually all Muslim Israelis are
Arab and more than eighty percent of Arabs are Muslim. Third, the Israeli-Palestinian conflict – and the wider Arab-Israeli
conflict – has a strong religious component. For example, the most extreme Palestinian organizations involved in the conflict,
such as Hamas, have an Islamist agenda. Similarly, among Jewish Israelis religiosity is strongly associated with hard-line
views toward the Palestinians.
In recent decades the Israeli-Palestinian and Arab-Israeli conflicts have been characterized by low-intensity warfare.
In particular, since the turn of the century most Israeli conflict-related fatalities were the result of sporadic attacks by
Palestinians against civilian and military targets inside Israel and in the Occupied Territories (the West Bank and the Gaza
Strip). This period was also characterized by a growing missile threat, as exemplified by the Hizbullah (a Lebanese militia
group) attacks on northern Israel during the Second Lebanon War of 2006. This paper builds on the exogenous nature of
the temporal and geographical variation in the intensity of this form of politically motivated violence to identify the causal
effect of the political environment on religiosity.
The analysis, which covers the period 2002–2010, relies on comprehensive interview-based Social Surveys conducted
by the Israeli Central Bureau of Statistics. The surveys contain, among other things, questions on religion and religiosity.
Using (confidential) information on interview date and location, I link the surveys to detailed data on Israeli fatalities from
politically motivated violence associated with the Israeli-Palestinian and Arab-Israeli conflicts.
I find robust evidence that political violence makes both Jewish and Muslim Israelis identify themselves as more religious.
The main regression specifications show that, all else being equal, a single additional fatality from politically motivated
violence in the vicinity of the survey participant’s locality in the 30 days preceding the interview date lowers the likelihood
that a Jewish survey participant will self-identify as secular by 0.33 percentage points (0.7 percent); the corresponding figure
for Muslims is 0.34 percentage points (3.1 percent).
The rest of the paper is organized as follows. Section 2 describes the main sources of data used in the analysis and highlights
some broad patterns found in them. Section 3 presents the results of the econometric analysis. Section 4 deals with issues
concerning interpretation of the results. Section 5 provides concluding remarks and discusses potential implications of the
paper’s findings.
2. Data
The analysis in this paper relies on two data sources. The first is Israeli Social Surveys, currently available for all years
from 2002 to 2010. The Social Survey is conducted throughout the year with an in-person interview by employees of the
Israeli Central Bureau of Statistics. It covers annually a randomly selected sample of roughly 7000 individuals aged 20 or
older. The sampled individuals vary by year, i.e. this is not a panel. Like similar cross-sectional surveys in other countries
(e.g. the General Social Survey in the United States and the European Social Survey), the Israeli Social Survey collects detailed
data on participants’ socio-demographic and economic characteristics as well as on their attitudes on various issues.
For the purposes of this study the most relevant questions in the survey are those that ask the participant about her or
his religion and degree of religiosity. In terms of religion, participants can choose among the following five options: Jewish,
Muslim, Christian, Druze4 and “other”. If they pick the last option, a follow-up question asks them to either choose “No
religion/Atheist” or to specify a religion other than the four mentioned above.
To describe their degree of religiosity, Jewish survey participants can choose among the following five options: (1)
ultra-orthodox; (2) religious; (3) traditional/religious; (4) traditional/not particularly religious; (5) not religious/secular.
Muslim survey participants can choose among the following four options: (1) very religious; (2) religious; (3) not particularly
religious; (4) not religious. In the analysis that follows I define as secular Jews and Muslims who chose the last option available
to them (5 and 4, respectively).
The second main data source used in the analysis is a self-constructed dataset on politically motivated violence. It contains
information on all Israeli civilian and security forces fatalities from politically motivated violence since January 1, 1997. For
each fatality there is information about the date and location of the relevant fatal incident. The main source of these data
is B’Tselem, the Israeli Information Center for Human Rights in the Occupied Territories. The B’Tselem data are thought to
be accurate and reliable and have been used by other researchers (e.g. Jaeger and Paserman, 2008; Gould and Klor, 2010).
However, the B’Tselem data do not contain information on attacks carried out by non-Palestinians. For this reason, I turned
to information provided by the Israeli Ministry of Defense (for security forces fatalities) and by the Israeli National Insurance
Institute (for civilian fatalities). I also used these sources to cross-check the B’Tselem data.
4
The Druze religion has its roots in the Ismailism school of Shia Islam; the Druze community in Israel is officially recognized as a separate religious
entity.
66 A. Zussman / Journal of Economic Behavior & Organization 104 (2014) 64–83
Table 1
Distribution of religions.
Jewish Muslim Christian Druze Other religions Atheist N
2002 0.83 0.10 0.03 0.02 0.00 0.02 6793

2003 0.84 0.09 0.03 0.02 0.00 0.02 7036
2004 0.84 0.10 0.03 0.01 0.00 0.01 7391
2005 0.81 0.12 0.04 0.02 0.00 0.02 7251
2006 0.81 0.12 0.03 0.02 0.00 0.02 7099
2007 0.80 0.12 0.04 0.02 0.00 0.02 7167
2008 0.80 0.13 0.03 0.01 0.00 0.02 7075
2009 0.81 0.13 0.03 0.01 0.00 0.02 7200
2010 0.79 0.13 0.03 0.02 0.00 0.02 7290
Total 0.82 0.12 0.03 0.02 0.00 0.02 64,302
Sources. Israeli Central Bureau of Statistics: Social Surveys, 2002–2010.

Notes: The table reports the distribution of self-identified religion among survey participants.
Table 2A
Distribution of religiosity – Jews.
Ultra-orthodox Religious Traditional/religious Traditional/not Not religious/secular N

particularly religious
2002 0.05 0.10 0.13 0.29 0.43 5591

2003 0.06 0.08 0.12 0.28 0.46 5909
2004 0.07 0.08 0.12 0.27 0.46 6160
2005 0.06 0.09 0.13 0.25 0.47 5879
2006 0.07 0.09 0.14 0.26 0.45 5730
2007 0.06 0.09 0.14 0.26 0.45 5749
2008 0.07 0.09 0.14 0.27 0.43 5656
2009 0.07 0.11 0.13 0.26 0.43 5801
2010 0.08 0.09 0.13 0.25 0.45 5774
Total 0.07 0.09 0.13 0.26 0.45 52,249

