Gender Differences in How Parents, Peers, and Exposure to Sexually Explicit Materials Influence the Intention to Engage in Casual Sex among Adolescents and Young Adults in Taiwan: Applying the Theory of Planned Behavior

Round 1

Reviewer 1 Report

Line 48, is this study a representative sample of young adults throughout the USA? I think the year of this study would be salient to include

Given your background that “A good parent–child relationship reduces the chance of a first sexual experience occurring in a casual relationship context among adolescents [15]. It is reasonable to hypothesize that sufficient parent–child discussions of sexual issues reduce the incidence of casual sex; however, the relationship between parent–child discussions of sexual issues  (hereafter referred to as “parent–child discussion”) and its TPB constructs warrants further study.” I am uncertain where the evidence for hypothesis one comes from, if parent-child discussions may reduce the incidence of casual sex then why is your hypothesis that parent-child discussion leads to casual sex intention?

 

Might there be a bias towards students who have more sex positive parents given a large proportion of the sample were children under 20 who needed to get written parental approval to participate. It is possible parents who do not discuss sex with their children would not be comfortable with their children taking a survey that asks about sexual practice. This should be explored as a limitation.

Author Response

We appreciated your valuable comments. As discussed below, we have revised our manuscript with underlines based on your suggestions. Please let us know if we need to provide anything else regarding this revision.

 

Comment 1

Line 48, is this study a representative sample of young adults throughout the USA? I think the year of this study would be salient to include

Response:

Toledo Adolescent Relationships Study (TARS) is a longitudinal, four Wave dataset originally collected in 2000 based on a random sample derived from the school records of 7^th, 9th, and 11^th grades in Lucas County, OH. This study was based on participants from Wave IV, which was collected in 2006-2007 and included young adults (aged 18-24 years) whose trajectories reflected a wider spectrum of education al experiences. We added the explanation as below into Introduction of the revised manuscript. Please refer to line 51-54.

“The 2006-2007 Toledo Adolescent Relationships Study using the fourth wave dataset originally collected in 2000 based on a random school sample of youth in Lucas County, Ohio found that 59% of young adults (aged 18-24 years) had engaged in casual sex [2].”

 

Comment 2

Given your background that “A good parent–child relationship reduces the chance of a first sexual experience occurring in a casual relationship context among adolescents [15]. It is reasonable to hypothesize that sufficient parent–child discussions of sexual issues reduce the incidence of casual sex; however, the relationship between parent–child discussions of sexual issues (hereafter referred to as “parent–child discussion”) and its TPB constructs warrants further study.” I am uncertain where the evidence for hypothesis one comes from, if parent-child discussions may reduce the incidence of casual sex then why is your hypothesis that parent-child discussion leads to casual sex intention?

Response:

Thank you for your comment. We revised the Hypothesis 1 (H1) and Hypothesis 2 (H2) as below to make it clear. Please refer to line 130-138.

“Hypothesis 1 (H₁): Parent–child discussion on sex-related topics is directly associated with reduced intention to engage in casual sex, whereas peer interaction around sex-related topics and exposure to sexually explicit materials are directly associated with increased intention to engage in casual sex.

Hypothesis 2 (H₂): The three TPB constructs (favorable attitude toward casual sex, subjective norms of accepting casual sex, and perceived control over involvement) mediate the associations of parent–child discussion on sex-related topics, peer interaction around sex-related topics, and exposure to sexually explicit materials with the intention to engage in casual sex.”

 

Comment 3

Might there be a bias towards students who have more sex positive parents given a large proportion of the sample were children under 20 who needed to get written parental approval to participate. It is possible parents who do not discuss sex with their children would not be comfortable with their children taking a survey that asks about sexual practice. This should be explored as a limitation.

Response:

Thanks for your suggestion. We have added it as one of the limitations of this study.

Please refer to line 467-470.

“Fifth, parents who did not discuss sex with their children might not agree their children younger than 20 years old to participate in this survey; it might increase the possibility of the bias in sample representation.”

Reviewer 2 Report

The study examined how parent-child communication, peer influences, and exposure to sexually explicit material were associated with intention to engage in casual sex among adolescents and young adults in Taiwan. It also explored gender differences of these associations. In general, I believe the paper has potential to contribute significantly to the literature. I have some concerns/questions about the mediation analysis though. Please see my comments below.

