Jump to ContentJump to Main Navigation

## Mark D. Regnerus

Print publication date: 2007

Print ISBN-13: 9780195320947

Published to British Academy Scholarship Online: May 2007

DOI: 10.1093/acprof:oso/9780195320947.001.0001

Show Summary Details

# (p.215) Appendix A Regression Models

Source:
Forbidden Fruit
Publisher:
Oxford University Press

Regression analysis is a statistician's way of performing a controlled experiment. In order to isolate the effects of one variable on another, I must account for, or hold constant, other factors that might confound the association. Regression allows me to do this by simultaneously evaluating the independent effect of each variable on the outcome of interest. When coefficients are presented, a number greater than zero means that an increase in that variable has a positive association with the dependent variable (outcome) under examination. If the coefficient is less than zero, then an increase in that variable has a negative association with the outcome. When odds ratios are presented, a value greater than one indicates an increase in the odds of a higher level of the dependent variable, while values less than one are indicative of a reduction in the odds of a higher level of the outcome.

These tables also bring up the question of statistical significance. Because the surveys I use are samples of the American adolescent population, there remains the possibility that findings are a result of chance due to sampling error. That is, results may vary slightly because different samples of the population would yield slightly different results. Because of the large number of respondents in the NSYR, Add Health, and NSFG data sets, however, we can be confident that the results are similar to what would be obtained from analyses of the total population of American adolescents. Furthermore, I have performed tests of statistical significance that determine the actual likelihood that my findings are due to chance, or sampling error. A coefficient or odds ratio with a + next to it suggests that there is less than a 10 percent chance that the difference is due to sampling and is not a “real difference.” One star (*) means there is less than a 5 percent chance, two stars (**) indicates less than a 1 percent chance, and three stars (***) signifies that there is less than a one‐tenth of 1 percent chance that the finding is due to sampling error. If nothing (p.216) appears next to a coefficient or odds ratio, it is implied that there is no statistically significant effect of that variable on the outcome.

Performing multivariate analyses like these boosts confidence that the associations between religion and sexual outcomes are actually the result of religion and not some other variable, such as race/ethnicity, gender, or age.

Table A3.1 Odds Ratios from Ordered Logit Regression Estimates of the Parent‐Reported Frequency of Discussion about Sexual Morality

Model 1

Model 2

Parental Religion Measures

Church attendance

1.148**

Importance of religion

1.448***

Evangelical Protestant

0.811*

1.139

Mainline Protestant

0.502***

0.819

Catholic

0.491***

0.770+

Jewish

0.189***

0.357***

Mormon (LDS)

0.993

1.364

Other religion

0.650***

0.998

No religion

0.254***

1.157

Parental Controls

Parents' average education

0.819**

White

0.804+

Hispanic

0.757+

Asian American

0.559**

Age

0.979***

Female

1.306**

Bio‐intact, two‐parent family

0.974

Disapproves of sex at child's age

1.188***

Thinks their child has already had sex

1.244***

Adolescent Controls

Female

1.648***

Age

1.066**

School has sex education

Curriculum

0.928

Has taken abstinence pledge

1.330***

Family well‐being

1.026**

Number of recent romantic partners

1.076**

Model Fit Statistics

−2 log likelihood

37,706.5

36,595.3

Pseudo R 2

0.016

0.045

N

13,726

13,726

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

(p.217)

Table A3.2 Odds Ratios from Ordered Logit Regression Estimates of the Parent‐Reported Frequency of Discussion about Sex and Birth Control with Their Adolescent Child

Frequency of talk about sex

Frequency of talk about birth control

Model 1

Model 2

Model 3

Model 1

Model 2

Model 3

Parental Religion Measures

Church attendance

0.902***

0.796***

0.893***

0.816***

Importance of religion

1.224***

0.970

1.101*

0.887**

Evangelical Protestant

0.516***

0.757+

0.637**

0.501***

0.671**

0.582***

Mainline Protestant

0.428***

0.671*

0.698*

0.484***

0.705**

0.722**

Catholic

0.369***

0.648**

0.689*

0.424***

0.664**

0.701**

Jewish

0.394***

0.638+

1.305

0.517***

0.814

1.531+

Mormon (LDS)