Notes: The table reports the distribution of self-identified degree of religiosity among Jewish survey participants.
To link the survey and fatalities datasets I rely on restricted-use information provided by the Central Bureau of Statistics on
the date and location in which each of the Social Survey interviews was conducted. This information enables me to establish
the intensity of violence around a particular survey location during a particular period preceding the survey interview date.
The analysis presented in this paper excludes the Occupied Territories.5 I do so because the Territories are an outlier in
several respects. First, the Israeli settlers in the Territories are all Jews. Second, the share of seculars among the settlers is
particularly low. Third, the settlers were subject to an exceptionally high level of politically motivated violence during the
period examined here.
2.1. Broad patterns
I now present some broad patterns concerning religion, religiosity and politically motivated violence. Table 1 displays
the distribution of religions by year in the Social Survey data. The distribution is generally quite stable with the possible
exception that there seems to be a slight decrease in the share of Jews and a slight increase in the share of Muslims over
time.
Table 2 displays, for Jews and Muslims separately, the distribution of the self-reported degree of religiosity. The share of
secular survey participants out of the total over the entire 2002–2010 period was much higher for Jews (45 percent) than
for Muslims (11 percent).
Fig. 1 displays the share of seculars in each quarter during this period.6 For both religious groups the share of seculars
seems to exhibit a downward trend. This downward trend in secularism, or rise in religiosity, likely reflects long-term
demographic factors. For example, for both Jews and Muslims the fertility rate is substantially higher among the non-
secular population, a pattern which works to decrease the share of seculars over time. The main goal of this paper is not to
account for long-term trends in religiosity but rather to test whether shorter-term, region-specific, fluctuations in the share
of seculars are related to temporal and geographical variation in the intensity of violence.
To what extent do the secular and the non-secular differ in their religious practices? Table 3 addresses this ques-
tion by reporting, for Jews and Muslims separately, the distribution of prayer practices by the self-reported degree of
5
In Table 6 I include the Territories in the analysis as a robustness check.
6
The figure does not display the share of seculars among Muslims in the first quarter of 2002 because there was a particularly low number of observations
for Muslims in that quarter.
Table 2B
Distribution of religiosity – Muslims.
Very religious Religious Not particularly religious Not religious N
2002 0.11 0.55 0.21 0.13 706

2003 0.13 0.56 0.21 0.09 622
2004 0.10 0.55 0.21 0.14 745
2005 0.10 0.56 0.22 0.12 833
2006 0.06 0.54 0.26 0.14 866
2007 0.05 0.52 0.33 0.09 859
2008 0.07 0.53 0.31 0.09 896
2009 0.10 0.51 0.28 0.11 908
2010 0.10 0.62 0.18 0.10 954
Total 0.09 0.55 0.25 0.11 7389

Notes: The table reports the distribution of self-identified degree of religiosity among Muslim survey participants.
Jews (le axis) Muslims (right axis)

50% 25%
45% 20%
40% 15%
35% 10%
30% 5%
2002 Q1
2002 Q2
2002 Q3
2002 Q4
2003 Q1
2003 Q2
2003 Q3
2003 Q4
2004 Q1
2004 Q2
2004 Q3
2004 Q4
2005 Q1
2005 Q2
2005 Q3
2005 Q4
2006 Q1
2006 Q2
2006 Q3
2006 Q4
2007 Q1
2007 Q2
2007 Q3
2007 Q4
2008 Q1
2008 Q2
2008 Q3
2008 Q4
2009 Q1
2009 Q2
2009 Q3
2009 Q4
2010 Q1
2010 Q2
2010 Q3
2010 Q4
Fig. 1. Share of seculars in Israel, 2002–2010.
Table 3
Religiosity and prayer practices.
Do you pray? N
Always Once in a while Infrequently Never
Jews
Ultra-orthodoxa 1.00 0.00 0.00 0.00 429
Religiousa 1.00 0.00 0.00 0.00 619
Traditional/Religious 0.37 0.39 0.14 0.10 757
Traditional/Not particularly religious 0.11 0.27 0.30 0.32 1482
Not religious/Secular 0.02 0.08 0.15 0.75 2511
Muslims
Very religious 0.94 0.02 0.00 0.03 90
Religious 0.77 0.13 0.03 0.07 471
Not particularly religious 0.48 0.14 0.16 0.21 263
Not religious 0.04 0.09 0.05 0.82 99
Sources. Israeli Central Bureau of Statistics, 2009 Social Survey.

Notes: The table reports, for Jews and Muslims separately, the distribution of prayer practices by self-identified degree of religiosity.
a
“Ultra-orthodox” and “religious” Jewish survey participants were not asked about their prayer practices.
civilian security forces

120
Second Infada Second Lebanon War
100
80
60
40
20
0
2003 Q1
2003 Q2
2004 Q1
2005 Q3
2005 Q4
2006 Q2
2006 Q4
2007 Q1
2008 Q1
2008 Q3
2009 Q1
2010 Q1
2002 Q1
2002 Q2
2002 Q3
2002 Q4
2003 Q3
2003 Q4
2004 Q2
2004 Q3
2004 Q4
2005 Q1
2005 Q2
2006 Q1
2006 Q3
2007 Q2
2007 Q3
2007 Q4
2008 Q2
2008 Q4
2009 Q2
2009 Q3
2009 Q4
2010 Q2
2010 Q3
2010 Q4
Fig. 2. Israeli fatalities from political violence.
religiosity. Data come from the 2009 Social Survey which focused on religion and religiosity. The table establishes that
there is a very tight link between the self-reported degree of religiosity and prayer practices. Especially noteworthy is
the large difference found for both religious groups between the secular and the “not particularly religious” in the share
of those who never pray. This difference is especially large for Muslims; however, as Appendix Table 1 demonstrates, the
secular/non-secular dichotomy among Jews is just as striking when one examines the observance of other key Jewish religious
practices.
Fig. 2 reveals substantial temporal variation in the number of Israeli civilian and security forces fatalities from politically
motivated violence. The intensity of violence was especially high in early 2002, at the height of the Second Intifada (which
erupted in the fall of 2000), and declined sharply later. The Second Lebanon War (July–August 2006) saw another peak in the
number of Israeli fatalities. Overall, from the first quarter of 2002 to the fourth quarter of 2010 there were 638 Israeli fatalities
in Israel proper – roughly 1.0 fatalities per 100,000 population per year on average – 83 percent of them civilian.7 As Table 4
demonstrates, the number of Israeli fatalities was characterized not only by temporal variation but also by geographical
variation. During the period under investigation the number of fatalities within Israel proper was especially high in the
Jerusalem district.
3. The effect of political violence on religiosity
I now turn to an econometric analysis of the effect of politically motivated violence on religiosity. Relying on the temporal
and geographical variation in the number of fatalities to identify the causal effect of violence on religiosity, I estimate the
following model separately for Jews and for Muslims:
Seculari = ˛ + ˇFatalitieslt + l + ıt + Xi + εi , (1)
where Seculari is an indicator variable that takes the value of 1 if survey participant i self-identified as secular and the value
of 0 otherwise. The variable Fatalitieslt is defined in three alternative ways. As a default, Fatalitieslt is defined as the number of
civilian Israeli fatalities from politically motivated violence in the sub-district surrounding the survey participant’s locality
in the 30 days preceding the survey interview date (the number of fatalities was divided by 100 for ease of exposition).8 In
the second definition the absolute number of fatalities is further divided by sub-district population in the relevant year (in
thousands). In the third alternative, Fatalitieslt is defined as an indicator variable taking the value of 1 if the absolute number
of fatalities is positive and the value of 0 otherwise.
7
To put these figures in perspective, the September 11, 2001 attacks in the United States resulted in almost 3000 fatalities, or about 1.1 fatalities per
100,000 population in that year.
8
The Israeli Central Bureau of Statistics divides Israel proper into 6 districts and 15 sub-districts.
Table 4
Fatalities from political violence by district and year.
Year Jerusalem North Haifa Center Tel Aviv South Total
2002 88 22 75 57 17 7 266
(0.95) (0.82) (0.56) (0.96) (0.94) (0.00) (0.81)
2003 61 8 39 14 32 2 156
(0.97) (0.75) (0.95) (0.36) (0.88) (0.50) (0.87)
2004 22 4 0 1 4 33 64
(0.86) (0.00) (–) (1.00) (0.75) (1.00) (0.88)
2005 2 1 8 10 5 3 29
(1.00) (0.00) (1.00) (0.90) (1.00) (1.00) (0.93)
2006 1 45 17 1 11 4 79
(1.00) (0.62) (1.00) (1.00) (1.00) (0.50) (0.76)
2007 0 0 0 0 0 6 6
(–) (–) (–) (–) (–) (1.00) (1.00)
2008 14 0 0 0 0 18 32
(0.86) (–) (–) (–) (–) (0.67) (0.75)
2009 1 1 0 1 0 0 3
(1.00) (1.00) (–) (1.00) (–) (–) (1.00)
2010 2 0 0 0 0 1 3
(1.00) (–) (–) (–) (–) (1.00) (1.00)
Total 191 81 139 84 69 74 638

(0.94) (0.65) (0.75) (0.86) (0.91) (0.78) (0.83)
Sources. See text for details.

