Disentangling Actual Peer, Perceived Peer, and Adolescent ...program2/ubafdp/SRCD... · 1. W1...

1
Disentangling Actual Peer, Perceived Peer, and Adolescent Substance Use with Latent Class and Latent Transition Analysis Matthew D. Scalco, Samuel N. Meisel, & Craig R. Colder Psychology Department, State University of New York at Buffalo BACKGROUND RESULTS Control Condition Control Condition Aim 1: Structure of the Final Model for the LCAs from W1-W3 Aim 2: Validity Tests To examine whether latent classes differed on constructs that have previously been linked to adolescent, peer, and perceived peer SU, we related the latent classes at each wave to several external criterion variables, including parent reported demographics, temperament and personality (Early Adolescent Temperament Questionnaire-Revised), peer closeness, parental SU, peer delinquency, and pubertal development. CONCLUSIONS Relationship Between Peer, Perceived Peer, and Adolescent SU (cross-sectional; LCA) 1. While some adolescents accurately perceive their peers’ SU, others over- or underestimate peer SU. A. Documenting that adolescents do not uniformly overestimate their peers’ SU is important considering that normative feedback interventions can have iatrogenic effects for individuals who underestimate actual levels of SU (Schultz et al., 2007). Relationship Between Peer, Perceived Peer, and Adolescent SU (longitudinal; LTA) 1. W1 misperception+ ALC class were more likely to transition into AP-users class at W2 than misperception-. A. Adolescents associate alcohol use with popular status (Allen et al., 2005; Balsa et al., 2011). B. Adolescents who use alcohol as well as over-perceive the alcohol use of their peers may be highly motivated to seek out peers who use to obtain the positive social benefits associated with drinking. 2. Individuals in misperception+ ALC and misperception- were equally likely to transition to the AP- user at W3. 1. AP-users class = highest risk of later adolescent SU. A. Peer group that supports and reinforces poly SU mutually over time may be instrumental in establishing more habitual patterns of SU and a peer culture which revolves around SU. B. Overestimating SU in the absence of peers who actually endorsed SU was associated with heavier rates of later SU than both AP-non-users and misperception- classes, yet less than AP- users. These results suggest that while misperceived social norms do affect later SU behavior, lack of a peer context that supports and reinforces SU results in less overall SU. C. Youth least at risk for heavy SU later in adolescence were individuals in the accurate perception- non-users and misperception- classes. 2. There were three distinct pathways to the class at highest risk for later escalations in SU: A. Adolescent and perceived peer SU precedes peer SU in late childhood (age=10-12). B. Adolescent and perceived peer SU precedes peer SU in early adolescence (age=12-14). C. Peer SU precedes adolescent SU and perceptions of peer SU in early adolescence (age=12-14). Note. PR = peer self-reported use, TR = target reported peer use, T = target’s own use, Alc =alcohol, Cig = cigarette, Mar = Marijuana, SU = substance use, AP = accurate perception, + and – reflect the direction of misperception. Values on the y-axis reflect the probability of observing peer, perceived peer, or target adolescent SU for each latent class. Percentages in parentheses reflect the proportion of sample in each class (W1 N=765; W2 N=737; W3 N=715). There was no target or peer self reported marijuana use at W1 and hence these variables were not included in the analysis at W1 and can therefore be assumed to have a probability of 0. Note. SES = socioeconomic status, EC = effortful control, Peer Close = Peer Closeness Rating, SU = substance use, Peer Delin = perceived peer delinquency, Pubertal = pubertal status, and v = versus. The second number at the top of each column (after the “v”) is the referent class for that column. Values in table are odds ratios. Bolded values fall in the range of small effect sizes, while bolded and italicized values fall in the range of medium effect sizes (Cohen, 1988). The following results were consistent across Waves: 1. High levels of surgency and frustration, low levels of effortful control and shyness, and high levels of peer delinquency and pubertal development were associated with the AP-users class when compared to AP-non-users. 2. Misperception+ youth tended to have higher levels of peer delinquency and sensation seeking than AP-non-users and misperception- youth, but generally lower levels on these factors than AP-users. 