0.508**

0.935

0.696

0.379***

0.634*

0.497***

Other religion

0.454***

0.708*

0.635**

0.481***

0.677**

0.629***

No religion

0.365***

0.679+

0.531**

0.528***

0.676*

0.559**

Parental Controls

Parents' average education

1.063

1.241***

0.946

1.036

White

0.755*

0.802*

0.841

0.891

Hispanic

0.542***

0.549***

0.660**

0.701**

Asian American

0.295***

0.340***

0.407***

0.479***

Age

0.965***

0.972***

0.962***

0.968***

Female

1.300**

1.159+

1.231*

1.102

Bio‐intact, two‐parent family

0.780***

0.751***

0.768***

0.745***

Disapproves of sex at child's age

1.094***

0.988

1.011

0.921***

Thinks child has already had sex

2.359***

2.542***

2.949***

3.116***

Frequency of talk about morality of sex

3.787***

2.841***

Adolescent Controls

Female

1.544***

1.201***

1.228***

0.977

Age

1.096***

1.070***

1.134***

1.120***

School has sex education curriculum

1.132

1.230

1.143

1.205

Has taken abstinence pledge

1.092

0.920

0.983

0.839*

Family well‐being

1.036**

1.027**

1.028**

1.018*

Number of recent romantic partners

1.154***

1.138***

1.128***

1.105***

Model Fit Statistics

−2 log likelihood

36,581.1

35,159.6

30,060.2

39,603.6

37,806.5

34,324.0

Pseudo R 2

0.012

0.051

0.188

0.008

0.053

0.140

N

13,726

13,726

13,726

13,726

13,726

13,726

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

(p.218)

(p.219)

Table A3.3 Odds Ratios from Ordered Logit Regression Estimates of the Parent‐Reported Frequency and Ease/Difficulty of Talking about Sex

Frequency of talking about sex

Model 1

Model 2

Model 3

Parental Religion Measures

Church attendance

0.952*

0.980

Importance of religion

1.107*

1.091*

Evangelical Protestant

0.700*

0.708*

0.587*

Mainline Protestant

0.514***

0.525***

0.444**

Catholic

0.533***

0.544***

0.519*

Jewish

0.625*

0.641*

0.585+

Mormon (LDS)

0.720

0.730

0.544

Other religion

0.474*

0.478*

0.582

No religion

0.702+

0.727

0.728

Parental Controls

Parents' average education

1.007

White

1.403

Hispanic

0.937

Asian American

0.323**

Age

0.977***

Female

1.982***

Respondent parent is married

0.862

Thinks people should wait to have sex until married

1.037

Knows or thinks child is dating

1.200***

Frequency of talking about sex

Adolescent Controls

Female

1.703***

Age

1.084*

Has already had sex

1.313*

Thinks people should wait until married to have sex

0.999

Quality of relationship with parents

1.042

Model Fit Statistics

−2 log likelihood

6,391.7

6,381.1

6,157.8

Pseudo R 2

0.005

0.007

0.041

N

3,089

3,089

3,089

Ease of talking with child about sex

Model 1

Model 2

Model 3

Model 4

Parental Religion Measures

Church attendance

0.920***

0.936**

0.938*

Importance of religion

1.115**

1.080+

1.058

Evangelical Protestant

0.415***

0.414***

0.577+

0.677

Mainline Protestant

0.273***

0.266***

0.414**

0.485*

Catholic

0.334***

0.330***

0.484*

0.574+

Jewish

0.334***

0.322***

0.585

0.655

Mormon (LDS)

0.303***

0.312***

0.466+

0.561

Other religion

0.438**

0.425**

0.695

0.719

No religion

0.489***

0.438***

0.670

0.683

Parental Controls

Parents' average education

0.940**

0.938**

White

0.757

0.646

Hispanic

0.728

0.693

Asian American

0.515

0.685

Age

0.991

0.998

Female

1.494***

1.220*

Respondent parent is married

0.902

0.908

Thinks people should wait to have sex until married

1.132

1.136

Knows or thinks child is dating

1.138*

1.078

Frequency of talking about sex

2.923***

Adolescent Controls

Female

1.546***

1.341***

Age

1.029

1.005

Has already had sex

1.457***

1.407**

Thinks people should wait until married to have sex

1.058

1.070

Quality of relationship with parents

1.094***

1.093***

Model Fit Statistics

−2 log likelihood

6,344.4

6,326.8

6,191.1

5,776.0

Pseudo R 2

0.017

0.020

0.041

0.105

N

2,840

2,840

2,840

2,840

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Survey of Youth and Religion

(p.220)

(p.221)

Table A3.4 Race/Religion Interaction Effects from Ordered Logit Regression Estimates of Frequency of Discussion about Sex and Birth Control with Their Adolescent Child

Frequency of talk about sex

Frequency of talk about birth control

Parental Measures

Weekly church attendance

−0.303**

−0.412***

Parent is white

−0.484***

−0.483***

Interaction Effect

Weekly church attendance×white

−0.330**

−0.286*

Model Fit Statistics

−2 log likelihood

27,173.5

31,227.1

Pseudo R 2

0.179

0.123

N

12,233

12,233

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

Note: Control variables are included but not shown and are identical to those displayed in Table A3.2.