Notes: The table reports total (civilian and security forces) Israeli fatalities from politically motivated violence by district and year (share of civilian fatalities
in total is in parentheses).
Additional variables included in Eq. (1) are defined as follows: l is a fixed-effect for the sub-district surrounding the
survey participant’s locality; ıt includes a linear time trend (to capture long-term demographic factors) and sets of indicators
for the month and day of week of the survey interview; Xi is a vector of participant characteristics which includes sets of
indicators for gender, age group, marital status, number of children, highest educational degree, continent of birth, father’s
continent of birth, health status, income, and employment status; and εi is a well-behaved error term. The model is estimated
by OLS (i.e. I use a linear probability model). I focus on the coefficient ˇ which measures the marginal effect of an additional
fatality on the likelihood that the survey participant will self-identify as secular.
The results presented in column 1 of Table 5 indicate that, all else being equal, a single additional civilian fatality from
politically motivated violence in the sub-district surrounding the survey participant’s locality in the 30 days preceding the
interview date lowers the likelihood that a Jewish survey participant will self-identify as secular by 0.33 percentage points.
Since during the period under investigation 45 percent of Jewish survey participants self-identified as secular (Table 2A), this
represents a 0.7 percent decline. Similarly, the results in column 2 indicate that an additional fatality per 1000 population in
the sub-district lowers the likelihood that a Jewish survey participant will self-identify as secular by 0.19 percentage points
(0.4 percent); the results in column 3 indicate that a fatal incident in the sub-district lowers the likelihood that a Jewish
survey participant will self-identify as secular by 1.8 percentage points (4 percent).
As for Muslims, the results presented in column 4 indicate that a single additional civilian fatality from politically moti-
vated violence in the sub-district surrounding the survey participant’s locality in the 30 days preceding the interview date
lowers the likelihood that a Muslim survey participant will self-identify as secular by 0.34 percentage points. Since during
the period under investigation 11 percent of Muslim survey participants self-identified as secular (Table 2B), this represents
a 3.1 percent decline. Similarly, the results in column 5 indicate that an additional fatality per 1000 population in the sub-
district lowers the likelihood that a Muslim survey participant will self-identify as secular by 0.21 percentage points (1.9
percent); the results in column 6 indicate that a fatal incident in the sub-district lowers the likelihood that a Muslim survey
participant will self-identify as secular by 1.0 percentage points (9 percent).
Considering that some regions of the country experienced bouts of high-intensity violence during the period under
investigation – the maximum value of the main explanatory variable in the analysis is 0.3, i.e. 30 civilian fatalities in a sub-
district in a 30 days window – it seems fair to conclude that, for both Jews and Muslims, the effect of violence on religiosity
is quite large.9
Table 3 and Appendix Table 1 demonstrate that in terms of the observance of religious practices, for both Jews and
Muslims the “not particularly religious” have much more in common with the “religious” than with the “secular”. This is
why the analysis is focused on the contrast between the “secular” and the rest. Nevertheless, it is useful to examine the effect
of violence on religiosity in a more general framework. I do so by replacing the Seculari indicator in Eq. (1) with an ordinal
variable “degree of religiosity” which can take five values for Jews and four values for Muslims; in both cases a lower value is
9
The mean and standard deviation of the main explanatory variable in the analysis are, respectively, 0.004 and 0.021.
70
Table 5
The effect of political violence on secularity.
A. Zussman / Journal of Economic Behavior & Organization 104 (2014) 64–83

Dependent variable: secular
Jews Muslims
Absolute number Number adjusted for Indicator for a Absolute number Number adjusted for Indicator for a
sub-district population positive number sub-district population positive number
(1) (2) (3) (4) (5) (6)
Fatalities −0.334*** −0.193*** −0.018** −0.340** −0.206*** −0.010

(0.077) (0.049) (0.007) (0.132) (0.062) (0.012)
Sub-district fixed effects Yes Yes Yes Yes Yes Yes

Time controls Yes Yes Yes Yes Yes Yes
Participant characteristics Yes Yes Yes Yes Yes Yes
R2 0.218 0.218 0.217 0.141 0.141 0.141