3. Misperception- youth were not different than AP-non-users, although they did have lower levels of peer closeness than youth in misperception+ classes and AP-users. 4. AP-users tended to be higher than other use groups on peer closeness, frustration, and peer delinquency. 5. Overall, these results provide support for the validity of the latent classes and help with interpretation of subsequent results. Note. All values significant unless marked by ns, non-significant. The overall Χ 2 for mean differences on total frequency and quantity of SU from W4-W6 was significant. AP = accurate perception, Misper = misperception, ALC = alcohol, and SU = substance use. Values in matrices are Cohen’s d with mean (M) frequency and quantity for each class below ds. Above diagnoal reflects ds for W4-W6 total frequency comparisons, while below the diagonal reflects ds for W4-W6 total average quantity comparisons. Italicized values fall in the range of small effect sizes, bolded values in the medium effect size range, and bold and italicized values in the large effect size range (Cohen, 1988). Social norms are strong influences of adolescent behavior (Collins & Steinberg, 2006), including drug and alcohol use (Borsari & Carey, 2001). They influence behavior by providing informal rules and standards that guide behavior, and communicate the frequency and acceptability of various behaviors. Descriptive norms, perceptions of the extent to which peers engage in a behavior, such as peer substance use (SU), are consistent determinants of later adolescent SU (Elek et al., 2006). Despite widespread support for the relation between descriptive norms and adolescent SU, this literature suffers three noteworthy gaps. First, studies often fail to include actual peer SU, making it difficult to gauge the degree to which adolescents misperceive peer behavior. It is important to understand misperception and it’s development because evidence suggests that adolescents who engage in SU tend to overestimate peer SU (misperception +) while adolescents who do not use tend to underestimate peer SU (misperception -; Henry et al., 2011). Second, Pape (2012) argued that misperception may be an exaggerated phenomenon and be largely a consequence of the methodology used to assess social norms. That is, Pape (2012) noted that using close friends as the referent (versus same aged peers) when asking adolescents about peer SU behavior may reduce the degree of misperception. Third, studies have not considered the emergence of misperceptions of norms over time. Understanding how these perceptions develop and influence use is important for designing effective social norms interventions. Our study addressed these gaps using latent class analysis (LCA) of adolescent SU, peer SU, and perceptions of peer SU at three waves to model the different response profiles across these variables as suggested by prior research (Henry et al., 2011). How perceptions develop and influence use over time were then tested using latent transition analysis (LTA). Social Norms, Actual Peer Substance Use, and Adolescent Substance Use Aim 1: Test cross-sectional structure of peer, perceived peer, and adolescent SU at Waves (W) 1-3. We predicted that class structure at each wave would replicate Henry et al., (2011) with 4-classes: Aims and Study Hypotheses 1. Latent Class Analyses (LCA) were conducted on W1-W3 peer, perceived peer, and adolescent SU. 2. Youth were assigned to most likely class membership and the classes were related to known correlates to test the validity of the LCA. A. Given the large number of tests, odds ratios were computed for effect size and cutoffs labeled small (ORs > 1.5 but < 2.5 or ORs < 0.67 but > 0.4), medium (ORs > 2.5 but < 4.3 or ORs < 0.4 but > 0.23), or large (ORs > 4.3 or ORs < 0.23) were used to evaluate effects instead of using p-values. 3. Latent Transition Analysis (LTA) was estimated across the classes. A. Prior to estimating LTA, measurement invariance was tested using nested log-likelihood tests (Collins & Lanza, 2010). Originally, gender and age were included as covariates but gender did not predict class membership at any wave and was excluded in the final model. 