Table A3.5 Odds Ratios from Ordered Logit Regression Estimates of Parental Misgivings about Discussing Sexual Issues with Their Adolescent Child

Respondent doesn't know enough about topic

Difficult to explain things to child

Child will learn about sex elsewhere

Talking about birth control would only encourage sex

Parental Religion Measures

Church attendance

1.043

1.066*

0.985

1.059*

Importance of religion

1.039

1.029

1.060

1.135**

Model Fit Statistics

−2 log likelihood

28,527.7

29,624.5

28,527.4

27,813.5

Pseudo R 2

0.100

0.110

0.116

0.077

N

13,726

13,726

13,726

13,726

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

Note: Model includes (but does not display) controls for parent religious affiliation; parent race, age, and gender; parents' average education; adolescent gender, age, and family satisfaction; intact family; exposure to a school sex education curriculum; whether the parent thinks sex is inappropriate during adolescence; parent's communication about sex and sexual morality; and whether the parent thinks his/her adolescent child has already had sex.

(p.222)

Table A3.6 Odds Ratios from Logistic Regression Estimates of First Sex (Virginity Loss) as a Function of Parent‐Child Communication about Sex, Contraception, Sexual Morality, and Several Control Variables, Wave I Virgins Only

Model 1

Model 2

Parent talks about sex

1.147+

1.129

Parent talks about contraception

1.265***

1.228***

Parent talks about sexual morality

0.903*

0.961

Parent thinks adolescent has had sex

2.386**

Adolescent's church attendance

0.918+

Adolescent's importance of religion

0.914

Adolescent has taken an abstinence pledge

0.796*

Model Fit Statistics

−2 log likelihood

7,664.1

7,573.0

Pseudo R 2

0.132

0.142

N

6,385

6,385

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

Note: Model includes controls for age, gender, parents’ average education level, biologically intact two‐parent family, race/ethnicity, and number of adolescent's recent romantic partners.

Table A3.7 Odds Ratios from Ordered Logit Regression Estimates of the Score on Adolescent Pregnancy Awareness Quiz

Wave I

Wave II

Church attendance

1.000

1.019

Importance of religion

0.948+

0.925*

Identifies as “born again”

0.909

0.875

Female

1.406***

1.476***

Age

1.096**

1.058+

White

1.145+

1.460***

Asian American

1.513**

1.443*

Hispanic

1.081

1.292*

Parents' average education

1.444***

1.281**

Has had sexual intercourse (by Wave I)

1.595***

1.218**

Has had sexual intercourse (by Wave II)

1.383***

Frequency of talking to parent about sex

0.990

1.013

Frequency of talking to parent about contraception

1.021

1.004

Frequency of talking to parent about sexual morality

0.963

0.990

Has taken an abstinence pledge

0.853*

0.762**

Model Fit Statistics

−2 log likelihood

32,972.8

20,901.5

Pseudo R 2

0.011

0.010

N

9,716

6,297

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

(p.223)

Table A3.8 Odds Ratios from Ordered Logit Regression Estimates of Attraction to Members of the Same Sex, among Unmarried 15‐ to 19‐Year‐Olds

Boys

Girls

Model 1

Model 2

Model 3

Model 1

Model 2

Model 3

Church attendance

0.762*

0.960

0.743**

0.863

Importance of religion

0.472**

0.443*

0.549***

0.894

Evangelical Protestant

1.502

1.114

Black Protestant

1.112

1.060

Catholic

1.408

0.807

Other religion

7.828**

2.308*

No religion

1.550

2.622*

Age

1.145

1.153

1.132

1.052

1.087

1.073

African American

1.821

2.213

2.718

0.787

0.850

0.813

Hispanic

2.098+

2.460*

2.963**

0.483*

0.513+

0.605

Other race

1.965

2.190

1.313

0.783

0.789

0.664

Lives in the suburbs

2.089+

2.199*

1.944+

0.756

0.723

0.679+

Lives in rural area

2.711*

2.869*

3.039*

0.602

0.589

0.623

Model Fit Statistics

−2 log likelihood

540.5

531.6

508.5

1,096.0

1,088.3

1,066.7

Pseudo R 2

0.047

0.063

0.103

0.058

0.065

0.084

N

1,068

1,068

1,068

1,083

1,083

1,083

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Survey of Family Growth, Cycle 6

Note: Models include controls for parents’ average education level, biologically intact two‐parent family, whether the respondent still lives with his/her parent(s), educational status, and whether the respondent has ever had heterosexual sex.