Observations 52,249 52,249 52,249 7389 7389 7389
Sources. Israeli Central Bureau of Statistics: Social Surveys, 2002–2010; information on fatalities from politically motivated violence was collected by the author as described in the text.
Notes: “Secular” is an indicator variable that takes the value of 1 if the survey participant self-identified as secular and the value of 0 otherwise. In columns 1 and 4 “fatalities” is the number of civilian Israeli
fatalities from politically motivated violence in the sub-district surrounding the survey participant’s locality in the 30 days preceding the survey interview date. Fatalities figures were divided by 100 for ease
of exposition. In columns 2 and 5 the number of fatalities was further divided by sub-district population (in thousands). In columns 3 and 6 “fatalities” is an indicator for a positive number of fatalities in the
sub-district surrounding the survey participant’s locality in the 30 days preceding the survey interview date. “Time controls” include a linear time trend and sets of indicators for the month and day of week
of the survey interview date. “Participant characteristics” include sets of indicators for gender, age group, marital status, number of children, highest educational degree, continent of birth, father’s continent of
birth, health status, income, and employment status.
Estimated by OLS. Robust standard errors, clustered by sub-district, in parentheses.
*
Represent statistical significance at the 10 percent level.
**
***
associated with a greater degree of self-identified religiosity. The regression is estimated using Ordered Probit. The results,
presented in Appendix Table 2, are qualitatively identical to those presented in Table 5.
Table 6A (Jews) and Table 6B (Muslims) test the robustness of the baseline results to several changes in the speci-
fication of Eq. (1). To facilitate comparison, column 1 in each table replicates the results of the baseline specifications
(Table 5).
Thus far the analysis has included a linear time trend to account for the possible effect of demographic factors on religiosity.
In column 2 of Tables 6A and 6B I allow for a more flexible approach by replacing this variable with a set of sub-district-specific
linear time trends. This has almost no effect on the results.
The reaction of Israelis to security forces fatalities may differ from their reaction to civilian fatalities – e.g. because the
latter are perceived as an illegitimate target of politically motivated violence. To examine this issue, in column 3 of both
tables violence intensity is measured with total (civilian and security forces) fatalities instead of civilian fatalities only. This
change has only a modest effect on the results for both Jews and Muslims.
In column 4 of both tables I include in the analysis survey participants residing in the West Bank and the Gaza Strip. The
results indicate that including this set of observations in the regressions has only a small effect on the size and statistical
significance of the coefficient of interest.
As shown in Fig. 2, the number of fatalities reached two peaks in the period examined here: one in the second quarter of
2002 and the other during the Second Lebanon War of July–August 2006. To examine whether these periods have an undue
influence on the results, I exclude them in turn from the analysis. The results (columns 5 and 6) indicate that excluding both
periods from the analysis has only a small effect for Jews and the same is true for the second quarter of 2002 for Muslims;
however, when excluding the Second Lebanon War period, the coefficient of interest becomes much larger in absolute size
for Muslims. This result is sensible: it implies that Israeli Muslims, who are part of the Palestinian people, are influenced
more by violence associated with the Israeli-Palestinian conflict than by other types of violence (in this case associated with
the Lebanese militia Hizbullah).
Does the effect of politically motivated violence on religiosity decrease with distance from the location of the fatal
incident? So far the analysis has focused on the immediate vicinity of the survey participant’s locality – the sub-district. To
explore the role of geographical distance, I augment Eq. (1) with two additional fatalities variables. The first is the number
of fatalities inside the district but outside the sub-district surrounding the survey participant’s locality. The second is the
number of fatalities within Israel proper but outside the district surrounding the survey participant’s locality (in both cases,
as with the original variable, the number of fatalities was divided by 100 for ease of exposition). Results are presented in
Table 7. To facilitate comparison, columns 1 and 3 replicate the baseline results (Table 5).
For Jewish survey participants I find that the effect of conflict intensity on religiosity decays monotonically with distance
(column 2). The marginal negative effect of an additional fatality on the likelihood that the survey participant will self-
identify as secular declines from 0.35 percentage point (for fatalities suffered in own sub-district) to 0.17 percentage points
(for fatalities suffered in own district but outside own sub-district) and to 0.07 percentage points (for fatalities suffered in
Israel proper but outside own district). For Muslims, the marginal effect of violence on religiosity ceases to be significant
for fatalities suffered outside own sub-district (column 4). The results therefore suggest that physical distance mitigates the
effect of violence on religiosity.
So far the analysis has examined the effect of violence intensity on religiosity using a short, 30 days, window. This
naturally raises the question whether the effect of violence is persistent. To answer this question, I augment Eq. (1) with two
additional fatalities variables. The first is the number of fatalities in the sub-district surrounding the survey participant’s
locality in the 180 days preceding the survey interview date, not including the most recent 30 days. The second is the number
of fatalities in the sub-district surrounding the survey participant’s locality in the 360 days preceding the survey interview
date, not including the most recent 180 days (as before, the number of fatalities was divided by 100). Results are presented
in Table 8.
For both Jewish and Muslim survey participants, I find that the effect of violence on religiosity decays strongly with time.
For Jews, the marginal negative effect of an additional fatality on the likelihood that the survey participant will self-identify
as secular declines from 0.33 percentage points for fatalities in the previous 30 days to 0.09 percentage points for fatalities
in days 181–360 prior to the survey interview date (column 2). For Muslims, the marginal negative effect declines from 0.34
to 0.03 percentage points (column 4). The results therefore suggest that temporal distance has a mitigating influence on the
effect of violence on religiosity.
4. Interpretation
Can the paper’s main finding of a relationship between political violence and religiosity be interpreted as causal? Such an
interpretation relies on the assumption that the location and timing of attacks are not associated with pre-existing patterns
of religiosity. To test the validity of this assumption I run a placebo test where religiosity is regressed on future fatalities
rather than on past fatalities. Specifically, I re-estimate Eq. (1) where the variable Fatalitieslt is now defined as the number of
civilian Israeli fatalities from politically motivated violence in the sub-district surrounding the survey participant’s locality
in the 30 days following the survey interview date. The results (Appendix Table 3A) show that future fatalities are not
associated with secularity for Jews but are negatively associated with secularity for Muslims. However, additional analysis
shows that the association for Muslims disappears when excluding from the analysis the period of the Second Lebanon War
72
Table 6A

Additional specifications – Jews.
Baseline Sub-district Total fatalities Including WB and GS Excluding 2nd Excluding Second
specific time trends quarter of 2002 Lebanon War period
(1) (2) (3) (4) (5) (6)
Fatalities −0.334 ***

−0.331 ***
−0.295 ***
−0.297*** −0.281*** −0.324***
(0.077) (0.081) (0.074) (0.081) (0.081) (0.079)

R2 0.218 0.218 0.218 0.221 0.217 0.217

Observations 52,249 52,249 52,249 54,059 50,746 51,295
Notes: “Secular” is an indicator variable that takes the value of 1 if the survey participant self-identified as secular and the value of 0 otherwise. “Fatalities” is the number of Israeli fatalities from politically
motivated violence in the sub-district surrounding the survey participant’s locality in the 30 days preceding the survey interview date; in columns 1–2 and 4–6 the figure refers to civilians only while in column
3 it refers to civilians and members of the security forces. Fatalities figures were divided by 100 for ease of exposition. “Time controls” include a linear time trend (in column 2 – a set of sub-district-specific linear
time-trends) and sets of indicators for the month and day of week of the survey interview date. “Participant characteristics” include sets of indicators for gender, age group, marital status, number of children,
highest educational degree, continent of birth, father’s continent of birth, health status, income, and employment status. In column 4 the analysis includes survey participants residing in the West Bank and the
Gaza Strip. In column 5 the analysis excludes the second quarter of 2002. In column 6 the analysis excludes the period of the Second Lebanon War (July 12, 2006–August 14, 2006).
Estimated by OLS. Standard errors, clustered by sub-district, in parentheses.
*
**
***
Table 6B

Additional specifications – Muslims.
Baseline Sub-district Total fatalities Including WB and GS Excluding 2nd Excluding Second
specific time trends quarter of 2002 Lebanon War period
(1) (2) (3) (4) (5) (6)
Fatalities −0.340 **
−0.321 **
−0.353 ***
−0.349** −0.332** −0.488**
(0.132) (0.123) (0.068) (0.133) (0.127) (0.200)

R2 0.141 0.145 0.141 0.142 0.139 0.141

Observations 7389 7389 7389 7392 7368 7239
Notes: “Secular” is an indicator variable that takes the value of 1 if the survey participant self-identified as secular and the value of 0 otherwise. “Fatalities” is the number of Israeli fatalities from politically
motivated violence in the sub-district surrounding the survey participant’s locality in the 30 days preceding the survey interview date; in columns 1–2 and 4–6 the figure refers to civilians only while in column
3 it refers to civilians and members of the security forces. Fatalities figures were divided by 100 for ease of exposition. “Time controls” include a linear time trend (in column 2 – a set of sub-district-specific linear
time-trends) and sets of indicators for the month and day of week of the survey interview date. “Participant characteristics” include sets of indicators for gender, age group, marital status, number of children,
highest educational degree, continent of birth, father’s continent of birth, health status, income, and employment status. In column 4 the analysis includes survey participants residing in the West Bank and the
Gaza Strip. In column 5 the analysis excludes the second quarter of 2002. In column 6 the analysis excludes the period of the Second Lebanon War (July 12, 2006–August 14, 2006).
*
**
***
73
Table 7
The effect of violence decays with distance.
Jews Muslims
(1) (2) (3) (4)
Fatalities in sub-district −0.334*** −0.351*** −0.340** −0.336**