4. Test whether the latent classes differ on W4-W6 total frequency and quantity of SU. A. Wald tests of mean equality (pseudo class method) were used to test for mean differences and Cohen’s d was calculated for effect size. METHOD Participants were drawn from two six wave longitudinal studies designed to test risk and protective factors for adolescent SU. Inclusion criteria were a 10 to 12 year-old child at the time of recruitment with no language or physical disabilities that would preclude participation. The community samples included 387 and 378 families for a total of 765 families. The mean age for adolescents at the first assessment was 11.8 (SD = .79). Subsequent assessments occurred annually. At W1-W3 for both samples, target adolescents provided the names of four close friends and one was recruited into the study (close friend peer) to provide collateral reports. Peers were required to be within two years of age of the target adolescent and could not be a sibling. Attrition rate by W3 was 6.5% and completers vs. non-completers did not differ on any demographic or study variables. Measures Adolescent and Peer SU. Items taken from The National Youth Survey (NYS) were used to assess self-reported lifetime and past year alcohol, tobacco, and marijuana use (Elliott & Huizinga, 1983) at each wave of assessment. Lifetime use of alcohol without parental permission, cigarettes, and marijuana were each assessed with a dichotomous item at W1-W3 (see Table 2). Higher rates of target SU were observed between W4-W6 and quantity and frequency indices were summed across W4-W6, respectively and then summed across substances to represent total frequency of SU from W4-W6 (M = 37.57; SD = 75.28; range = 0 – 281; skew = 2.34; kurtosis = 4.37) and total quantity of average SU per occasion (M = 1.62; SD = 2.35; range = 0 – 9.64; skew = 1.77; kurtosis = 2.64). Data Analytic Plan This research was supported by two grants from NIDA (R01 DA020171 and R01 DA019631) awarded to Craig Colder. The content of this poster is solely the responsibility of the authors and does not necessarily represent the official views of NIDA. Correspondence should be addressed to Matthew Scalco, B.A., Psychology Department, Park Hall, University at Buffalo, State University of New York, Buffalo, NY 14260. Email: [email protected] Aim 3: Estimated latent transitions between the latent classes. Hypothesized Transitions: Accurate perception (AP)-users and AP-non-users would be highly stable. Probability of transitioning into the AP-users class would be highest in the misperception+ class (selection). Followed by a significant but smaller chance of the misperception- class (socialization). Finally AP-non-users were hypothesized to be least likely to transition to the AP-users class. Aim 2: Test Validity of the latent structure at each wave using known correlates of peer, perceived peer, and adolescent SU. Aim 4: Test whether latent classes differ on later adolescent SU (W4-6). We hypothesized that all mean differences would be significant and would rank from highest to lowest in the following order: AP-users, misperception+, AP-non-users, and misperception-. Perceived Peer SU. Target report of peer SU was assessed using 3 items taken from Fergusson, Woodward, and Howard (1999). The instructions asked the adolescent to “Tell whether or not any of your three close friends have ever used” and then assessed alcohol, tobacco, and marijuana. Responses were keyed as lifetime use (1) or no lifetime use (0). Rates of perceived peer SU can be found in Table 2. Note. P = Peer self-report, T = Target self-report and TR = Target Report of peer use. Percentages represent proportion of youth who endorsed initiation or perception of peer initiation of alcohol, tobacco, and marijuana use form W1-W3. Number in parentheses is the N who endorsed initiation or perception of peer initiation. Wave 1 Wave 2 Wave 3 P Alcohol 10% (70) 17% (112) 28% (174) P Tobacco 4% (25) 8% (51) 11% (69) P Marijuana 0% (0) 5% (34) 10% (61) T Alcohol 6% (43) 16% (114) 29% (208) T Tobacco 3% (22) 6% (43) 11% (47) T Marijuana 0% (0) 3% (19) 9% (63) TR P Alcohol 10% (79) 21% (153) 31% (225) TR P Tobacco 5% (40) 10% (74) 18% (132) TR P Marijuana 2% (12) 7% (55) 18% (127) Wave 1 (W1) 1 v 4 2 v 4 3 v 4 1 v 3 2 v 3 1 v 2 Age 3.04 1.44 2.25 1.42 0.67 2.12 Gender 0.90 0.75 1.80 0.50 0.42 1.20 SES 1.04 0.75 0.71 1.46 1.06 1.38 Surgency 1.42 1.48 2.10 0.68 0.71 0.96 Shyness 1.44 0.93 2.01 0.72 0.46 1.54 Affiliation 1.01 0.79 0.87 1.15 0.91 1.27 EC 0.70 0.87 0.85 0.83 1.03 0.81 Fear 0.94 0.97 1.22 0.77 0.79 0.97 Frustration 1.61 1.47 1.24 1.30 1.18 1.10 Peer Close 1.73 1.07 1.62 1.07 0.66 1.62 Parent SU 1.25 1.24 1.28 0.98 0.98 1.00 Peer Delin 1.57 1.21 2.35 0.69 0.51 1.30 Pubertal 1.48 1.06 1.38 1.07 0.76 1.40 Wave 2 (W2) 1 v 4 2 v 4 3 v 4 1 v 3 2 v 3 1 v 2 Age 2.14 3.01 3.10 0.69 0.97 0.71 Gender 1.24 0.70 1.52 0.82 0.46 1.76 SES 0.79 0.53 0.66 