(p.224)

Table A3.9 Odds Ratios from Logistic Regression Estimates of Ever Having Had Same‐Sex Relations among Unmarried 15‐ to 19‐Year‐Olds

Boys

Girls

Model 1

Model 2

Model 3

Model 1

Model 2

Model 3

Church attendance

0.842

0.984

0.805+

0.953

Importance of religion

0.653

0.636

0.581**

0.706

Evangelical Protestant

0.917

1.579

Black Protestant

0.561

3.345+

Catholic

2.848+

0.782

Other religion

8.704***

1.681

No religion

1.556

2.076

Age

1.506*

1.528*

1.547*

0.943

0.960

0.950

African American

1.769

1.960

3.071*

0.832

0.921

0.420

Hispanic

1.718

1.904

1.554

0.403*

0.425*

0.499+

Other race

1.143

1.216

0.663

0.321+

0.318+

0.315+

Lives in the suburbs

1.945

2.002+

2.096+

0.819

0.797

0.773

Lives in rural area

2.789*

2.899**

3.367**

0.682

0.703

0.696

Has had heterosexual sex

2.759*

2.675*

2.926*

3.151***

2.977**

2.957**

Model Fit Statistics

−2 log likelihood

339.3

337.2

312.7

643.2

623.6

634.9

Pseudo R 2

0.136

0.142

0.204

0.114

0.141

0.125

N

1,079

1,079

1,079

1,100

1,100

1,100

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Survey of Family Growth, Cycle 6

Note: Models include controls for parents’ average education level, biologically intact two‐parent family, whether the respondent still lives with his/her parent(s), and educational status.

(p.225)

Table A3.10 Odds Ratios from Ordered Logit Regression Estimates of Homosexual Sexual Orientation, among Unmarried 15‐ to 19‐Year‐Olds

Boys

Girls

Model 1

Model 2

Model 3

Model 1

Model 2

Model 3

Church attendance

0.694**

0.914

0.777

0.933

Importance of religion

0.416

0.399

0.560

0.575

Evangelical Protestant

0.184

1.010

Black Protestant

0.143

1.105

Catholic

0.653

0.776

Other religion

5.514

0.632

No religion

0.436

0.829

Age

1.355

1.378

1.295

1.409

1.443

1.427

Intact family

0.723

0.786

0.642

3.124

3.412

3.278

Lives with parents

1.767

1.662

1.659

0.156*

0.133*

0.142*

Model Fit Statistics

−2 log likelihood

167.5

164.3

144.7

103.1

102.2

101.9

Pseudo R 2

0.110

0.127

0.231

0.078

0.086

0.088

N

1,047

1,047

1,047

1,070

1,070

1,070

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Survey of Family Growth, Cycle 6

Note: Race and parents’ education predict statistical failure perfectly among girls. They are dropped from the analyses. All models include controls for race, suburban/rural residence, parents’ average education level, educational status, and whether the respondent has ever had heterosexual sex.

(p.226)

Table A3.11 Odds Ratios from Ordered Logit Regression Estimates of Bisexual Sexual Orientation, among Unmarried 15‐ to 19‐Year‐Olds

Boys

Girls

Model 1

Model 2

Model 3

Model 1

Model 2

Model 3

Church attendance

0.914

0.795

0.758+

0.886

Importance of religion

1.216

1.242

0.555**

0.727

Evangelical Protestant

2.287

1.564

Black Protestant

2.895

2.210

Catholic

3.329

0.729

Other religion

16.601**

1.948

No religion

2.327

1.996

Age

1.419

1.401

1.351

0.961

0.881

0.861

African American

1.151

0.930

0.708

0.638

0.679

0.451

Hispanic

5.064*

4.730*

5.181*

0.635

0.678

0.825

Other race

2.419

2.222

1.515

0.294

0.303

0.289

In high school

4.174

4.400

3.828

1.513

1.489

1.310

High school degree or less

1.191

1.294

0.841

5.471**

5.622**

5.071**

Has had heterosexual sex

0.817

0.896

0.944

1.477

1.463

1.396

Model Fit Statistics

−2 log likelihood

128.6

128.6

118.4

444.8

440.1

432.3

Pseudo R 2

0.136

0.136

0.204

0.094

0.104

0.120

N

1,047

1,047

1,047

1,065

1,065

1,065

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Survey of Family Growth, Cycle 6

Note: Models include controls for suburban/rural residence, parents’ average education level, biologically intact two‐parent family, and whether the respondent still lives with his/her parent(s).

