Professional Documents
Culture Documents
Article views: 88
1
Scholars use a variety of terms to refer to pornography, including
sexually explicit media/materials or erotica (Carroll et al., 2008;
Lindgren, 1993; Manning, 2006; Short, Black, Smith, Wetterneck, &
Wells, 2012). Because a number of widely used national data sets use the
term pornography (e.g., General Social Surveys, National Study of Youth
and Religion, Portraits of American Life Study), I also use the term here.
Throughout the study, pornography or porn will refer to any sexually
explicit media (videos, Web sites, magazines, etc.) intended to arouse the
viewer.
Data and coding specications for replication are available from the
author upon request. The author wishes to thank the editor and anonymous
reviewers for their helpful feedback. All errors are my own.
Correspondence should be addressed to Samuel L. Perry, Department of
Sociology, University of Oklahoma, 780 Van Vleet Oval, Kaufman Hall,
Norman, OK 73019. E-mail: samperry@ou.edu
Color versions of one or more of the gures in the article can be found
online at www.tandfonline.com/HJSR.
PERRY
Method
Data
Data for this study were drawn from two waves of the
Portraits of American Life Study (PALS), which was elded
in 2006 and 2012 (Emerson & Sikkink, 20062012). PALS is
a nationally representative panel survey with questions focusing on a variety of topics, including social networks, moral and
political attitudes, and religious life. The original PALS sampling frame includes the civilian, noninstitutionalized household population in the continental United States who were
18 years of age or older at the time the survey was conducted.
Cluster sampling was used to achieve the goal of racially
diverse oversamples. Surveys were administered in English
or Spanish. For Wave 1, face-to-face interviews were conducted with 2,610 respondents in their homes, from April to
October 2006. The response rate was 58%. Interviewers used
audio computer-assisted self-interviewing (ACASI) for more
sensitive questions (e.g., how often they view pornography).
The second wave was conducted from March to September
2012, with 1,314 respondents successfully reinterviewed.
After accounting for respondents from 2006 who died or
were mentally incapacitated, the Wave 2 response rate was
53%. The second wave was administered through self-administered Web survey, computer-assisted telephone interviewing, and face-to-face interviewing. To assess whether attrition
or the different methods of survey administration in Wave 2
may have introduced bias into the sample, Table 1 provides a
comparison of Wave 1 to Wave 2 (after attrition) for all variables in the analysis. The results are virtually identical across
all variables, suggesting the attrition of respondents between
Wave 1 and 2 did not follow a clear pattern that would imply
3
PERRY
Variables
Outcome variables
Religious doubts b
Religious salience b
Religious service attendance b
Prayer b
Focal predictor variables
Pornography viewing at all a
Pornography viewing frequency
Religious doubts a
Religious salience a
Religious service attendance a
Prayer a
Religion variables
Religious tradition b
Conservative Protestant (ref.)
Mainline Protestant
Other Protestant
Catholic
Jewish
Other religion
Unafliated
Scripture fully inspired b
Sociodemographic variables
Age b
Male b
Married b
Number of children b
Bachelors or better b
Household income b
Race b
White (ref.)