(0.077) (0.070) (0.132) (0.133)
Fatalities in rest of district −0.171 0.165
(0.164) (0.187)
Fatalities in rest of country −0.070** 0.109
(0.030) (0.064)
Sub-district fixed effects Yes Yes Yes Yes

Time controls Yes Yes Yes Yes
Participant characteristics Yes Yes Yes Yes
R2 0.218 0.218 0.141 0.142

Observations 52,249 52,249 7389 7389
Sources. Israeli Central Bureau of Statistics: Social Surveys, 2002–2010; information on fatalities from politically motivated violence was collected by the
author as described in the text.
Notes: “Secular” is an indicator variable that takes the value of 1 if the survey participant self-identified as secular and the value of 0 otherwise. “Fatalities”
is the number of civilian Israeli fatalities from politically motivated violence in the 30 days preceding the survey interview date in three distinct areas: (1)
inside the sub-district surrounding the survey participant’s locality; (2) inside the district but outside the sub-district surrounding the survey participant’s
locality; (3) inside Israel proper (i.e. excluding the West Bank and the Gaza Strip) but outside the district surrounding the survey participant’s locality.
Fatalities figures were divided by 100 for ease of exposition. “Time controls” include a linear time trend and sets of indicators for the month and day of
week of the survey interview date. “Participant characteristics” include sets of indicators for gender, age group, marital status, number of children, highest
educational degree, continent of birth, father’s continent of birth, health status, income, and employment status.
*
**
***
Table 8
The effect of violence decays with time.
Jews Muslims
(1) (2) (3) (4)
Fatalities in days t−1 to t−30 −0.334*** −0.329*** −0.340** −0.335**

(0.077) (0.074) (0.132) (0.134)
Fatalities in days t−31 to t−180 0.013 −0.021
(0.045) (0.046)
Fatalities in days t−181 to t−360 −0.086* −0.034
(0.042) (0.031)
Sub-district fixed effects Yes Yes Yes Yes

Time controls Yes Yes Yes Yes
Participant characteristics Yes Yes Yes Yes
R2 0.218 0.218 0.141 0.141

Observations 52,249 52,249 7389 7389
Notes: “Secular” is an indicator variable that takes the value of 1 if the survey participant self-identified as secular and the value of 0 otherwise. “Fatalities”
is the number civilian Israeli fatalities from politically motivated violence in the sub-district surrounding the survey participant’s locality in three distinct
periods: (1) days 1–30 prior to the survey interview date; (2) days 31–180 prior to the survey interview date; (3) days 181–360 prior to the survey interview
date. Fatalities figures were divided by 100 for ease of exposition. “Time controls” include a linear time trend and sets of indicators for the month and day of
*
**
***
(Appendix Table 3B).10 The results thus lend support to a causal interpretation of the relationship between violence and
religiosity.
So far the analysis has assumed that the decline in the share of seculars following acts of violence reflects a shift
toward greater religiosity. However, at least in principle, this decline could instead reflect a shift toward atheism. To
verify that this is not the case, I re-estimate Eq. (1) after replacing the variable Seculari with an indicator for atheism.
The sample includes members of all religions. The results show that violence does not affect atheism (Appendix Table
4).
Another possible concern is selective migration: Israelis may migrate in order to keep away from the areas most
affected by violence and this could lead to biased estimates of the coefficient of interest – ˇ in Eq. (1) – if the
migration response depends on religiosity. Such association could arise, for example, if the religious population is
more willing to stay in violent areas (e.g. Jerusalem) for religious reasons. Unfortunately, the Social Surveys do not
contain detailed information about migration patterns. The most relevant question in the Surveys asks participants
how long they have been residing in the locality. Based on the response to this question, I construct an indicator
for survey participants who have been residing in the locality for a year or less (“new in town”) and then regress
it on the Seculari indicator, the Fatalitieslt variable, and the interaction between the two. The results, presented in
Appendix Table 5, show that the secular indeed tend to migrate more than the non-secular (this is only true for
Jews; the association for Muslims in not significant). Importantly, however, local violence does not seem to deter
in-migration of either the secular or the non-secular. This provides some evidence against the selective migration hypoth-
esis.
What are the channels linking exposure to violence and religiosity? Two main potential channels suggest them-
selves. The first is increased sense of identity. Violence increases the salience of religion (and ethnicity) and this leads
both Jews and Muslims to identify themselves more with their group. This interpretation is consistent with theory
and evidence according to which salience of group membership enhances social identification. For example, Shayo
and Zussman (2011) study ingroup bias using data from Israeli small claims courts during 2000–2004, where the
assignment of a case to an Arab or Jewish judge is essentially random. They find evidence for judicial ingroup bias
and demonstrate that the bias is strongly associated with the intensity of politically motivated violence in the vicin-
ity of the court in the period preceding the ruling. Similarly, analyzing a dataset covering the universe of private
owner transactions in the Israeli market for used cars during 1998–2010, Zussman (2013) finds robust evidence that
escalations in politically motivated violence lead to increased segregation between Israeli Arabs and Jews in the market-
place.
One might argue, however, that the effect of violence on religiosity does not reflect changing perceptions of identity but
rather fear of death. By being exposed to the risk of death, an individual might realize that she or he should behave more
religiously in order to buy a place in heaven. Indeed, a large literature studies the relationship between religiosity and fear
of death using individual level survey data; some of this research finds a non-linear cross-sectional association between the
two variables: fear of death is highest for the moderately religious.11 However, the finding in this paper that the effect of
violence on religiosity is as strong – if not stronger – for Muslims than for Jews seems on the face of it inconsistent with
the fear of death explanation. This is because the politically motivated fatal attacks examined in this paper were targeted
almost exclusively at Jews.
In order to analyze this issue more rigorously I rely on the responses to the Social Survey question “do you feel safe
walking alone at night in your neighborhood?” as a (crude) proxy for fear of death. Based on the responses I construct a
variable “safety” which can take four values ranging from “not safe at all” (1) to “very safe” (4). I then re-estimate Equation
(1) after replacing the variable Seculari with the “safety” variable. The results (Appendix Table 6) show that violence leads
to a decline in perceptions of safety among Jews but not among Muslims. This makes perfect sense given that, as mentioned
above, violence is mostly directed at Jews. In the next stage I re-estimate Eq. (1) – with the original Seculari indicator on
the left hand side – while controlling for the response to the “safety” question. The results (Appendix Table 7) are almost
identical to the baseline results (Table 5). This is consistent with increased sense of identity, rather than fear of death, being
the main channel of influence.12
5. Conclusion
A vast social science literature studies religion and religiosity. However, there has been relatively little rigorous research
on the causal effect of short-term fluctuations in the political environment on religiosity. This paper contributes to the
literature by analyzing how variation in the intensity of the Israeli-Palestinian conflict and the wider Arab-Israeli conflict
affects self-identified religiosity among Jews and Muslims in Israel.
10
Recall in addition that for both Jews and Muslims past fatalities do explain religiosity when excluding from the analysis the Second Lebanon War period
(Tables 6A and 6B).
11
See Neimeyer et al. (2004) for a review of this literature.
12
It is possible that for Muslims violence not only increases the salience of religion (and ethnicity) but also raises concerns about retaliation and
discrimination by Jews. This, in turn, may lead to greater identification with own religious (and ethnic) group.
The analysis, which covers the period 2002–2010, relies on comprehensive Social Surveys which contain, among
other things, questions on religion and religiosity. Using (confidential) information on interview date and loca-
tion, I link the surveys to detailed data on Israeli fatalities from politically motivated violence associated with
the Israeli-Palestinian and Arab-Israeli conflicts. The exogenous nature of the temporal and geographical variation
in violence intensity facilitates credible identification of the causal effect of the political environment on religios-
ity.
I find robust evidence that politically motivated violence makes both Jewish and Muslim Israelis identify them-
selves as more religious. The main regression specifications show that, all else being equal, a single additional
fatality from politically motivated violence in the vicinity of the survey participant’s locality in the 30 days pre-
ceding the interview date lowers the likelihood that a Jewish survey participant will self-identify as secular by 0.7
percent; the corresponding figure for Muslims is 3.1 percent. Considering that some regions of the country experi-
enced bouts of high-intensity violence during the period under investigation, these effects – which are consistent with
theory and evidence according to which salience of group membership enhances social identification – seem quite
large.
What are the implications of these findings? There is a lot evidence in the Israeli context that religious Jews
hold less tolerant attitudes toward Arab Israelis and exhibit greater pessimism regarding the possibility of a
peaceful resolution of the Israeli-Palestinian and Arab-Israeli conflicts. Most of this evidence is correlational (e.g.
Zussman, forthcoming), but some of it suggests that the relationship is causal (e.g. Cohen-Zada et al., 2013). The
findings in this paper thus raise the possibility that via its effect on religiosity, political violence may increase
tensions between Israel’s main religious and ethnic groups and heighten hostility between Israel and its neigh-
bors.
More generally, a large literature studies the role of religion in ethnic conflicts, civil wars, and other forms
of politically motivated violence (see, for example, Berman, 2009). This literature typically treats religiosity as
exogenously given. However, as this paper demonstrates, religiosity may be endogenous to the level of violence.
Thus, the paper’s main findings highlight a potential feedback mechanism which could exacerbate violent con-
flict.
Appendix A.
See Appendix Tables 1–8.
Appendix Table 1
Religiosity and observance of Jewish religious practices.
Observance of practice N
Very strict Strict Lax Not observed
Keep Kosher diet Ultra-orthodoxa 1.00 0.00 0.00 0.00 429