(p.227)

Table A4.1 Odds Ratios from Logistic Regression Estimates of Taking a Pledge of Sexual Abstinence until Marriage

Model 1

Church attendance

1.251***

Importance of religion

1.460***

Mainline Protestant

0.577***

Black Protestant

0.512**

Catholic

0.664**

Mormon (LDS)

1.437

Jewish

0.184*

Other religion

0.788

No religion

2.048**

Identifies as “born again”

2.169***

Age

1.011

Female

1.704***

Family satisfaction

1.051**

Level of autonomy

0.915**

Strategic

1.066***

Model Fit Statistics

−2 log likelihood

10,123.5

Pseudo R 2

0.156

N

14,501

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

Note: Model also includes controls for parents’ average education level, southern residence, biologically intact two‐parent family, social desirability, race/ethnicity, virginity status, and number of adolescent's recent romantic partners.

(p.228)

Table A4.2 Odds Ratios from Ordered Logit Regression Estimates of Motivations to Have or Avoid Sex

After sex, you would feel guilty

Friends would respect you more

Partner would lose respect for you

Having sex would make you attractive

Sex would give you much pleasure

Having sex would upset your mother

Pregnancy would embarrass you

Church attendance

1.267***

0.950+

1.148***

0.978

0.987

1.310***

1.171***

Importance of religion

1.176**

0.954

1.088+

0.911+

0.956

1.157**

0.972

Mainline Protestant

0.806*

1.031

0.842+

1.123

0.887

0.789*

0.995

Black Protestant

0.657*

0.977

0.763+

1.068

0.833

0.544***

0.808

Catholic

0.735**

1.100

0.854

0.979

1.044

0.852

0.974

Mormon (LDS)

1.475

0.683

1.102

0.772

0.928

1.711+

1.005

Jewish

0.543**

1.274

0.656+

0.776

1.510

0.420***

1.678*

Other religion

0.905

0.805+

0.816

0.879

1.008

0.949

0.997

No religion

1.230

0.817

1.134

0.756+

0.878

1.022

0.927

Identifies as “born again”

1.419***

0.831*

1.260*

0.905

0.855+

1.332***

1.076

Age

0.947*

1.004

0.932*

0.934*

1.167***

0.841***

0.917***

Female

1.915***

0.278***

1.213***

0.346***

0.283***

2.086***

1.361***

African American

0.792*

1.732***

0.837+

1.079

1.008

0.859

0.516***

Bio‐intact, two‐parent family

1.198***

0.861*

1.101+

0.939

1.044

1.733***

1.351***

Strategic

1.034***

0.950***

0.994

0.950***

1.019*

1.030**

1.056***

Has already had sex

0.266***

1.557***

0.465***

1.192*

1.843***

0.456***

0.464***

Model Fit Statistics

−2 log likelihood

29379.7

27171.0

28706.3

26891.8

27001.9

25846.5

28820.1

Pseudo R 2

0.087

0.059

0.029

0.042

0.072

0.093

0.060

N

9,688

9,688

9,688

9,688

9,688

9,688

9,688

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

Note: All models include controls for parents’ average education level, southern residence, level of autonomy, social desirability, other races/ethnicities, and number of adolescent's recent romantic partners.

(p.229)

Table A5.1 Odds Ratios from Logistic Regression Estimates of Having Experienced Sexual Intercourse by Wave I

Model 1

Model 2

Model 3

Model 4

Church attendance

0.821***

0.828***

0.856***

0.927

Importance of religion

0.883***

0.929*

0.876**

0.969

Mainline Protestant

0.671***

0.744+

Black Protestant

1.246

1.213

Catholic

0.610***

0.600***

Mormon (LDS)

0.333*

0.289*

Jewish

0.459*

0.330*

Other religion

0.691*

0.702*

No religion

0.589**

0.716

Identifies as “born again”

0.981

0.973

Age

1.811***

1.601***

1.567***

1.452***

Female

0.937

0.886+

0.892

1.371***

Parents’ average education

0.501***

0.557***

0.532***

0.522***

Lives in the South

1.309**

1.234**

1.192*

1.237*

African American

2.718***

2.416***

2.115***

1.594*

Hispanic

1.090

1.210

1.540***

1.028

Asian American

0.599+

0.597+

0.732

0.818

Bio‐intact, two‐parent family

0.557***

0.593***

0.630***

0.671***

Family satisfaction

0.879***

0.896***

0.895***

Level of autonomy

1.085***

1.054

Strategic

0.942***

0.952***

0.962**

Social desirability

1.126

1.183*

1.146

Number of recent romantic partners

1.609***

1.609***

School percentage nonvirgins

9.118***

7.709***

5.373***

Has taken an abstinence pledge

0.246***

0.298***

Sex would bring guilt

0.584***

Sex would upset your mother

0.817***

Sex would bring friends’ respect

1.230***

Model Fit Statistics

−2 log likelihood

12081.9

11706.2

11006.0

7386.7

Pseudo R 2

0.171

0.197

0.245

0.253

N

10,757

10,757

10,757

6,845

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

(p.230)