Black
Hispanic
Asian
Native American
South b
Range
Full Sample
Full Sample
Men
Women
2006
20062012
20062012
20062012
M/%
SD
15
15
18
17
01
18
15
15
18
19
37%
2.0
1.7
3.3
3.6
4.8
01
01
01
01
01
01
01
01
33%
14%
4%
27%
2%
6%
16%
58%
1880
01
01
09
01
119
45%
48%
57%
NA
26%
9.2
01
01
01
01
01
01
70%
11%
13%
5%
1%
34%
1.7
.9
1.2
2.2
2.7
M/%
SD
M/%
SD
M/%
SD
1.7
3.3
3.7
3.9
0.9
1.3
2.3
2.0
1.8
3.1***
3.6**
3.6***
0.8
1.4
2.4
2.0
1.7
3.5
3.7
4.3
0.9
1.3
2.3
1.9
39%
1.8
1.7
3.4
3.8
4.9
1.6
0.9
1.2
1.9
2.6
59%***
2.7***
1.8
3.2***
3.7*
4.4***
2.1
0.9
1.3
2.3
2.7
20%
1.3
1.7
3.5
3.8
5.5
0.9
0.9
1.3
2.3
2.5
31%
10%
10%
24%
2%
5%
18%
70%
16
4.6
51
48%
58%
2.2
32%
9.4
69%
12%
14%
4%
1%
33%
29%*
10%
10%
24%
1%
6%
20%*
70%***
16
1.9
4.8
51
65%***
2.2
37%*
10.2***
71%
9%*
14%
5%
1%
34%
34%
9%
9%
25%
2%
4%
17%
66%
16
1.9
4.8
50
52%
2.1
31%
8.7
16
1.9
4.8
69%
14%
13%
3%
1%
33%
a
2006 PALS; b2012 PALS.
*Difference between men and women signicant at .05 level.
**Difference between men and women signicant at. 01 level.
***Difference between men and women signicant at. 001 level.
Pitcher, 1986). Thus, to assess the potential effects of pornography use on religiosity over time, outcomes for this study
included four measures of religiosity that tap these three
domains: consistency of belief in ones faith, or conversely,
the prevalence of religious doubt (belief); religious salience
(commitment); and corporate and private religious practice
(behavior). To measure religious doubts, respondents were
asked: In the past 6 years, how often have you had doubts
about your religious faith? Responses ranged from 1 = Never
Outcome Variables
3
In 2012, respondents were asked about religious doubts in the past
6 years (i.e., between the 2012 wave and the 2006 wave), while in 2006
respondents were asked about religious doubts in the past 3 years. While
the wording of the question is changed slightly, the coding of responses is
the same.
Control Variables
Sociodemographic controls are all from Wave 2 of PALS.
These controls were selected based on previous studies of
religiosity over time (Froese & Bader, 2007; Hill, 2011;
Regnerus & Burdette, 2006; Schleifer & Chaves, 2014;
Snell, 2009; Vaidyanathan, 2011) and because of their correlation to both porn viewing and religiosity. Age is measured in years, from 19 to 80. Dummy variables are
constructed for gender (male = 1), education (bachelors
degree or higher = 1), and region (South = 1), and a series
of dummy variables were used for race (White = reference).
Household income is measured in categories from (1) less
than $5,000 to (19) $200,000 or more. Being married and
having children have both long been associated with religious behavior (Schleifer & Chaves, 2014), and thus I
included a dummy variable for whether a respondent was
married (married = 1) and a continuous measure of how
many children a respondent had (range 09).
Pornography attitudes and consumption, as well as
beliefs and behaviors regarding human sexuality more generally, differ across religiocultural milieus, and conservative
Protestants are known to have the strongest anti-pornography attitudes (Sherkat & Ellison, 1997). Moreover, it has
been shown that conservative Protestants tend to practice
their religion more faithfully than, say, mainline Protestants
or Catholics (Putnam & Campbell, 2010). And thus religiocultural conservatism may confound the relationship
between porn use and religiosity measures. Thus, I also
included additional religion controls for religious tradition
and theological conservatism, both of which are associated
with porn use (Perry, 2016) and religiosity (Froese & Bader,
2007; Schleifer & Chaves, 2014).4 Religious tradition is
measured with a modied version of the RELTRAD classication scheme (Steensland et al., 2000). Categories include
conservative Protestant, mainline Protestant, other
Protestant, Catholic, Jewish, other, and None or
unafliated.5 Conservative Protestants are the reference
category. Theological conservatism was measured with a
PALS question asking respondents about whether they
believed their religious text to be fully inspired by God.
Responses included 1 = Fully inspired by God, 2 = Partly
4
Because of the obvious correlation between these controls for religious conservatism and the religiosity outcomes, models were run with and
without the religious conservatism measures. In each model, the effects of
earlier porn use on every religiosity outcome were stronger without these
controls in place, and thus the inclusion of these controls actually made the
results presented in the analysis here more conservative.