Religiousa 1.00 0.00 0.00 0.00 619
Traditional/Not 0.43 0.35 0.14 0.08 1481
a
Hold Kiddush on eve of Ultra-orthodox 1.00 0.00 0.00 0.00 429
Sabbath Religiousa 1.00 0.00 0.00 0.00 619
Traditional/Not 0.56 0.16 0.06 0.22 1483
Fast on Yom Kippur Ultra-orthodoxa 1.00 0.00 0.00 0.00 429

Religiousa 1.00 0.00 0.00 0.00 619
Traditional/Not 0.75 0.15 0.05 0.06 1347
Sources. Israeli Central Bureau of Statistics, 2009 Social Survey.

Notes: The table reports the distribution of strictness in the observance of three key Jewish religious practices by self-identified degree of religiosity.
a
“Ultra-orthodox” and “religious” survey participants were not asked about their observance of these practices.
Appendix Table 2
Effect of political violence on degree of religiosity.

Dependent variable: degree of religiosity
Jews Muslims
(1) (2) (3) (4) (5) (6)
Fatalities −0.872*** −0.476*** −0.064** −2.984*** −1.597*** −0.180

(0.291) (0.146) (0.028) (1.117) (0.558) (0.124)

Pseudo R2 0.132 0.132 0.132 0.099 0.099 0.098

Observations 52,249 52,249 52,249 7389 7389 7389
Notes: “Degree of religiosity” can take 5 values for Jews and 4 values for Muslims; in both cases a lower value is associated with a greater degree of self-identified religiosity. In columns 1 and 4 “fatalities” is the
number of civilian Israeli fatalities from politically motivated violence in the sub-district surrounding the survey participant’s locality in the 30 days preceding the survey interview date. Fatalities figures were
divided by 100 for ease of exposition. In columns 2 and 5 the number of fatalities was further divided by sub-district population (in thousands). In columns 3 and 6 “fatalities” is an indicator for a positive number
of fatalities in the sub-district surrounding the survey participant’s locality in the 30 days preceding the survey interview date. “Time controls” include a linear time trend and sets of indicators for the month and
day of week of the survey interview date. “Participant characteristics” include sets of indicators for gender, age group, marital status, number of children, highest educational degree, continent of birth, father’s
continent of birth, health status, income, and employment status.
Estimated by Ordered Probit. Robust standard errors, clustered by sub-district, in parentheses.
*
**
***
77
78
Appendix Table 3A
Placebo test – all observations.

Jews Muslims
(1) (2) (3) (4) (5) (6)
Future fatalities −0.116 −0.054 −0.003 −0.491** −0.261*** −0.022

(0.074) (0.051) (0.008) (0.184) (0.060) (0.013)

R2 0.217 0.217 0.217 0.141 0.141 0.141

Observations 52,249 52,249 52,249 7389 7389 7389
Notes: “Secular” is an indicator variable that takes the value of 1 if the survey participant self-identified as secular and the value of 0 otherwise. In columns 1 and 4 “future fatalities” is the number of civilian
Israeli fatalities from politically motivated violence in the sub-district surrounding the survey participant’s locality in the 30 days following the survey interview date. Fatalities figures were divided by 100 for
ease of exposition. In columns 2 and 5 the number of fatalities was further divided by sub-district population (in thousands). In columns 3 and 6 “future fatalities” is an indicator for a positive number of fatalities
in the sub-district surrounding the survey participant’s locality in the 30 days following the survey interview date. “Time controls” include a linear time trend and sets of indicators for the month and day of week
of the survey interview date. “Participant characteristics” include sets of indicators for gender, age group, marital status, number of children, highest educational degree, continent of birth, father’s continent of
birth, health status, income, and employment status.
*
**
***
Appendix Table 3B
Placebo test – excluding Second Lebanon War Period.

Jews Muslims
(1) (2) (3) (4) (5) (6)
Future fatalities −0.109 −0.044 −0.002 −0.248 −0.171 −0.014

(0.076) (0.051) (0.008) (0.317) (0.113) (0.014)

R2 0.218 0.218 0.218 0.142 0.142 0.142

Observations 51,343 51,343 51,343 7304 7304 7304
Notes: The Second Lebanon War started in July 12, 2006 and ended in August 14, 2006. “Secular” is an indicator variable that takes the value of 1 if the survey participant self-identified as secular and the value
of 0 otherwise. In columns 1 and 4 “future fatalities” is the number of civilian Israeli fatalities from politically motivated violence in the sub-district surrounding the survey participant’s locality in the 30 days
following the survey interview date. Fatalities figures were divided by 100 for ease of exposition. In columns 2 and 5 the number of fatalities was further divided by sub-district population (in thousands). In
columns 3 and 6 “future fatalities” is an indicator for a positive number of fatalities in the sub-district surrounding the survey participant’s locality in the 30 days following the survey interview date. “Time
controls” include a linear time trend and sets of indicators for the month and day of week of the survey interview date. “Participant characteristics” include sets of indicators for gender, age group, marital status,
number of children, highest educational degree, continent of birth, father’s continent of birth, health status, income, and employment status.
*
**
***
79
Appendix Table 4
The effect of political violence on atheism.
Dependent variable: Atheist
Absolute number Number adjusted for sub- district population Indicator for a positive number
(1) (2) (3)
Fatalities 0.029 0.010 0.001