Table A5.2 Odds Ratios from Logistic Regression Estimates of Having Had Sexual Intercourse by Wave II, Using Full Sample and Parent Religiosity Proxies

Model 1

Model 2

Model 3

Model 4

Parent's church attendance

0.843***

0.855***

0.889*

0.854**

Parent's importance of religion

1.002

1.040

0.986

1.047

Model Fit Statistics

−2 log likelihood

12930.1

12143.2

8174.2

7706.4

Pseudo R 2

0.177

0.227

0.182

0.229

N

11,456

11,456

7,245

7,245

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

Note: All models include controls for age, gender, religious affiliation, parents’ average education level, southern residence, biologically intact two‐parent family, family satisfaction, social desirability, proclivity for risk taking, strategic orientation, school percentage nonvirgins, and race/ethnicity. Model 4 also controls for number of adolescent's recent romantic partners and three attitudes (sex would bring guilt, sex would upset mother, sex would bring respect of friends) about sex.

(p.231)

Table A5.3 Odds Ratios from Logistic Regression Estimates of Having Had Sexual Intercourse by Wave II, Wave I Virgins Only

Model 1

Model 2

Model 3

Model 4

Church attendance

0.880*

0.890*

0.888*

0.914

Importance of religion

0.851*

0.864*

0.814*

0.803*

Mainline Protestant

0.906

0.876

Black Protestant

1.511+

2.112*

Catholic

1.094

1.037

Mormon (LDS)

0.798

0.981

Jewish

0.509+

0.220***

Other religion

1.014

1.203

No religion

0.886

0.996

Identifies as “born again”

0.951

Parents’ average education

0.601***

0.653***

0.595***

0.557***

African American

1.441***

1.361***

1.190

0.875

Hispanic

1.072

1.165

1.204

0.862

Asian American

0.854

0.869

0.969

0.930

Bio‐intact, two‐parent family

0.562***

0.581***

0.602***

0.723**

Strategic

0.954**

0.956**

0.973

Likes taking risks

1.202***

1.153***

1.036

Number of recent romantic partners

1.625***

1.791***

School percentage nonvirgins

4.739***

4.667***

3.254*

Has taken an abstinence pledge

0.909

0.862

Sex would bring guilt

0.689***

Sex would upset your mother

0.974

Sex would bring friends’ respect

1.204***

Model Fit Statistics

−2 log likelihood

7003.2

6878.8

6550.1

3901.9

Pseudo R 2

0.067

0.083

0.127

0.126

N

7,117

7,117

7,117

3,833

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

Note: All models include controls for age, gender, and southern residence. Some models also control for level of autonomy, family satisfaction, and social desirability.

(p.232)

Table A5.4 Odds Ratios from Logistic Regression Estimates of Having Had Sexual Intercourse by Wave II, Using Religious Change Measures

Model 1

Model 2

Model 3

Model 4

Change in attendance

1.016 (0.889**)

Change in importance

0.905 (0.958)

Drastic increase in attendance

1.097 (0.918)

1.056 (.892)

Drastic increase in importance

1.176 (0.943)

1.176 (.936)

Drastic decrease in attendance

1.082 (1.400*)

1.082 (1.398*)

Drastic decrease in importance

1.514* (1.369*)

1.496* (1.380*)

Model Fit Statistics

−2 log likelihood

6613.6 (12217.4)

6603.1 (12212.5)

6618.1 (12246.8)

6604.9 (12212.7)

Pseudo R 2

0.112 (0.209)

0.113 (0.209)

0.111 (0.207)

0.113 (0.209)

N

7,430 (11,266)

7,430 (11,266)

7,430 (11,266)

7,430 (11,266)

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

Note: Coefficients in parentheses are for regression models using the virgin‐only sample. Models also control for age, gender, region, race/ethnicity, intact family, planful personality, aversion to risk taking, social desirability, number of adolescent's recent romantic partners, Wave I religious service attendance, and Wave I religious salience.