5
I include those who indicated none or unafliated with respect to
religious tradition because these persons often hold religious beliefs and
practice a religion, though they do not necessarily identify as such (Putnam
& Campbell, 2010). For example, in Wave 2 of PALS, over 60% of
unafliated persons considered God or spirituality at least somewhat
important, over 50% prayed at least occasionally, and about one-third
attended worship services. Moreover, supplementary analyses without
these unafliated respondents included in models made the effects of
pornography on religiosity outcomes stronger, and thus I present the more
conservative estimates with them included.
PERRY
Analytic Strategy
Table 1 presents basic descriptive statistics for all variables in the analysis. This provides a general overview of
pornography viewing, focal predictors, and controls for the
full sample and by gender. Table 2 presents zero-order
correlations to establish important bivariate relationships
between pornography viewing and religiosity measures at
both Wave 1 and Wave 2. Multivariate models in Tables 3
through 5 consist of OLS LDV models estimating the
effect of porn viewing at Wave 1 on later religiosity measures at Wave 2, while controlling for previous scores on
these religiosity measures at Wave 1 (lagged dependent
variable) and other relevant correlates. These models
have been frequently used in previous studies of change
in religiosity measures over time (e.g., Hill, 2011;
Regnerus & Burdette, 2006; Snell, 2009; Vaidyanathan,
2011). While change-score models (e.g., xed effects) are
helpful for handling the problem of omitted variable bias
caused by time-invariant omitted variables (Schleifer &
Chaves, 2014), estimating these models would not be
possible since the question about pornography was not
asked in 2012. Thus I could not test for whether a change
in respondents frequency of porn consumption between
2006 and 2012 inuenced a change in religiosity measures
during the same time frame. While some scholars have
recommended change-score models over LDV models
(e.g., Johnson, 2005), other scholars (e.g., Keele & Kelly,
2006; Wilkins, 2014) have shown with Monte Carlo simulations that LDV models with OLS generally produce
accurate estimates with no more bias introduced than alternative estimation strategies. And studies using both LDV
and xed effects modeling strategies report similar outcomes for each (e.g., Audia & Teckchandani, 2010).
Results
Descriptive results in Table 1 demonstrate that about 39%
of American adults in Wave 1 viewed some sort of pornographic materials in the previous 12 months. Consistent
with previous research, almost three times as many men
(59%) viewed pornography as women (20%). Comparing
1
2
3
4
5
6
7
8
9
10
a
Predictors
10
1.00
.74***
.15***
.27***
.24***
.29***
.15***
.32***
.24***
.31***
1.00
.16***
.28***
.25***
.32***
.16***
.31***
.24***
.27***
1.00
.21***
.08**
.15***
.40***
.21***
.12***
.18***
1.00
.58***
.64***
.12***
.68***
.52***
.60***
1.00
.52***
.06*
.52***
.74***
.51***
1.00
.09**
.61***
.45***
.65***
1.00
.23***
.12***
.17***
1.00
.56***
.67***
1.00
.61***
1.00
Predictors
Pornography
viewing at all a
Religious
doubts
Religious
Service
Attendanceb
Table 3. (Continued)
Religious
Doubts b
Religious
Salience b
.126*
.202**
.149
.245**
Native
American
(.064)
(.106)
(.094)
South
(.061)
.344***
Predictors
Prayerb
Constant
(.029)
Religious
salience
Adjusted R2
N
.547***
a
Religious
Doubts b
Religious
Salience b
.660
1.101*
(.486)
.028
(.055)
1.360***
(.155)
.196
960
(.502)
.044
(.057)
1.162***
(.168)
.561
1160
Religious
Service
Attendanceb
.217
(.712)
.086
(.096)
.284
(.248)
.605
1161
Prayerb
1.014
(.628)
.166*
(.084)
1.613***
(.230)
.555
1156
(.026)
Religious service
attendance
.653***
(.022)
Prayer
.374***
(.017)
Conservative
Protestant b
(ref.)
Mainline
Protestant b
Other Protestant b
b
Catholic
Jewish
Other religion
Unafliated
Scripture fully
inspired b
Age
Male
Married
Number of
children b
Bachelors or
better b
Household
income b
White b (ref.)