1.20 0.81 1.48 Surgency 1.22 0.84 1.60 0.76 0.52 1.45 Shyness 1.28 0.86 1.61 0.79 0.54 1.48 Affiliation 0.70 0.64 0.77 0.91 0.84 1.09 EC 0.74 0.81 0.57 1.29 1.41 0.92 Fear 0.81 0.89 0.71 1.15 1.26 0.91 Frustration 1.37 1.11 1.56 0.88 0.72 1.23 Peer Close 1.16 0.75 1.18 0.98 0.64 1.54 Parent SU 1.13 1.15 1.29 0.88 0.89 0.99 Peer Delin 2.01 1.23 2.72 0.74 0.45 1.62 Pubertal 1.38 1.46 1.60 0.86 0.91 0.95 Wave 3 (W3) 1 v 5 2 v 5 3 v 5 4 v 5 1 v 4 2 v 4 3 v 4 1 v 3 2 v 3 1 v 2 Age 1.35 2.52 3.27 2.57 0.53 0.98 1.27 0.41 0.77 0.54 Gender 1.23 0.81 1.42 1.93 0.64 0.42 0.74 0.86 0.57 1.52 SES 0.90 0.62 0.59 0.80 1.13 0.78 0.73 1.54 1.06 1.45 Surgency 1.25 0.89 1.70 1.84 0.68 0.49 0.92 0.74 0.53 1.40 Shyness 1.04 1.06 1.38 1.32 0.79 0.80 1.04 0.76 0.77 0.98 Affiliation 0.90 0.58 0.85 0.72 1.25 0.81 1.17 1.07 0.69 1.55 EC 0.64 0.75 0.52 0.99 0.65 0.76 0.53 1.22 1.44 0.85 Fear 0.92 1.20 1.12 0.60 1.52 1.99 1.86 0.82 1.07 0.77 Frustration 1.23 1.07 1.87 2.42 0.51 0.44 0.77 0.66 0.57 1.16 Peer Close 0.82 0.90 1.35 2.49 0.33 0.36 0.54 0.61 0.67 0.91 Parent SU 1.13 1.12 1.24 1.46 0.77 0.77 0.85 0.91 0.91 1.00 Peer Delin 2.04 2.07 3.04 2.22 0.92 0.48 1.37 0.67 0.35 1.91 Pubertal 1.29 1.08 1.53 1.27 1.01 0.85 1.21 0.84 0.71 1.19 Aim 2: Summary of Validity Results W1 Classes 1. 2. 3. 4. - 1. AP-Non-Users - 0.30 0.72 0.54 - 2. Misper- 0.31 - 0.32 ns 0.19 ns - 3. Misper+ ALC 0.84 0.45 - -0.14 ns - 4. Misper+ SU 0.67 0.31 ns -0.15 ns - - M Frequency 30.58 52.54 83.51 69.93 - M Quantity 3.61 5.43 8.55 7.49 - W2 Classes 1. 2. 3. 4. - 1. AP-Non-Users - 0.31 ns 0.73 1.62 - 2. Misper- 0.31 ns - 0.32 * 0.88 - 3. Misper+ ALC 1.01 0.58 - 0.58 - 4. AP-Users 1.65 1.22 0.48 - - M Frequency 24.53 44.8 75.04 134.98 - M Quantity 3.02 4.66 8.62 11.85 - W3 Classes 1. 2. 3. 4. 5. 1. AP-Non-Users - -0.04 ns 0.52 1.52 1.56 2. Misper- 0.19 ns - 0.42 * 0.99 0.99 3. Misper+ ALC 0.95 0.55 - 0.63 0.71 4. Misper + SU 1.72 1.02 0.37 - 0.11 ns 5. AP-Users 2.04 1.30 0.61 0.24 ns - M Frequency 16.88 14.93 47.47 105.4 117.67 M Quantity 2.12 2.87 6.67 9.35 10.98 Aim 4: Matrix of effect sizes (Cohen d) for Pairwise Mean Comparisons with Frequency above and Quantity below the Diagonal and Augmented with Latent Class Means. Wave 2 Wave 1 AP non-users Misper - Misper + ALC - AP-users AP non-users 0.84 0.06 0.09 - 0.01 Misper - 0.41 0.38 0.16 - 0.05 Misper + ALC 0.00 0.00 0.61 - 0.39 Misper + SU 0.00 0.00 0.65 - 0.36 Wave 3 Wave 2 AP non-users Misper - Misper + ALC Misper + SU AP-users AP non-users 0.74 0.03 0.19 0.02 0.02 Misper - 0.05 0.48 0.20 0.03 0.24 Misper + ALC 0.00 0.00 0.57 0.15 0.28 AP-users 0.00 0.00 0.05 0.06 0.89 Aim 3: Latent Transition Probabilities Note. AP = accurate perception, Misper = misperception, ALC = alcohol, and SU = substance use. 1. High stability for AP-non-users and even higher for AP-users. 2. Misperception- and + youth had much smaller stabilities. 3. Misperception+ youth (both ALC and SU) were more likely to transition into the AP-users class than misperception- youth from W1 to W2. 4. From W2 to W3 misperception- and + youth were near equally likely to transition into the AP-users class. Aim 3: Summary of LTA Results Aim 2: Multinomial Logistic Regression Results (Odds Ratios) Aim 4: Summary of Outcome Results 1. AP-users class had the highest levels of later SU. A. Followed by the misperception+ SU, misperception+ ALC, misperception-, and AP-non-users. 2. Effects on quantity were consistently larger than effects on frequency regardless of comparison or wave. 3. Large effects consistently occurred when comparing AP-users to both AP non-users and misperception-. 4. Small to medium sized effects occurred when comparing misperception+ to misperception- and AP-non- users, respectively. 5. Effects were medium when comparing AP-users at W2 and W3 and misperception+ SU at W3 to misperception+ ALC. University at Buffalo State University of New York University at Buffalo State University of New York Aim 1: Model Implied Response Probabilities by Latent Class AIC, BIC, Adjusted-BIC, entropy, N of classes, Lo-Mendell-Rubin test (LMR), and bootstrapped likelihood ratio test (BLRT) were all used to determine the number of classes to extract. At W1, W2, and W3, 4, 4, and 5 class solutions, respectively, were retained. Model implied response probabilities from the final model (N = 765) are plotted below. 1 = AP-non-users 2 = Misperception- 3 = Misperception+ ALC 4 = Misperception+ SU 1 = AP-non-users 2 = Misperception- 3 = Misperception+ ALC 4 = AP-users 1 = AP-non-users 2 = Misperception- 3 = Misperception+ ALC 4 = Misperception+ SU 5 = AP-users

