Zagreb Panel ------------ Given the diverse findings and the conceptual and methodological limitations in this area, we began by exploring cross-lagged models involving pornography use and psychological wellbeing over time using data from the Zagreb panel. On the basis of the current weight of evidence, we expected to find negative associations between adolescents’ pornography use and indicators of their psychological wellbeing (e.g., depression and anxiety, self-esteem, well-being). However, no firm hypotheses concerning directionality of these associations were adopted. **Method** In April of 2015, N = 2,655 2nd year high school students registered online for a 3 year longitudinal study about the effects of sexualized media on young people’s beliefs, attitudes, and behaviors. These students were recruited from 59 of 90 schools in the city of Zagreb (Croatia) and the surrounding county and were sampled from a total population of 11,768 2nd year students enrolled in that area. Data collection occurred in 6 month intervals (T1-T6). Due to a substantially different notification procedure (associated with the fact that participants finished their secondary education between T5 and T6), the data from the final wave (T6) is not directly comparable to the previous waves and was not used in this study. More panel details and a full list of measures administered to this panel can be found here: [https://osf.io/hf4k2/][1]. Our exploratory analyses made use of the following measures, all of which were presented in Croatian: **Pornography use.** The frequency of pornography use was assessed at each wave with the item, “How often have you used pornography during the last 6 months?” Scale response options included: 1 – “Not once”; 2 – “Several times”; 3 – “Once a month”; 4 – “2-3 times a month”; 5 – “Once a week”; 6 – “Several times a week”; 7 – “Every day or almost every day”; 8 – “Several times a day. ” Stability coefficients for the indicator ranged from r = .75 to r = .81 (p < .001). In the questionnaire, pornography was defined for participants as “any material which openly (i.e., in an uncensored manner) depicts sexual activity; material which shows naked bodies but not sexual intercourse or other sexual activity does not belong to pornography as here defined.” **Depression and Anxiety.** We assessed a negative dimension of psychological wellbeing with Kroenke et al.’s (2009) 4-item measure of Depression and Anxiety at waves T2, T3, and T4. Following the stem “During the last two weeks, how often have you experienced...” participants responded to items such as “Feeling down, depressed, or hopeless” and “Feeling nervous, anxious or on edge” with 4-point scales that ranged from “Not at all” (1) to “Nearly every day” (4). Cronbach’s α coefficients for the scale ranged from .85 to .86. Stability coefficients of this state measure of depression and anxiety were in the r = .56 to r = .60 range (p < .001). **Self-Esteem.** Self-esteem was one of two indicators used to assess a positive dimension of psychological wellbeing. It was measured with Marsh et al.’s (2014) 4-item inventory at waves T2, T3, and T4. Participants were asked to “Estimate how the following statements relate to you” using responses that ranged from “It does not relate to me at all” (1) to “It relates to me completely” (5). Example items included “In general, I like myself the way I am” and “When I do something, I do it well.” Cronbach’s α coefficients for the scale ranged from .81 to .85, while stability coefficients ranged from r = .68 to r = .72 (p < .001). Subjective Wellbeing . Subjective wellbeing was the other indicator that was used to assess a positive dimension of psychological health. It was measured with an adapted 4-item version of the Personal Wellbeing Inventory – School Children (PWI-SC)(Tomyn & Cummins, 2011) at waves T3 and T5. Participants were asked to indicate how satisfied they were with various facets of their life, such as their health, and what they had achieved so far. Responses were collected with 10-point scales that ranged from “Completely Unsatisfied” (1) to “Completely Satisfied” (10). Reliability of this measure was satisfactory (Cronbach’s α = .80 and .81), as was its stability across time (r = .74, p < .001). Impulsiveness was also included to allow use to examine conservative tests of the associations between pornography use and psychological wellbeing. This variable was included because it has been found to be correlated with both pornography use (Wetterneck et al., 2012) and psychological wellbeing (Goodwin, Browne, Hing, & Russell, 2017), and because it seemed unlikely that impulsiveness would mediate causal links between pornography use and indicators of psychological wellbeing. **Impulsiveness.