(p.233)

Table A5.5 Reverse Causation–Odds Ratios from Ordered Logit Regression Estimates of Wave II Attendance and Importance of Religion on Having Experienced Sexual Intercourse between Study Waves, Wave I Virgins Only

Attendance

Importance

Male

Female

Male

Female

Had sex, Wave II

0.890

0.945

0.774+

0.749**

Model Fit Statistics

−2 log likelihood

6942.2

7670.5

6481.5

6800.7

Pseudo R 2

0.246

0.227

0.238

0.235

N

3,313

3,689

3,312

3,689

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

Note: Models include but do not display estimated coefficients from lagged dependent variable, demographic covariates, social desirability, personality traits, etc.

Table A5.6 Negative Binomial Regression Estimates of Reported Number of Times Having Sex on Respondent Characteristics and Behaviors

Model 1

Model 2

Model 3

Church attendance

−0.080**

−0.073**

−0.043+

Importance of religion

−0.158**

−0.059

0.052

White

−0.223+

−0.093

−0.182

Hispanic

−0.395*

−0.197

−0.310

Asian American

0.089

0.041

−0.126

Age

0.736***

0.646***

0.642***

Female

−0.080

−0.100

−0.107

Parents’ average education

−0.099***

−0.088***

−0.087***

Parent respondent is married

−0.344**

−0.217*

−0.208*

Parent perceives adolescent's friends as positive

−0.237***

−0.220***

Level of autonomy

0.119***

0.104***

Is currently in a dating relationship

0.894***

0.912***

Parent respondent considers adolescent to be rebellious

0.239***

0.225***

Adolescent is considered popular

0.285***

0.270***

Spiritual but not religious

0.001

Frequency of Bible reading

−0.120**

Do what makes me happy

0.697***

Do what gets me ahead

0.642**

Do what an adult or parent says

0.212

Model Fit Statistics

−2 log likelihood

4848.4

4600.7

4556.3

Pseudo R 2

0.076

0.123

0.132

N

2,973

2,973

2,973

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Survey of Youth and Religion

(p.234)

Table A5.7 Odds Ratios from Ordered Logit Regression Estimates of Adolescent Attitudes about Birth Control, Wave I

Using birth control is morally wrong (full sample)

Friends might think respondent is looking for sex if respondent uses birth control (adolescent girls only)

Model 1

Model 2

Model 1

Model 2

Church attendance

1.166***

1.147***

1.168***

1.185***

Importance of religion

1.069+

1.144**

1.004

1.078

Mainline Protestant

0.883

1.012

Black Protestant

0.632***

1.039

Catholic

1.190+

0.887

Mormon (LDS)

1.142

0.706

Jewish

0.440+

0.687

Other religion

0.775*

0.942

No religion

1.281+

1.310

Identifies as “born again”

0.939

1.225*

Age

0.968

0.966

0.869***

0.884**

Female

0.593***

0.583***

Parents’ average education

0.673***

0.704***

0.786*

0.819*

Lives in the South

1.054

1.032

1.096

1.051

African American

1.242*

1.609***

1.149

1.050

Hispanic

1.556***

1.407***

1.880***

1.747***

Asian American

2.222**

2.119**

2.746***

2.589***

Bio‐intact, two‐parent family

1.019

1.012

1.105

1.095

Family satisfaction

0.980+

0.976*

0.969*

0.964*

Level of autonomy

0.937**

0.910**

Strategic

0.937***

0.934***

0.955**

0.954**

Social desirability

1.045

1.023

0.902

0.893

Number of adolescent's recent romantic partners

0.900***

0.956

Has taken an abstinence pledge

1.305**

1.447**

School percentage nonvirgins

1.108

1.153

0.994

0.985

Has had sexual intercourse

0.792***

0.852**

0.512***

0.546***

Model Fit Statistics

−2 log likelihood

28684.7

28539.6

16087.5

16027.9

Pseudo R 2

0.022

0.027

0.028

0.031

N

10,852

10,852

5,314

5,314

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

(p.235)

Table A5.8 Odds Ratios from Logistic Regression Estimates of Birth Control Practices among Adolescents Who First Had Sex between Waves I and II

Used birth control at first intercourse

Used birth control at most recent intercourse

Church attendance

1.154

1.169

Importance of religion

0.793*

0.967

Identifies as “born again”

1.371

0.888

Age

1.128

1.045

Female

0.991

1.072

Model Fit Statistics

−2 log likelihood

1086.3

1364.2

Pseudo R 2

0.037

0.027

N

910

1,140

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Longitudinal Study of Adolescent Health

Note: Models also include controls for age, gender, parents' average education, southern residence, race/ethnicity, biologically intact two‐parent family, family satisfaction, strategic orientation, social desirability, and school percentage nonvirgins.