Transcript of Disentangling Actual Peer, Perceived Peer, and Adolescent ...program2/ubafdp/SRCD... · 1. W1...

Page 1: Disentangling Actual Peer, Perceived Peer, and Adolescent ...program2/ubafdp/SRCD... · 1. W1 misperception+ ALC class were more likely to transition into AP-users class at W2 than

Disentangling Actual Peer, Perceived Peer, and Adolescent Substance Use

with Latent Class and Latent Transition Analysis Matthew D. Scalco, Samuel N. Meisel, & Craig R. Colder

Psychology Department, State University of New York at Buffalo

BACKGROUND

RESULTS

Control Condition

Control Condition

Social Exclusion Condition

Aim 1: Structure of the Final Model for the LCAs from W1-W3

Aim 2: Validity Tests To examine whether latent classes differed on constructs that have previously been linked to adolescent, peer, and perceived peer SU, we related the latent classes at each wave to several external criterion variables, including parent reported demographics, temperament and personality (Early Adolescent Temperament Questionnaire-Revised), peer closeness, parental SU, peer delinquency, and pubertal development.

CONCLUSIONS

Relationship Between Peer, Perceived Peer, and Adolescent SU (cross-sectional; LCA) 1. While some adolescents accurately perceive their peers’ SU, others over- or underestimate peer SU.

A. Documenting that adolescents do not uniformly overestimate their peers’ SU is important considering that normative feedback interventions can have iatrogenic effects for individuals who underestimate actual levels of SU (Schultz et al., 2007).

Relationship Between Peer, Perceived Peer, and Adolescent SU (longitudinal; LTA) 1. W1 misperception+ ALC class were more likely to transition into AP-users class at W2 than

misperception-. A. Adolescents associate alcohol use with popular status (Allen et al., 2005; Balsa et al., 2011). B. Adolescents who use alcohol as well as over-perceive the alcohol use of their peers may be

highly motivated to seek out peers who use to obtain the positive social benefits associated with drinking.

2. Individuals in misperception+ ALC and misperception- were equally likely to transition to the AP-

user at W3. 1. AP-users class = highest risk of later adolescent SU.

A. Peer group that supports and reinforces poly SU mutually over time may be instrumental in establishing more habitual patterns of SU and a peer culture which revolves around SU.

B. Overestimating SU in the absence of peers who actually endorsed SU was associated with heavier rates of later SU than both AP-non-users and misperception- classes, yet less than AP-users. These results suggest that while misperceived social norms do affect later SU behavior, lack of a peer context that supports and reinforces SU results in less overall SU.

C. Youth least at risk for heavy SU later in adolescence were individuals in the accurate perception-non-users and misperception- classes.

2. There were three distinct pathways to the class at highest risk for later escalations in SU:

A. Adolescent and perceived peer SU precedes peer SU in late childhood (age=10-12). B. Adolescent and perceived peer SU precedes peer SU in early adolescence (age=12-14). C. Peer SU precedes adolescent SU and perceptions of peer SU in early adolescence (age=12-14).

Note. PR = peer self-reported use, TR = target reported peer use, T = target’s own use, Alc =alcohol, Cig = cigarette, Mar = Marijuana, SU = substance use, AP = accurate perception, + and – reflect the direction of misperception. Values on the y-axis reflect the probability of observing peer, perceived peer, or target adolescent SU for each latent class. Percentages in parentheses reflect the proportion of sample in each class (W1 N=765; W2 N=737; W3 N=715). There was no target or peer self reported marijuana use at W1 and hence these variables were not included in the analysis at W1 and can therefore be assumed to have a probability of 0.

Note. SES = socioeconomic status, EC = effortful control, Peer Close = Peer Closeness Rating, SU = substance use, Peer Delin = perceived peer delinquency, Pubertal = pubertal status, and v = versus. The second number at the top of each column (after the “v”) is the referent class for that column. Values in table are odds ratios. Bolded values fall in the range of small effect sizes, while bolded and italicized values fall in the range of medium effect sizes (Cohen, 1988).

The following results were consistent across Waves: 1. High levels of surgency and frustration, low levels of effortful control and

shyness, and high levels of peer delinquency and pubertal development were associated with the AP-users class when compared to AP-non-users.

2. Misperception+ youth tended to have higher levels of peer delinquency and sensation seeking than AP-non-users and misperception- youth, but generally lower levels on these factors than AP-users.

3. Misperception- youth were not different than AP-non-users, although they did have lower levels of peer closeness than youth in misperception+ classes and AP-users.

4. AP-users tended to be higher than other use groups on peer closeness, frustration, and peer delinquency.

5. Overall, these results provide support for the validity of the latent classes and help with interpretation of subsequent results.

Note. All values significant unless marked by ns, non-significant. The overall Χ2 for mean differences on total frequency and quantity of SU from W4-W6 was significant. AP = accurate perception, Misper = misperception, ALC = alcohol, and SU = substance use. Values in matrices are Cohen’s d with mean (M) frequency and quantity for each class below ds. Above diagnoal reflects ds for W4-W6 total frequency comparisons, while below the diagonal reflects ds for W4-W6 total average quantity comparisons. Italicized values fall in the range of small effect sizes, bolded values in the medium effect size range, and bold and italicized values in the large effect size range (Cohen, 1988).