 

Intro – I like the literature review part. It’s well-written and concise. The flow of the entire section makes sense to me. I also appreciate clear the aims and hypotheses.

 

Methods

 

-p.4 “participants” – I understand that the study took place in Taiwan but can the authors state this again in this section?

 

- p.4, line 146-148 – Some explanation may be needed for why students younger than 20 need parental consent. Is this because of the law? Are people younger than 20 considered a "minor" in Taiwan? This can be different from many countries where people aged 18+ are not considered "adolescent" anymore (and thus no parental consent is needed). The authors should explicitly mention this, so the reader know who are considered “adolescents” and who are considered “young adults” in the context of this study.

 

Results

 

- what is the age range?

 

- Table 2 is confusing and misleading even. First, please number all rows (1, 2, 3, 4, etc.) so it’s easier to see which variable correlates with which variable. Second, at first glance it seems like ALL correlations (except the correlation between attitude and parent-child discussion) are significant because they all have asterisks next to them. The note under the table also doesn’t helpful at all because it still makes it feel like they all indicate significance. Then I realized that it needs to be <.002 to be considered significant. Please rearrange the symbols to better reflect this. For example, use a different symbol from near-but-not significant results (e.g., “‡” for p < .05, < .01). Then put one asterisk (*) for p < .002 (because this is the significance cut-off) and two asterisks (**) for p < .001.

 

- My major concern is the mediation analysis. Have the authors tried running a model where there is no mediators included (only the IVs and DV)? If there is no significant relationship between the IVs and the DV, then there is no point of testing a mediation (you don’t have a “relationship” to mediate in the first place). Of course, suppression effect is still possible, but with a correlational study like this, I wouldn't test mediation if there is no relationship between IVs and DV. It is VERY easy to find statistically significant mediation effects after throwing some mediators in the model (especially if there is a strong relationship between the mediator and the DV). Too many correlational studies have abused mediation analysis this way. Although the authors stated that no causation relationships could be implied, the nature of SEM makes it easy to mistake the path models as causational. I would be very careful when testing mediation with a correlational study. So, at least please show us that there is a relationship to mediate in the first place (i.e., relationship between (1) parent-child discussion and casual sex intention, (2) peer interaction and casual sex intension, and (3) exposure to explicit material and casual sex intension).

 

- Another major concern is the authors didn’t report the indirect effect test. How did they even know the mediations exist then? It’s possible X is significantly related to M, and M is significantly related to Y, but still there is no mediating relationship among X, M, and Y. I cannot take these SEM figures seriously without seeing the results of the indirect effects.

 

Discussion

 

- Showing correlations between variables is important, but instead of making conclusions based on correlational results, why don’t the authors run a multiple regression to test the relationship between IV and DV controlling for the other IVs (especially when SEM show non-significant results for all main IVs)? I’m asking this because the authors talk about correlational results in their discussion. Multiple regression would be more appropriate.

 

- I feel some of the summaries in the discussion is a bit misleading. Like they talked as if the main IVs predict DV (but in fact it’s the mediator). I’d say run the model without the mediators, and if there is enough justification to run a mediation, then test the indirect effect. After that, please consider if the summary needs to be rewritten (e.g., p.10 line 315-317).

 

General comment

 

- Why didn't the authors look at the effect of age too? Like what they did with the gender differences. Casual sex can have very different meanings for high-school-age adolescents and college-age young adults. High school students and college-age adults may undergo different sexual developmental milestones that could impact intention to have casual sex. It also may be less socially acceptable for adolescents to have casual sex (compared to young adults). I understand that not a lot of high school students may want to engage in casual sex so there may be a problem with power. But would it be possible to at least compare between people aged less than 20 and people aged 20 or older?

 

- Using intention to have casual sex may be challenging because it implies that casual sex is planned and intentional. For a lot of people (especially young people), casual sex is situational (e.g., in a party, bar, club where alcohol is involved). People who intend to do it may not actually do it, while people who don’t intend to do it may end up doing it a lot. I know that measuring real behavior may not be ideal either given the young age and cultural restraints. Anyway, the authors should mention this in the limitation and future direction section.

Author Response

We appreciated your valuable comments. As discussed below, we have revised our manuscript with underlines based on your suggestions. Please let us know if we need to provide anything else regarding this revision.

 

Comment 1 

Methods

 -p.4 “participants” – I understand that the study took place in Taiwan but can the authors state this again in this section?