(0.022) (0.016) (0.002)
Sub-district fixed effects Yes Yes Yes

Time controls Yes Yes Yes
Participant characteristics Yes Yes Yes
R2 0.062 0.062 0.062

Observations 64,302 64,302 64,302
Notes: “Atheist” is an indicator variable that takes the value of 1 if the survey participant self-identified as atheist and the value of 0 otherwise. In column
1 “fatalities” is the number of civilian Israeli fatalities from politically motivated violence in the sub-district surrounding the survey participant’s locality
in the 30 days preceding the survey interview date. Fatalities figures were divided by 100 for ease of exposition. In column 2 the number of fatalities
was further divided by sub-district population (in thousands). In column 3 “fatalities” is an indicator for a positive number of fatalities in the sub-district
surrounding the survey participant’s locality in the 30 days preceding the survey interview date. “Time controls” include a linear time trend and sets of
indicators for the month and day of week of the survey interview date. “Participant characteristics” include sets of indicators for gender, age group, marital
status, number of children, highest educational degree, continent of birth, father’s continent of birth, health status, income, and employment status.
*
**
***
Appendix Table 5
Selective migration.
Dependent variable: new in town
Jews Muslims
Window length (in days) 30 180 360 30 180 360

(1) (2) (3) (4) (5) (6)
Secular 0.014*** 0.014*** 0.014*** 0.002 0.001 0.002

(0.002) (0.002) (0.002) (0.006) (0.006) (0.006)
Fatalities −0.065 −0.007 −0.002 −0.055 −0.048 −0.016
(0.055) (0.014) (0.008) (0.037) (0.028) (0.018)
Secular × fatalities 0.043 0.006 0.001 0.221 0.047 −0.010
(0.105) (0.034) (0.021) (0.250) (0.104) (0.049)

R2 0.066 0.066 0.066 0.088 0.088 0.088

Observations 52,214 52,214 52,214 7389 7389 7389
Notes: “New in town” is an indicator for survey participants who have been residing in the locality for a year or less. “Secular” is an indicator variable that
takes the value of 1 if the survey participant self-identified as secular and the value of 0 otherwise. “Fatalities” is the number of civilian Israeli fatalities
from politically motivated violence in the sub-district surrounding the survey participant’s locality in the 30/180/360 days preceding the survey interview
date. Fatalities figures were divided by 100 for ease of exposition. “Time controls” include a linear time trend and sets of indicators for the month and day of
*
**
***
Appendix Table 6
The effect of political violence on perceptions of safety.

Dependent variable: safety
Jews Muslims
(1) (2) (3) (4) (5) (6)
Fatalities −0.821** −0.386 −0.065** 0.041 0.194 0.049

(0.408) (0.237) (0.032) (0.816) (0.289) (0.037)

Pseudo R2 0.088 0.088 0.088 0.048 0.048 0.048

Observations 50,837 50,837 50,837 7316 7316 7316
Notes: “Safety” captures the survey participant’s response to the question “do you feel safe walking alone at night in your neighborhood?”; the variable can take four values ranging from 1 = “not safe at all” to
4 = “very safe”. In columns 1 and 4 “fatalities” is the number of civilian Israeli fatalities from politically motivated violence in the sub-district surrounding the survey participant’s locality in the 30 days preceding
the survey interview date. Fatalities figures were divided by 100 for ease of exposition. In columns 2 and 5 the number of fatalities was further divided by sub-district population (in thousands). In columns 3
and 6 “fatalities” is an indicator for a positive number of fatalities in the sub-district surrounding the survey participant’s locality in the 30 days preceding the survey interview date. “Time controls” include a
linear time trend and sets of indicators for the month and day of week of the survey interview date. “Participant characteristics” include sets of indicators for gender, age group, marital status, number of children,
highest educational degree, continent of birth, father’s continent of birth, health status, income, and employment status.
Estimated by Ordered Probit. Robust standard errors, clustered by sub-district, in parentheses.
*
**
***
81
82
Appendix Table 7
The effect of political violence on secularity controlling for perceptions of safety.

Jews Muslims
(1) (2) (3) (4) (5) (6)
Fatalities −0.321*** −0.186*** −0.018** −0.333** −0.197*** −0.009

(0.076) (0.046) (0.008) (0.136) (0.064) (0.011)
“Safety” indicators Yes Yes Yes Yes Yes Yes