Social norms are strong influences of adolescent behavior (Collins & Steinberg, 2006), including drug and alcohol use (Borsari & Carey, 2001). They influence behavior by providing informal rules and standards that guide behavior, and communicate the frequency and acceptability of various behaviors. Descriptive norms, perceptions of the extent to which peers engage in a behavior, such as peer substance use (SU), are consistent determinants of later adolescent SU (Elek et al., 2006). Despite widespread support for the relation between descriptive norms and adolescent SU, this literature suffers three noteworthy gaps. First, studies often fail to include actual peer SU, making it difficult to gauge the degree to which adolescents misperceive peer behavior. It is important to understand misperception and it’s development because evidence suggests that adolescents who engage in SU tend to overestimate peer SU (misperception +) while adolescents who do not use tend to underestimate peer SU (misperception -; Henry et al., 2011). Second, Pape (2012) argued that misperception may be an exaggerated phenomenon and be largely a consequence of the methodology used to assess social norms. That is, Pape (2012) noted that using close friends as the referent (versus same aged peers) when asking adolescents about peer SU behavior may reduce the degree of misperception. Third, studies have not considered the emergence of misperceptions of norms over time. Understanding how these perceptions develop and influence use is important for designing effective social norms interventions. Our study addressed these gaps using latent class analysis (LCA) of adolescent SU, peer SU, and perceptions of peer SU at three waves to model the different response profiles across these variables as suggested by prior research (Henry et al., 2011). How perceptions develop and influence use over time were then tested using latent transition analysis (LTA).

Social Norms, Actual Peer Substance Use, and Adolescent Substance Use

Aim 1: Test cross-sectional structure of peer, perceived peer, and adolescent SU at Waves (W) 1-3. We predicted that class structure at each wave would replicate Henry et al., (2011) with 4-classes:

Aims and Study Hypotheses

1. Latent Class Analyses (LCA) were conducted on W1-W3 peer, perceived peer, and adolescent SU. 2. Youth were assigned to most likely class membership and the classes were related to known correlates to test the validity of the LCA.

A. Given the large number of tests, odds ratios were computed for effect size and cutoffs labeled small (ORs > 1.5 but < 2.5 or ORs < 0.67 but > 0.4), medium (ORs > 2.5 but < 4.3 or ORs < 0.4 but > 0.23), or large (ORs > 4.3 or ORs < 0.23) were used to evaluate effects instead of using p-values.

3. Latent Transition Analysis (LTA) was estimated across the classes. A. Prior to estimating LTA, measurement invariance was tested using nested log-likelihood tests (Collins & Lanza, 2010). Originally, gender

and age were included as covariates but gender did not predict class membership at any wave and was excluded in the final model. 4. Test whether the latent classes differ on W4-W6 total frequency and quantity of SU.

A. Wald tests of mean equality (pseudo class method) were used to test for mean differences and Cohen’s d was calculated for effect size.

METHOD

Participants were drawn from two six wave longitudinal studies designed to test risk and protective factors for adolescent SU. Inclusion criteria were a 10 to 12 year-old child at the time of recruitment with no language or physical disabilities that would preclude participation. The community samples included 387 and 378 families for a total of 765 families. The mean age for adolescents at the first assessment was 11.8 (SD = .79). Subsequent assessments occurred annually. At W1-W3 for both samples, target adolescents provided the names of four close friends and one was recruited into the study (close friend peer) to provide collateral reports. Peers were required to be within two years of age of the target adolescent and could not be a sibling. Attrition rate by W3 was 6.5% and completers vs. non-completers did not differ on any demographic or study variables.

Measures Adolescent and Peer SU. Items taken from The National Youth Survey (NYS) were used to assess self-reported lifetime and past year alcohol, tobacco, and marijuana use (Elliott & Huizinga, 1983) at each wave of assessment. Lifetime use of alcohol without parental permission, cigarettes, and marijuana were each assessed with a dichotomous item at W1-W3 (see Table 2). Higher rates of target SU were observed between W4-W6 and quantity and frequency indices were summed across W4-W6, respectively and then summed across substances to represent total frequency of SU from W4-W6 (M = 37.57; SD = 75.28; range = 0 – 281; skew = 2.34; kurtosis = 4.37) and total quantity of average SU per occasion (M = 1.62; SD = 2.35; range = 0 – 9.64; skew = 1.77; kurtosis = 2.64).

Data Analytic Plan

This research was supported by two grants from NIDA (R01 DA020171 and R01 DA019631) awarded to Craig Colder. The content of this poster is solely the responsibility of the authors and does not necessarily represent the official views of NIDA. Correspondence should be addressed to Matthew Scalco, B.A., Psychology Department, Park Hall, University at Buffalo, State University of New York, Buffalo, NY 14260. Email: [email protected]

Aim 3: Estimated latent transitions between the latent classes. Hypothesized Transitions: • Accurate perception (AP)-users and AP-non-users would be highly stable. • Probability of transitioning into the AP-users class would be highest in the misperception+ class (selection).