Black b
Hispanic
Asian
.051
.018
.379*
(.093)
.041
(.091)
.038
(.072)
.107
(.221)
.080
(.133)
.132
(.105)
.192**
(.101)
.234*
(.099)
.171*
(.079)
.580**
(.213)
.075
(.136)
.483***
(.091)
.605***
(.061)
.003
(.002)
.031
(.057)
.108+
(.063)
.020
(.065)
.005*
(.002)
.099+
(.058)
.110+
(.064)
.000
(.107)
.010**
(.003)
.052
(.097)
.198+
(.108)
.009
(.095)
.005+
(.003)
.047
(.087)
.192*
(.096)
.029
(.016)
.063
(.017)
.067
(.028)
.124
(.025)
.123
(.062)
.006
(.063)
.014*
(.106)
.005
(.093)
.023*
(.007)
(.007)
(.011)
(.010)
.139+
(.082)
.094
(.083)
.287*
(.130)
.344***
(.089)
.212*
(.087)
.058
(.135)
.544***
(.149)
.393**
(.145)
.580**
(.226)
.454***
(.132)
.202
(.128)
.078
(.211)
(.169)
.106
(.165)
.219+
(.132)
1.041**
(.353)
.484*
(.228)
.852***
(.153)
.554***
.172
(.150)
.060
(.146)
.169
(.117)
.677*
(.314)
.048
(.202)
.655***
(.136)
.779***
(Continued )
PERRY
Predictors
Pornography viewing frequency a
a
Religious doubts
Religious salience
Jewish
Other religion
Unafliated
Male
Married
Number of children
Bachelors or better
Household income
White b (ref.)
Black b
Hispanic
Asian
Native American
South
Catholic
Constant
Adjusted R2
N
.058***
(.018)
.554***
(.026)
Conservative Protestant
Mainline Protestant b
Other Protestant
.049**
(.018)
.342***
(.029)
(ref.)
.038
(.093)
.039
(.091)
.042
(.072)
.097
(.220)
.079
(.132)
.135
(.104)
.194***
(.061)
.003
(.002)
.018
(.057)
.104+
(.063)
.019
(.016)
.065
(.062)
.005
(.007)
.038
(.102)
.237*
(.099)
.181*
(.079)
.573**
(.213)
.081
(.136)
.480***
(.091)
.611***
(.064)
.005*
(.002)
.097+
(.058)
.101
(.064)
.001
(.017)
.066
(.063)
.014*
(.007)
.151+
(.082)
.107
(.083)
.282*
(.130)
.630
(.485)
.031
(.055)
1.328***
(.156)
.198
960
.355***
(.089)
.229**
(.089)
.060
(.135)
1.050*
(.502)
.041
(.057)
1.192***
(.171)
.561
1160
Model 1
Model 2
Model 3
Model 4
Pornography viewing
frequency a
.027
.225*
.007
.310**
(.030)
(.115)
.026+
(.027)
(.101)
.040**
Pornography viewing
(quadratic)
.654***
Religious service
attendance a
(.022)
Other Protestant
Other religion
Unafliated
Male
Married
Number of children
Bachelors or better
Household income
White b (ref.)
Black b
Hispanic
Asian
Native American
South
(.015)
Catholic
Jewish
(.013)
.380*** .378***
(.017)
(.017)
Conservative Protestant
(ref.)