Table A6.1 Odds Ratios from Logistic Regression Estimates of Oral Sexual Experience

Model 1

Model 2

Model 3

Spiritual but not religious

1.293***

1.073

0.993

Mainline Protestant

1.184

1.051

0.786

Black Protestant

0.445***

0.542+

0.386*

Catholic

0.736*

0.562***

0.400***

Mormon (LDS)

0.424+

0.375*

0.392*

Jewish

1.447

1.205

0.683

Other religion

0.963

0.826

0.645

No religion

1.426*

0.658+

0.533*

Church attendance

0.938+

1.024

Importance of religion

0.772***

1.024

Female

0.740**

0.765*

Quality of parent‐child relations

0.850***

Proponent of abstinence until marriage

0.248***

Is currently in a dating relationship

2.212***

Parents would be upset if respondent had sex

0.579***

Model Fit Statistics

−2 log likelihood

3157.7

2643.8

2173.6

Pseudo R 2

0.024

0.183

0.328

N

3,060

3,060

3,060

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Survey of Youth and Religion

Note: All models also include control variables for different races/ethnicities (coefficients not shown). Models 2 and 3 also include controls for parents' perceptions about respondent's friends, parents' average education, age, parents' marital status, and the perceived rebelliousness of the respondent.

(p.236)

Table A6.2 Odds Ratios from Logistic Regression Estimates of Having Given Oral Sex, Having Received Oral Sex, and Having Had Anal Sex

Has given oral sex

Has received oral sex

Has had anal sex

Church attendance

0.778***

0.832*

0.578***

Importance of religion

0.705*

0.698*

1.006

Evangelical Protestant

1.003

0.726

0.692

Black Protestant

1.252

1.059

1.025

Catholic

0.720

1.022

0.621

Other religion

0.787

0.576*

0.262*

No religion

1.006

1.152

1.407

Model Fit Statistics

−2 log likelihood

12,024.1

11,296.4

14,478.6

Pseudo R 2

0.146

0.117

0.057

N

1,270

1,270

1,263

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Survey of Family Growth, Cycle 6

Note: All models include controls for age, gender, parents' average education level, biologically intact two‐parent family, urbanicity of residence, and race/ethnicity. The reference category for religious tradition is mainline Protestant.

(p.237)

Table A6.3 Negative Binomial Regression Estimates of Number of Times Having Oral Sex on Respondent Characteristics and Behaviors

Model 1

Model 2

Model 3

Church attendance

−0.022

−0.027

−0.004

Importance of religion

−0.210***

−0.123*

−0.034

White

0.674***

0.875***

0.826***

Hispanic

0.299

0.522*

0.462+

Asian American

−0.102

0.175

−0.172

Age

0.637***

0.543***

0.538***

Female

−0.361***

−0.312**

−0.340***

Parents' average education

−0.068**

−0.046*

−0.046*

Parent respondent is married

−0.299**

−0.235*

−0.227*

Parent perceives adolescent's friends as positive

−0.198***

−0.177***

Level of autonomy

0.105***

0.088***

Is currently in a dating relationship

0.729***

0.744***

Parent respondent considers adolescent to be rebellious

0.200***

0.194***

Respondent is considered popular

0.378***

0.351***

Spiritual but not religious

0.019

Frequency of Bible reading

−0.077*

Do what makes me happy

0.545**

Do what gets me ahead

0.466*

Do what an adult or parent says

−0.012

Model Fit Statistics

−2 log likelihood

4935.7

4723.1

4683.2

Pseudo R 2

0.074

0.114

0.122

N

2,972

2,972

2,972

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Survey of Youth and Religion

(p.238)

Table A6.4 Odds Ratios from Ordered Logit Regression Estimates of Internet Pornography Use among Adolescent Boys

Model 1

Model 2

Spiritual but not religious

1.332**

1.132

Mainline Protestant

1.925**

1.437

Black Protestant

1.533

1.400

Catholic

1.890***

1.691*

Mormon (LDS)

1.292

0.902

Jewish

4.655***

3.161**

Other religion

1.375

1.108

No religion

1.978**

1.003

Church attendance

0.996

Importance of religion

0.858*

Parents perceive adolescent's friends as positive

0.755***

Parents’ average education

1.114**

Age

1.322***

Adolescent has autonomy with media

1.182**

Quality of parent‐child relations

0.866**

Proponent of abstinence until marriage

0.587**

Is currently in a dating relationship

1.534**

Model Fit Statistics

−2 log likelihood

2482.5

2328.8

Pseudo R 2

0.018

0.079

N

1,315

1,214

+ p < .10; * p < .05; ** p < .01; *** p < .001.

Source: National Survey of Youth and Religion