• Followed by a significant but smaller chance of the misperception- class (socialization). • Finally AP-non-users were hypothesized to be least likely to transition to the AP-users class.

Aim 2: Test Validity of the latent structure at each wave using known correlates of peer, perceived peer, and adolescent SU.

Aim 4: Test whether latent classes differ on later adolescent SU (W4-6). We hypothesized that all mean differences would be significant and would rank from highest to lowest in the following order: • AP-users, misperception+, AP-non-users, and misperception-.

Perceived Peer SU. Target report of peer SU was assessed using 3 items taken from Fergusson, Woodward, and Howard (1999). The instructions asked the adolescent to “Tell whether or not any of your three close friends have ever used” and then assessed alcohol, tobacco, and marijuana. Responses were keyed as lifetime use (1) or no lifetime use (0). Rates of perceived peer SU can be found in Table 2.

Note. P = Peer self-report, T = Target self-report and TR = Target Report of peer use. Percentages represent proportion of youth who endorsed initiation or perception of peer initiation of alcohol, tobacco, and marijuana use form W1-W3. Number in parentheses is the N who endorsed initiation or perception of peer initiation.

Wave 1 Wave 2 Wave 3 P Alcohol 10% (70) 17% (112) 28% (174) P Tobacco 4% (25) 8% (51) 11% (69) P Marijuana 0% (0) 5% (34) 10% (61) T Alcohol 6% (43) 16% (114) 29% (208) T Tobacco 3% (22) 6% (43) 11% (47) T Marijuana 0% (0) 3% (19) 9% (63) TR P Alcohol 10% (79) 21% (153) 31% (225) TR P Tobacco 5% (40) 10% (74) 18% (132) TR P Marijuana 2% (12) 7% (55) 18% (127)

Wave 1 (W1) 1 v 4 2 v 4 3 v 4 1 v 3 2 v 3 1 v 2

Age 3.04 1.44 2.25 1.42 0.67 2.12

Gender 0.90 0.75 1.80 0.50 0.42 1.20

SES 1.04 0.75 0.71 1.46 1.06 1.38

Surgency 1.42 1.48 2.10 0.68 0.71 0.96

Shyness 1.44 0.93 2.01 0.72 0.46 1.54

Affiliation 1.01 0.79 0.87 1.15 0.91 1.27

EC 0.70 0.87 0.85 0.83 1.03 0.81

Fear 0.94 0.97 1.22 0.77 0.79 0.97

Frustration 1.61 1.47 1.24 1.30 1.18 1.10

Peer Close 1.73 1.07 1.62 1.07 0.66 1.62

Parent SU 1.25 1.24 1.28 0.98 0.98 1.00

Peer Delin 1.57 1.21 2.35 0.69 0.51 1.30

Pubertal 1.48 1.06 1.38 1.07 0.76 1.40 Wave 2 (W2)