Mainline Protestant b
Age
(.015)
.026+
Prayer
Prayerb
Constant
Adjusted R2
N
.388*
.383*
(.170)
(.169)
.109
.099
(.165)
(.165)
.224+
.219+
(.132)
(.132)
1.038** 1.042**
(.353)
(.353)
.489*
.501*
(.228)
(.228)
.851*** .851***
(.153)
(.152)
.564***
.543***
(.106)
(.107)
.011***
.010**
(.003)
(.003)
.033
.064
(.098)
(.099)
.193+
.199+
(.108)
(.108)
.011
.009
(.028)
(.028)
.126
.116
(.106)
(.106)
.005
.006
(.011)
(.011)
.550***
(.149)
.401**
(.145)
.590**
(.226)
.212
(.712)
.084
(.096)
.255
(.252)
.605
1161
.544***
(.149)
.384**
(.145)
.579**
(.226)
.184
(.712)
.083
(.096)
.517+
(.291)
.606
1161
.171
(.151)
.065
(.146)
.170
(.117)
.670*
(.315)
.053
(.203)
.652***
(.136)
.810***
(.095)
.007*
(.003)
.125
(.087)
.189*
(.096)
.033
(.025)
.111
(.093)
.024*
(.010)
.164
(.150)
.049
(.146)
.162
(.117)
.676*
(.314)
.076
(.202)
.651***
(.136)
.775***
(.095)
.005+
(.003)
.077
(.088)
.199*
(.096)
.030
(.025)
.128
(.093)
.023*
(.010)
.457***
(.132)
.208
(.128)
.050
(.211)
.958
(.630)
.158+
(.084)
1.441***
(.236)
.552
1156
.448***
(.132)
.179
(.128)
.058
(.211)
.920
(.628)
.154+
(.084)
1.840***
(.268)
.556
1156
PERRY
Religious Doubts
Religious Salience
2
1
2.9*
1.6
No Porn
3.7
2.8**
2
1
No Porn
5
4.1
3
4.4
No Porn
3.2**
No Porn
Figure 1. Predicted values from OLS LDV models predicting religiosity measures at Wave 2 on whether one ever viewed pornography at all at Wave 1.
*Differences signicant at .05 level; **Differences signicant at .01 level.
Religious Doubts
Never
Once or
Twice
A Few
Times
Once a
Month
2-3 Times a
Month
Once a
Week
A Few
Times a
Week
Never
Once or
Twice
A Few
Times
Once a
Month
2-3 Times a
Month
Once a
Week
A Few
Times a
Week
Once a Day
or More
Never
Once or
Twice
A Few
Times
Once a
Month
2-3 Times a
Month
Once a
Week
A Few
Times a
Week
Once a Day
or More
10
Once a Day
or More
Prayer Frequency
Religious Salience
Prayer Frequency
8
6
Never
Once or
Twice
A Few
Times
Once a
Month
2-3 Times a
Month
Once a
Week
A Few
Times a
Week
Once a Day
or More
PERRY
Limitations
Several limitations are worth acknowledging in order to outline
directions for future research. The strength and contribution of
this analysis is that the panel design permits the determination of
temporal precedence and directionality of effect between porn
consumption in Wave 1 and religiosity measures in Wave 2. Yet
the fact that the question about porn use was not asked in Wave
2 precludes the possibility of determining whether and to what
extent measures of religiosity at Time 1 predict pornography use
at Time 2. While it is almost certainly the case that religiosity
does in fact predict porn use and frequency, it would be informative to compare the bidirectional effects of porn use and
religiosity over time. Future research would ideally make use
of data that contains measures for both pornography and religiosity at two different time periods so as to view which factor
more strongly predicts the other. This data limitation also precluded the possibility of other estimation procedures that rely on
changing scores, like xed or random effects. Though, as
explained, OLS LDV models generally provide accurate (even
conservative) estimates and are commonly used for studies of
religious change over time, change-score models would provide
another test to ensure that omitted variable bias was not inuencing the effects.
Second, while the pornography measure in PALS is an
improvement on other measures that ask only whether a respondent looks at pornography at all (e.g., the GSS), the measure
does not specify the type of pornographic materials that are
being viewed. It is possible that certain types of sexually explicit
media are more damaging to religiosity over time than others.
For instance, more standard fare or soft-core types of sexual
media are more culturally acceptable than, say, hard-core or
fetish types of media and thus may produce differing levels of
shame, cognitive dissonance, or scrupulosity, affecting religious
belief, commitment, and behavior in different ways. Future
quantitative research would thus benet from measures that
more explicitly dene what pornography means for the
respondent. Qualitative interviews would obviously also help
in this regard as well.
Last, as acknowledged, the religious outcome measures are
single items, which may not provide the most comprehensive
picture of pornographys inuence on religiosity over time.
Following the example of Grubbs et al. (2015), future studies
might effectively include scales or indexes based on a variety
of indicators to measure religiosity more comprehensively
rather than along particular dimensions as done here.
Supplemental Material
Supplementary data for this article can be accessed on
the publishers web site.