R2 0.219 0.220 0.219 0.145 0.145 0.145

Observations 50,837 50,837 50,837 7316 7316 7316
Notes: “Secular” is an indicator variable that takes the value of 1 if the survey participant self-identified as secular and the value of 0 otherwise. In columns 1 and 4 “fatalities” is the number of civilian Israeli fatalities
from politically motivated violence in the sub-district surrounding the survey participant’s locality in the 30 days preceding the survey interview date. Fatalities figures were divided by 100 for ease of exposition.
In columns 2 and 5 the number of fatalities was further divided by sub-district population (in thousands). In columns 3 and 6 “fatalities” is an indicator for a positive number of fatalities in the sub-district
surrounding the survey participant’s locality in the 30 days preceding the survey interview date. “Safety indicators” are dummy variables capturing the survey participant’s response to the question “do you feel
safe walking alone at night in your neighborhood?”. “Time controls” include a linear time trend and sets of indicators for the month and day of week of the survey interview date. “Participant characteristics”
include sets of indicators for gender, age group, marital status, number of children, highest educational degree, continent of birth, father’s continent of birth, health status, income, and employment status.
*
**
***
References
Akerlof, G.A., Kranton, R.E., 2010. Identity Economics: How our Identities Shape our Work, Wages, and Well-Being. Princeton University Press, Princeton.
Barro, R.J., McCleary, R.M., 2003. Religion and economic growth across countries. American Sociological Review 68 (5), 760–781.
Berman, E., 2009. Radical, Religious, and Violent: The New Economics of Terrorism. The MIT Press, Cambridge, MA.
Chen, D.L., 2007. Islamic resurgence and social violence during the indonesian financial crisis. In: Mark, G., Kai, A.K. (Eds.), Institutions and Norms in
Economic Development. MIT Press, Cambridge, MA, pp. 179–200.
Chen, D.L., 2010. Club goods and group identity: evidence from Islamic resurgence during the Indonesian financial crisis. Journal of Political Economy 118
(2), 300–354.
Clingingsmith, D., Khwaja, A.I., Kremer, M., 2009. Estimating the impact of the Hajj: religion and tolerance in Islam’s global gathering. Quarterly Journal of
Economics 124 (3), 1133–1170.
Cohen-Zada, D., Sander, W., 2011. Religious participation versus shopping: what makes people happier? Journal of Law and Economics 54 (4), 889–906.
Cohen-Zada, D., Margalit, Y., Rigbi, O., 2013. Does Religiosity Affect Support for Political Compromise. Ben-Gurion University and Columbia University
(Unpublished manuscript).
Gerber, A., Gruber, J., Hungerman, D.M., 2008. Does Church Attendance Cause People to Vote? Using Blue Laws’ Repeal to Estimate the Effect of Religiosity
on Voter Turnout. NBER Working Paper No. 14303.
Gould, E.D., Klor, E.F., 2010. Does terrorism work? Quarterly Journal of Economics 125 (4), 1459–1510.
Gould, E.D., Esteban, F.K., 2012. The Long-Run Effect of 9/11: Terrorism, Backlash, and the Assimilation of Muslim Immigrants in the West. CEPR Discussion
Paper No. 8797.
Gruber, J., Hungerman, D.M., 2008. The church versus the mall: what happens when religion faces increased secular competition? Quarterly Journal of
Economics 123 (2), 831–862.
Guiso, L., Sapienza, P., Zingales, L., 2003. People’s opium? Religion and economic attitudes. Journal of Monetary Economics 50 (1), 225–282.
Iannaccone, L.R., 1998. Introduction to the economics of religion. Journal of Economic Literature 36 (3), 1465–1495.
Jaeger, D.A., Paserman, D.M., 2008. The cycle of violence? An empirical analysis of fatalities in the Palestinian-Israeli conflict. American Economic Review
98 (4), 1591–1604.
Lee, D.N., 2013. The impact of repealing Sunday closing laws on educational attainment. Journal of Human Resources 48 (2), 286–310.
Lehrer, E.L., 2011. Religion, human capital investments and the family in the United States. In: McCleary R.M. (Ed.), The Oxford Handbook of the Economics
of Religion. Oxford University Press, New York, pp. 39–55.
Neimeyer, R.A., Wittkowski, J., Moser, R.P., 2004. Psychological research on death attitudes: an overview and evaluation. Death Studies 28 (4), 309–340.
Sen, A., 2006. Identity and Violence: The Illusion of Destiny. W.W. Norton, New York.
Shayo, M., Zussman, A., 2011. Judicial ingroup bias in the shadow of terrorism. Quarterly Journal of Economics 126 (3), 1447–1484.
Smith, A., 1776. An Inquiry into the Nature and Causes of the Wealth of Nations. Oxford University Press, Oxford, pp. 1993.
Weber, M., 1920. The Protestant Ethic and the Spirit of Capitalism. Charles Scribner’s Sons, New York, pp. 1958.
Zussman, A., 2013. Ethnic Conflict and Segregation in the Marketplace: Evidence from Israel. Hebrew University (Unpublished manuscript).
Zussman, A., 2013. Ethnic discrimination: lessons from the israeli online market for used cars. Economic Journal (forthcoming).

The Effect of Political Violence On Religiosity (Zussman, 2014)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Effect of Political Violence On Religiosity (Zussman, 2014)

Uploaded by

Copyright:

Available Formats

Journal of Economic Behavior & Organization 104 (2014) 64–83

Contents lists available at ScienceDirect

Journal of Economic Behavior & Organization

The effect of political violence on religiosity夽

Jewish Muslim Christian Druze Other religions Atheist N

2002 0.83 0.10 0.03 0.02 0.00 0.02 6793

Total 0.82 0.12 0.03 0.02 0.00 0.02 64,302

Sources. Israeli Central Bureau of Statistics: Social Surveys, 2002–2010.

Ultra-orthodox Religious Traditional/religious Traditional/not Not religious/secular N

2002 0.05 0.10 0.13 0.29 0.43 5591

Total 0.07 0.09 0.13 0.26 0.45 52,249

Sources. Israeli Central Bureau of Statistics: Social Surveys, 2002–2010.

2.1. Broad patterns

Very religious Religious Not particularly religious Not religious N

2002 0.11 0.55 0.21 0.13 706

Total 0.09 0.55 0.25 0.11 7389

Sources. Israeli Central Bureau of Statistics: Social Surveys, 2002–2010.

Jews (le axis) Muslims (right axis)

Fig. 1. Share of seculars in Israel, 2002–2010.

Always Once in a while Infrequently Never

Sources. Israeli Central Bureau of Statistics, 2009 Social Survey.

civilian security forces

3. The effect of political violence on religiosity

Seculari = ˛ + ˇFatalitieslt + l + ıt + Xi + εi , (1)

Year Jerusalem North Haifa Center Tel Aviv South Total

Total 191 81 139 84 69 74 638

Sources. See text for details.

A. Zussman / Journal of Economic Behavior & Organization 104 (2014) 64–83

Fatalities −0.334*** −0.193*** −0.018** −0.340** −0.206*** −0.010

Sub-district ﬁxed effects Yes Yes Yes Yes Yes Yes

R2 0.218 0.218 0.217 0.141 0.141 0.141

A. Zussman / Journal of Economic Behavior & Organization 104 (2014) 64–83

Dependent variable: secular

Fatalities −0.334 ***

Sub-district ﬁxed effects Yes Yes Yes Yes Yes Yes

R2 0.218 0.218 0.218 0.221 0.217 0.217

A. Zussman / Journal of Economic Behavior & Organization 104 (2014) 64–83

Dependent variable: secular

Sub-district ﬁxed effects Yes Yes Yes Yes Yes Yes

R2 0.141 0.145 0.141 0.142 0.139 0.141

Dependent variable: secular

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

Fatalities in sub-district −0.334*** −0.351*** −0.340** −0.336**

Sub-district ﬁxed effects Yes Yes Yes Yes

R2 0.218 0.218 0.141 0.142

Dependent variable: secular

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

Fatalities in days t−1 to t−30 −0.334*** −0.329*** −0.340** −0.335**

Sub-district ﬁxed effects Yes Yes Yes Yes

R2 0.218 0.218 0.141 0.141

See Appendix Tables 1–8.

Very strict Strict Lax Not observed

Keep Kosher diet Ultra-orthodoxa 1.00 0.00 0.00 0.00 429

Fast on Yom Kippur Ultra-orthodoxa 1.00 0.00 0.00 0.00 429

Sources. Israeli Central Bureau of Statistics, 2009 Social Survey.

A. Zussman / Journal of Economic Behavior & Organization 104 (2014) 64–83

Fatalities −0.872*** −0.476*** −0.064** −2.984*** −1.597*** −0.180

Sub-district ﬁxed effects Yes Yes Yes Yes Yes Yes

Pseudo R2 0.132 0.132 0.132 0.099 0.099 0.098

A. Zussman / Journal of Economic Behavior & Organization 104 (2014) 64–83

Future fatalities −0.116 −0.054 −0.003 −0.491** −0.261*** −0.022

Sub-district ﬁxed effects Yes Yes Yes Yes Yes Yes

R2 0.217 0.217 0.217 0.141 0.141 0.141

Fatalities −0.334* −0.193* −0.018 −0.340 −0.206*** −0.010

Fatalities in sub-district −0.334* −0.351* −0.340 −0.336

Fatalities in days t−1 to t−30 −0.334* −0.329* −0.340 −0.335

Fatalities −0.872* −0.476* −0.064 −2.984* −1.597*** −0.180

Future fatalities −0.116 −0.054 −0.003 −0.491 −0.261* −0.022

Secular 0.014* 0.014* 0.014*** 0.002 0.001 0.002

Fatalities −0.821 −0.386 −0.065 0.041 0.194 0.049

Fatalities −0.321* −0.186* −0.018 −0.333 −0.197*** −0.009