1 v 4 2 v 4 3 v 4 1 v 3 2 v 3 1 v 2

Age 2.14 3.01 3.10 0.69 0.97 0.71

Gender 1.24 0.70 1.52 0.82 0.46 1.76

SES 0.79 0.53 0.66 1.20 0.81 1.48

Surgency 1.22 0.84 1.60 0.76 0.52 1.45

Shyness 1.28 0.86 1.61 0.79 0.54 1.48

Affiliation 0.70 0.64 0.77 0.91 0.84 1.09

EC 0.74 0.81 0.57 1.29 1.41 0.92

Fear 0.81 0.89 0.71 1.15 1.26 0.91

Frustration 1.37 1.11 1.56 0.88 0.72 1.23

Peer Close 1.16 0.75 1.18 0.98 0.64 1.54

Parent SU 1.13 1.15 1.29 0.88 0.89 0.99

Peer Delin 2.01 1.23 2.72 0.74 0.45 1.62

Pubertal 1.38 1.46 1.60 0.86 0.91 0.95

Wave 3 (W3) 1 v 5 2 v 5 3 v 5 4 v 5 1 v 4 2 v 4 3 v 4 1 v 3 2 v 3 1 v 2

Age 1.35 2.52 3.27 2.57 0.53 0.98 1.27 0.41 0.77 0.54

Gender 1.23 0.81 1.42 1.93 0.64 0.42 0.74 0.86 0.57 1.52

SES 0.90 0.62 0.59 0.80 1.13 0.78 0.73 1.54 1.06 1.45

Surgency 1.25 0.89 1.70 1.84 0.68 0.49 0.92 0.74 0.53 1.40

Shyness 1.04 1.06 1.38 1.32 0.79 0.80 1.04 0.76 0.77 0.98

Affiliation 0.90 0.58 0.85 0.72 1.25 0.81 1.17 1.07 0.69 1.55

EC 0.64 0.75 0.52 0.99 0.65 0.76 0.53 1.22 1.44 0.85

Fear 0.92 1.20 1.12 0.60 1.52 1.99 1.86 0.82 1.07 0.77

Frustration 1.23 1.07 1.87 2.42 0.51 0.44 0.77 0.66 0.57 1.16

Peer Close 0.82 0.90 1.35 2.49 0.33 0.36 0.54 0.61 0.67 0.91

Parent SU 1.13 1.12 1.24 1.46 0.77 0.77 0.85 0.91 0.91 1.00

Peer Delin 2.04 2.07 3.04 2.22 0.92 0.48 1.37 0.67 0.35 1.91

Pubertal 1.29 1.08 1.53 1.27 1.01 0.85 1.21 0.84 0.71 1.19

Aim 2: Summary of Validity Results

W1 Classes 1. 2. 3. 4. - 1. AP-Non-Users - 0.30 0.72 0.54 - 2. Misper- 0.31 - 0.32ns 0.19ns - 3. Misper+ ALC 0.84 0.45 - -0.14ns - 4. Misper+ SU 0.67 0.31ns -0.15ns - -

M Frequency 30.58 52.54 83.51 69.93 - M Quantity 3.61 5.43 8.55 7.49 -

W2 Classes 1. 2. 3. 4. -

1. AP-Non-Users - 0.31ns 0.73 1.62 - 2. Misper- 0.31ns - 0.32* 0.88 - 3. Misper+ ALC 1.01 0.58 - 0.58 - 4. AP-Users 1.65 1.22 0.48 - -

M Frequency 24.53 44.8 75.04 134.98 - M Quantity 3.02 4.66 8.62 11.85 -

W3 Classes 1. 2. 3. 4. 5.

1. AP-Non-Users - -0.04ns 0.52 1.52 1.56 2. Misper- 0.19ns - 0.42* 0.99 0.99 3. Misper+ ALC 0.95 0.55 - 0.63 0.71 4. Misper + SU 1.72 1.02 0.37 - 0.11ns 5. AP-Users 2.04 1.30 0.61 0.24ns -

M Frequency 16.88 14.93 47.47 105.4 117.67 M Quantity 2.12 2.87 6.67 9.35 10.98

Aim 4: Matrix of effect sizes (Cohen d) for Pairwise Mean Comparisons with Frequency above and Quantity below the Diagonal and Augmented with Latent Class Means.

Wave 2

Wave 1 AP non-users Misper - Misper + ALC - AP-users

AP non-users 0.84 0.06 0.09 - 0.01

Misper - 0.41 0.38 0.16 - 0.05

Misper + ALC 0.00 0.00 0.61 - 0.39

Misper + SU 0.00 0.00 0.65 - 0.36

Wave 3

Wave 2 AP non-users Misper - Misper + ALC Misper + SU AP-users

AP non-users 0.74 0.03 0.19 0.02 0.02

Misper - 0.05 0.48 0.20 0.03 0.24

Misper + ALC 0.00 0.00 0.57 0.15 0.28

AP-users 0.00 0.00 0.05 0.06 0.89

Aim 3: Latent Transition Probabilities

Note. AP = accurate perception, Misper = misperception, ALC = alcohol, and SU = substance use.

1. High stability for AP-non-users and even higher for AP-users. 2. Misperception- and + youth had much smaller stabilities. 3. Misperception+ youth (both ALC and SU) were more likely to transition

into the AP-users class than misperception- youth from W1 to W2. 4. From W2 to W3 misperception- and + youth were near equally likely to

transition into the AP-users class.

Aim 3: Summary of LTA Results

Aim 2: Multinomial Logistic Regression Results (Odds Ratios)

Aim 4: Summary of Outcome Results 1. AP-users class had the highest levels of later SU.

A. Followed by the misperception+ SU, misperception+ ALC, misperception-, and AP-non-users. 2. Effects on quantity were consistently larger than effects on frequency regardless of comparison or wave. 3. Large effects consistently occurred when comparing AP-users to both AP non-users and misperception-. 4. Small to medium sized effects occurred when comparing misperception+ to misperception- and AP-non-

users, respectively. 5. Effects were medium when comparing AP-users at W2 and W3 and misperception+ SU at W3 to

misperception+ ALC.

University at Buffalo State University of New York

University at Buffalo State University of New York

Aim 1: Model Implied Response Probabilities by Latent Class

AIC, BIC, Adjusted-BIC, entropy, N of classes, Lo-Mendell-Rubin test (LMR), and bootstrapped likelihood ratio test (BLRT) were all used to determine the number of classes to extract. At W1, W2, and W3, 4, 4, and 5 class solutions, respectively, were retained. Model implied response probabilities from the final model (N = 765) are plotted below.

1 = AP-non-users 2 = Misperception- 3 = Misperception+ ALC 4 = Misperception+ SU

1 = AP-non-users 2 = Misperception- 3 = Misperception+ ALC 4 = AP-users

1 = AP-non-users 2 = Misperception- 3 = Misperception+ ALC 4 = Misperception+ SU 5 = AP-users