12
References
Abell, J. W., Steenbergh, T. A., & Boivin, M. J. (2006). Cyberporn use in
the context of religiosity. Journal of Psychology and Theology, 34,
165171.
Achen, C. (2000, July). Why lagged dependent variables can suppress the
explanatory power of other independent variables. Paper presented at the
annual meeting of the Political Methodological Section of the American
Political Science Association, Los Angeles, CA.
Audia, P. G., & Teckchandani, A. (2010). The effect of connected and
isolated voluntary associations on economic activity in the United
States, 19842000. Social Science Research, 39, 11531163.
doi:10.1016/j.ssresearch.2010.08.003
Avants, S. K., Warburton, L. A., & Margolin, A. (2001). Spiritual and
religious support in recovery from addiction among HIV-positive
injection drug users. Journal of Psychoactive Drugs, 33(1), 3945.
doi:10.1080/02791072.2001.10400467
Baker, J. O., & Smith, B. (2015). American secularism: Cultural contours
of religious nonbelief systems. New York, NY: NYU Press.
Baltazar, A., Helm, H. W., McBride, D., Hopkins, G., & Stevens, J. V.
(2010). Internet pornography use in the context of external and internal
religiosity. Journal of Psychology and Theology, 38, 3240.
Benda, B. B. (1997). An examination of a reciprocal relationship between
religiosity and different forms of delinquency with a theoretical model.
Journal of Research in Crime and Delinquency, 34, 163186.
doi:10.1177/0022427897034002001
Benda, B. B., & Corwyn, R. F. (1997). A test of a model with reciprocal
effects between religiosity and various forms of delinquency using 2stage least squares regression. Journal of Social Service Research, 22,
2752. doi:10.1300/J079v22n03_02
Benda, B. B., & Corwyn, R. F. (2000). A theoretical model of religiosity
and drug use with reciprocal relationships: A test using structural
equation modeling. Journal of Social Service Research, 26, 4367.
doi:10.1080/01488370009511336
Bock, E. W., Beeghley, L., & Mixon, A. J. (1983). Religion, socioeconomic
status, and sexual morality: An application of reference group theory.
Sociological Quarterly, 24, 545559. doi:10.1111/tsq.1983.24.issue-4
Bridges, A. J., & Morokoff, P. J. (2011). Sexual media use and relational
satisfaction in heterosexual couples. Personal Relationships, 18, 562
585. doi:10.1111/pere.2011.18.issue-4
Brown, A. E., Whitney, S. N., Schneider, M. A., & Vega, C. P. (2006).
Alcohol recovery and spirituality: Strangers, friends, or partners?
Southern Medical Journal, 99(6), 654657. doi:10.1097/01.
smj.0000198271.72795.ab
Carroll, J. S., Padilla-Walker, L. M., Nelson, L. J., Olson, C. D., McNamara
Barry, C., & Madsen, S. D. (2008). Generation XXX: Pornography
acceptance and use among emerging adults. Journal of Adolescent
Research, 23(1), 630. doi:10.1177/0743558407306348
Chaves, M. (2011). American religion: Contemporary trends. Princeton,
NJ: Princeton University Press.
Cornwall, M., Albrecht, S. L., Cunningham, P. H., & Pitcher, B. L. (1986).
The dimensions of religiosity: A conceptual model with an empirical
test. Review of Religious Research, 27(3), 226244. doi:10.2307/
3511418
Coughlin, P. (2009). Pornography and virtual indelity. Focus on the
Family. Retrieved from http://www.focusonthefamily.com/marriage/
divorce-and-indelity/pornography-and-virtual-indelity/virtual-inde
lity-and-marriage
Doran, K., & Price, J. (2014). Pornography and marriage. Journal of Family
and Economic Issues, 35, 489498. doi:10.1007/s10834-014-9391-6
Doring, N. M. (2009). The Internets impact on sexuality: A critical review
of 15 years of research. Computers in Human Behavior, 25(5), 1089
1101. doi:10.1016/j.chb.2009.04.003
Driscoll, M. (2009). Porn-again Christian: A frank discussion on pornography and masturbation. Retrieved from https://archive.org/details/
Porn-againChristian
13