Response

Thanks for your suggestion. We have added “Taiwan” in this section. Please refer to line 148.

 

Comment 2

- p.4, line 146-148 – Some explanation may be needed for why students younger than 20 need parental consent. Is this because of the law? Are people younger than 20 considered a "minor" in Taiwan? This can be different from many countries where people aged 18+ are not considered "adolescent" anymore (and thus no parental consent is needed). The authors should explicitly mention this, so the reader know who are considered “adolescents” and who are considered “young adults” in the context of this study.

Response

Participants younger than 20 years old were required to obtain written parental consent to participate into the studies because they are legally considered as the minors in Taiwan. We added the explanation as below into the revised manuscript. Please refer to line 154-156.

“Participants younger than 20 years old were required to obtain written parental consent to participate into the studies because they are legally considered as the minors in Taiwan.”

 

Comment 3

Results

- what is the age range?

Response

We added the range and mean of age as below in the Demographic Characteristics. Please refer to line 269-270.

“…age ranged between 15 and 24 years, with mean age of 17.38 years (standard deviation: 2.26).”

 

Comment 4

- Table 2 is confusing and misleading even. First, please number all rows (1, 2, 3, 4, etc.) so it’s easier to see which variable correlates with which variable. Second, at first glance it seems like ALL correlations (except the correlation between attitude and parent-child discussion) are significant because they all have asterisks next to them. The note under the table also doesn’t helpful at all because it still makes it feel like they all indicate significance. Then I realized that it needs to be <.002 to be considered significant. Please rearrange the symbols to better reflect this. For example, use a different symbol from near-but-not significant results (e.g., “‡” for p < .05, < .01). Then put one asterisk (*) for p < .002 (because this is the significance cut-off) and two asterisks (**) for p < .001.

Response

Thank you for your suggestion. In the revised Table 2 we numbered all rows and rearranged the symbols to reflect various levels of significances. Please refer to line 307.

 

Comment 5

- My major concern is the mediation analysis. Have the authors tried running a model where there is no mediators included (only the IVs and DV)? If there is no significant relationship between the IVs and the DV, then there is no point of testing a mediation (you don’t have a “relationship” to mediate in the first place). Of course, suppression effect is still possible, but with a correlational study like this, I wouldn't test mediation if there is no relationship between IVs and DV. It is VERY easy to find statistically significant mediation effects after throwing some mediators in the model (especially if there is a strong relationship between the mediator and the DV). Too many correlational studies have abused mediation analysis this way. Although the authors stated that no causation relationships could be implied, the nature of SEM makes it easy to mistake the path models as causational. I would be very careful when testing mediation with a correlational study. So, at least please show us that there is a relationship to mediate in the first place (i.e., relationship between (1) parent-child discussion and casual sex intention, (2) peer interaction and casual sex intension, and (3) exposure to explicit material and casual sex intension).

Response

Thank you for your comment. We reanalyzed the data based on your suggestions. Firstly, we examine the associations of parent–child discussion, peer interaction, and exposure to sexually explicit materials with casual sex intention using multiple regression analysis. The results support that after controlling for the effects of gender and age, parent–child discussion, peer interaction, and exposure to sexually explicit materials were significantly associated with the intention to engage in casual sex. Secondly, we used SEM to examine the mediation effects of the three TPB constructs on the associations. We rewrote the contents of Abstract, Statistical Analysis, and Results as below based on the results of analysis.

1. Abstract

“The results of multiple regression analysis revealed that parent–child discussions on sex issues, peer interactions around sexual issues, and exposure to sexually explicit materials were significantly associated with the intention to engage in casual sex. The results of structural equation modeling (SEM) further supported that favorable attitude, perceiving positive social norms toward casual sex, and control over involvement mediated the associations. For men, decreased favorable attitude mediated the negative association between parent–child discussions and casual sex intention; increased favorable attitudes and decreased control over involvement mediated the positive associations between peer interactions and casual sex intention. For women, decreased control over involvement mediated the positive association between exposure to sexually explicit materials and casual sex intention.” Please refer to line 30-39.

2. Statistical Analysis

“Multiple regression analysis was used to examine the associations of parent–child discussion, peer interaction, and exposure to sexually explicit materials with casual sex intention by controlling for the effects of gender and age. If parent–child discussion, peer interaction, and exposure to sexually explicit materials were significantly associated with casual sex intention, the mediation effects of the three TPB constructs on the associations were further examined using SEM performed with AMOS (v.24, SPSS Inc., Chicago, IL, USA).” Please refer to line 235-241.