** Impulsiveness was measured with an adapted 8-item version of the Barratt Impulsiveness Scale-Brief (Steinberg et al, 2013) at T4. Example items included “I do things without thinking” and “I am future oriented.” For structural equation modeling, the items were randomly paired in parcels. Responses were collected on 4-point scales ranging from 1 – “Never / Rarely” to 4 “Almost Always / Always”. Reliability of this measure was satisfactory (Cronbach’s α = .73). **Analytic Models** Because subjective wellbeing was measured at different waves than self-esteem and depression and anxiety, 2 separate multi-group (male and female adolescents as groups) cross-lagged path analytic models were carried out. Model 1 involved pornography use, self-esteem and depression and anxiety across three time points (T2, T3, & T4) and included data from participants who participated in at least two of these three waves (male group n = 202, female group n = 446). Model 2 involved pornography use and wellbeing across two time points (T3 & T5) and included only participants who provided data at both time points (male group n = 136, female group n = 336). A third subsequent model, Model 3, was used to explore the impact of impulsiveness as a confounding variable when significant path effects emerged between pornography use and wellbeing in Model 2. In assessing the path analytic models, χ2/df ratio ≤ 2, comparative fit index (CFI) values ≥ .95 and root mean square error of approximation (RMSEA) values ≤ .05 (with the upper 90% CI value ≤ .08) indicated excellent fit to the data (Byrne, 2013). Time invariance by gender was tested in progressively more restrictive steps (from configural to strong factorial invariance; cf. Little, 2013). Taking into account large sample size, particularly in the case of female adolescents, the standard χ2 difference test was replaced with the ∆CFI test, with values ≤ .002 indicating a non-significant difference between a less and a more constrained models (Meade, Johnson, & Braddy, 2008). Using this criterion, both cross-lagged models reached weak factorial invariance, indicating that direct comparisons between male and female adolescents should be avoided or discussed with great caution due to certain measurement discrepancies across time and gender. Based on the results of Little’s test, which suggested that data was missing completely at random in both the Model 1 and Model 2 samples (18% missing, χ2 (18) = 23.55, p < .17 and 14% missing, χ2 (27) = 7.37, p < .12, respectively), full information maximum likelihood estimation was used to handle missing values (Graham, 2012). **Results** **Model 1:** Model fit was acceptable: χ2 (582) = 1173, CFI = .93, RMSEA = .040 (90% CI = .036-.043). With respect to gender invariance, only weak factorial invariance was achieved. No significant covariances or paths of interest were found for either gender. With female adolescents, the correlation between pornography use and depression/anxiety approached significance, r = 0.11, p < .06 at T4, but as the female sample was large (n = 446), this may have simply reflected chance variation. *Male estimates:* ![Zagreb - Model 1 - Males][2] *Female estimates:* ![Zagreb - Model 1 - Females][3] **Model 2.** Model fit was excellent: χ2 (99) = 158, CFI = .98, RMSEA = .036 (90% CI = .025-.046). With respect to gender invariance, weak factorial invariance was achieved. Within the male group, a significant negative directional path was found between wellbeing at T3 and pornography use at T5, β = -0.13, p < .01, and a significant negative correlation was found between wellbeing at T5 and pornography use at T5, r = - .27. Further, a significant negative directional path was found also found within the female group between wellbeing at T3 and pornography use at T5, β = -0.10, p < .01. *Male estimates:* ![Zagreb - Model 2 - Males][4] *Female estimates:* ![Zabreb - Model 2 - Females][5] **Model 3.** To further investigate the significant paths and covariances between pornography use and wellbeing, impulsiveness was added to Model 2. Model fit was excellent: χ2(91) = 133, CFI = .97, RMSEA = .035 (90% CI = .021-.035). With impulsiveness controlled for, the direct paths between wellbeing at T3 and pornography use within males dropped to, β = -0.00, p > .05, and the covariance between pornography use and well-being became marginally significant, (p < .08). Within women, controlling for impulsiveness did not reduce the magnitude of the direct path between wellbeing at T3 and pornography use at T5, β = -0.18, p < .01. *Male estimates:* ![Zagreb - Model 3 - Males][6] *Female estimates:* ![Zagreb - Model 3 - Females][7] [1]: https://osf.io/hf4k2/ [2]: https://mfr.osf.io/export?url=https://osf.io/cy93r/?action=download&mode=render&direct&public_file=False&initialWidth=723&childId=mfrIframe&parentTitle=OSF%20%7C%20Zagreb%20-%20Model%201%20-%20Males.png&parentUrl=https://osf.io/cy93r/&format=1200x1200.jpeg [3]: https://mfr.osf.io/export?url=https://osf.io/bgj4a/?action=download&mode=render&direct&public_file=False&initialWidth=723&childId=mfrIframe&parentTitle=OSF%20%7C%20Zagreb%20-%20Model%201%20-%20Females.png&parentUrl=https://osf.io/bgj4a/&format=1200x1200.jpeg [4]: https://mfr.osf.io/export?url=https://osf.io/g35nj/?action=download&mode=render&direct&public_file=False&initialWidth=723&childId=mfrIframe&parentTitle=OSF%20%7C%20Zagreb%20-%20Model%202%20-%20Males.png&parentUrl=https://osf.io/g35nj/&format=1200x1200.jpeg [5]: https://mfr.osf.io/export?url=https://osf.io/7m3xr/?action=download&mode=render&direct&public_file=False&initialWidth=723&childId=mfrIframe&parentTitle=OSF%20%7C%20Zagreb%20-%20Model%202%20-%20Females.png&parentUrl=https://osf.io/7m3xr/&format=1200x1200.jpeg [6]: https://mfr.osf.io/export?url=https://osf.io/hf5zv/?action=download&mode=render&direct&public_file=False&initialWidth=723&childId=mfrIframe&parentTitle=OSF%20%7C%20Zagreb%20-%20Model%203%20-%20Males.png&parentUrl=https://osf.io/hf5zv/&format=1200x1200.jpeg [7]: https://mfr.osf.io/export?url=https://osf.io/z4ta6/?action=download&mode=render&direct&public_file=False&initialWidth=723&childId=mfrIframe&parentTitle=OSF%20%7C%20Zagreb%20-%20Model%203%20-%20Females.png&parentUrl=https://osf.io/z4ta6/&format=1200x1200.jpeg