3. Results

“3.4. Associations of Parent–Child Discussion, Peer Interaction, and Exposure to Sexually Explicit Materials with Casual Sex Intention

Table 3 shows the results of multiple regression analysis examining the associations of parent–child discussion, peer interaction, exposure to sexually explicit materials, sex, and age with casual sex intention. The results support that after controlling for the effects of sex and age, parent–child discussion on sexual issues was negatively associated with the intention to engage in casual sex, whereas peer interaction around sexual issues and exposure to sexually explicit materials wer positively associated with the intention to engage in casual sex.

3.5. Results of SEM

Given that parent–child discussion, peer interaction, exposure to sexually explicit materials were significantly associated with casual sex intention, the mediation effects of the TPB constructs on the associations were further examined by SEM.” Please refer to line 310-326.

 

Comment 6

- Another major concern is the authors didn’t report the indirect effect test. How did they even know the mediations exist then? It’s possible X is significantly related to M, and M is significantly related to Y, but still there is no mediating relationship among X, M, and Y. I cannot take these SEM figures seriously without seeing the results of the indirect effects.

Response

Thanks for your suggestion. We added the results of the tests for indirect effects of parent–child discussion, peer interaction, and exposure to sexually explicit materials on casual sex intention through the mediation of the three TPB constructs into the revised manuscript. We added the contents of Statistical Analysis, Results and Discussion as below based on the results of analysis.

1. Statistical Analysis

“In order to investigate the indirect effects of parent–child discussion, peer interaction, and exposure to sexually explicit materials on casual sex intention through the mediation of the three TPB constructs, we performed bias-corrected bootstrapping at a 95% confidence interval with 5,000 bootstrapping samples and calculated the confidence intervals of the lower and upper bounds; if the lower and upper bounds of the 95% bias-corrected bootstrap confidence interval did not contain zero, the paths between variables were statistically significant [35].” Please refer to line 254-260.

2. Results

“As shown in Table 4, the results of the bootstrap test confirmed the existence of a significantly positive direct effect between exposure to sexually explicit materials and casual sex intention. Moreover, there was a negative and significant mediating effect of all three TPB constructs on the association between parent-child sexual communication and casual sex intention. There were positive and significant mediating effects of all three TPB constructs on the associations of peer sexual communication and exposure sexually explicit materials with casual sex intention. The H₁ was partially supported and H₂ was supported.” Please refer to line 331-337 and Table 4.

3. Discussion

“The present study found that parent–child discussion, peer interaction, and exposure to sexually explicit materials were significantly associated with the intention to engage in casual sex among adolescents and young adults, though no direct effects were found in the SEM for men and women separately. Moreover, the mediating roles of favorable attitude, perceiving positive social norms toward casual sex, and control over involvement in the associations differed between men and women. For men, decreased favorable attitude mediated the negative association between parent–child discussions and casual sex intention; increased favorable attitudes and decreased control over involvement mediated the positive associations between peer interactions and casual sex intention. For women, decreased control over involvement mediated the positive association between exposure to sexually explicit materials and casual sex intention. Furthermore, the associations of peer interaction with intention to engage in casual sex and its related TPB constructs were stronger in male than in female adolescents and young adults.” Please refer to line 379-392.

 

Comment 7

Discussion

- Showing correlations between variables is important, but instead of making conclusions based on correlational results, why don’t the authors run a multiple regression to test the relationship between IV and DV controlling for the other IVs (especially when SEM show non-significant results for all main IVs)? I’m asking this because the authors talk about correlational results in their discussion. Multiple regression would be more appropriate.

Response

Thank you for your comment. As described in the response to Comment 5, we added the results of multiple regression analysis examining the associations of parent–child discussion, peer interaction, and exposure to sexually explicit materials with casual sex intention. As described in the response to Comment 6, we also examined the associations among independent and dependent variables and mediators.

 

Comment 8

- I feel some of the summaries in the discussion is a bit misleading. Like they talked as if the main IVs predict DV (but in fact it’s the mediator). I’d say run the model without the mediators, and if there is enough justification to run a mediation, then test the indirect effect. After that, please consider if the summary needs to be rewritten (e.g., p.10 line 315-317).

Response

Thank you for your suggestion. We rewrote the summary in the beginning of Discussion as below based on the results of new analyses.

“The present study found that parent–child discussion, peer interaction, and exposure to sexually explicit materials were significantly associated with the intention to engage in casual sex among adolescents and young adults, though no direct effects were found in the SEM for men and women separately. Moreover, the mediating roles of favorable attitude, perceiving positive social norms toward casual sex, and control over involvement in the associations differed between men and women. For men, decreased favorable attitude mediated the negative association between parent–child discussions and casual sex intention; increased favorable attitudes and decreased control over involvement mediated the positive associations between peer interactions and casual sex intention. For women, decreased control over involvement mediated the positive association between exposure to sexually explicit materials and casual sex intention. Furthermore, the associations of peer interaction with intention to engage in casual sex and its related TPB constructs were stronger in male than in female adolescents and young adults.” Please refer to line 379-392.

 

Comment 9

General comment

 - Why didn't the authors look at the effect of age too? Like what they did with the gender differences. Casual sex can have very different meanings for high-school-age adolescents and college-age young adults. High school students and college-age adults may undergo different sexual developmental milestones that could impact intention to have casual sex. It also may be less socially acceptable for adolescents to have casual sex (compared to young adults). I understand that not a lot of high school students may want to engage in casual sex so there may be a problem with power. But would it be possible to at least compare between people aged less than 20 and people aged 20 or older?

Response

Thank you for your suggestion. In the revised manuscript we added the results comparing Intention, three TPB constructs, parent–child discussion, peer interaction, and exposure to sexually explicit materials between participants aged 15-19 and aged 20-25 as below. Please refer to line 288-292 and Table 1.

“Only peer interaction around sexual issues and exposure to sexually explicit materials differed between participants aged 15-19 and 20-24. Those aged 20-24 years reported more peer interaction around sexual issues and exposure to sexually explicit materials than those aged 15-19 did.”

 

Comment 10

- Using intention to have casual sex may be challenging because it implies that casual sex is planned and intentional. For a lot of people (especially young people), casual sex is situational (e.g., in a party, bar, club where alcohol is involved). People who intend to do it may not actually do it, while people who don’t intend to do it may end up doing it a lot. I know that measuring real behavior may not be ideal either given the young age and cultural restraints. Anyway, the authors should mention this in the limitation and future direction section.

Response

Thanks for your suggestion. We have added it as one of the limitations of this study.

Please refer to line 470-474.

“Sixth, the present study examined the intention to have casual sex; however, some people might have situational casual sex (e.g., in a party, bar, or club where alcohol is involved) and might not have clear intentions to do causal sex. The discrepancy between the intention to have causal sex and real causally sexual behaviors warrants further study.”

 

Round 2

Reviewer 2 Report

I appreciate the authors' responses to my comments. In general, I'm very pleased with their responses and revision. Just some minor comments below:

• Table 3: Under the note section, please also state how the gender variable was coded (e.g., men were coded as ___ and women were coded as ___).
• Table 4: the exposure -> SN -> intention path seems to have a small effect since the CI is very close to 0. First, please increase the decimal points of the lower CI so that the number is not 0.000 (but keep the decimal points the same for the rest). If the number of this lower CI remains positive (i.e., still more than 0), I suggest mentioning the small indirect effect briefly in the discussion (so readers can take caution when interpreting the results).

 

Author Response

We appreciated your valuable comments. As discussed below, we have revised our manuscript with underlines based on your suggestions. Please let us know if we need to provide anything else regarding this revision.

 

Comment 1

Table 3: Under the note section, please also state how the gender variable was coded (e.g., men were coded as ___ and women were coded as ___).

Response

We have added the code (men =0; women = 1) for gender into Table 3.

 

Comment 2

Table 4: the exposure -> SN -> intention path seems to have a small effect since the CI is very close to 0. First, please increase the decimal points of the lower CI so that the number is not 0.000 (but keep the decimal points the same for the rest). If the number of this lower CI remains positive (i.e., still more than 0), I suggest mentioning the small indirect effect briefly in the discussion (so readers can take caution when interpreting the results).

Response

1. We have increased this decimal point to “0.0001” in Table 4.
2. We have added “the effect of subjective norm of acceptance was small” into Discussion. Please refer to line 438-439.