نگرش والدین نسبت به انحراف بعنوان پیش بینی کننده بزهکاری: ساخت اتصال از طریق فهم و شناخت

Abstract This study tested a core postulate of social cognitive theory: i.e., that perception precedes cognition in the development of behavior. Using data from four of the first five waves of the 1725-member (918 males, 807 females) National Youth Survey (NYS), youth perception of parental attitude toward deviance and youth attitude toward deviance at Waves 2 and 3 were tested as possible mediators of the relationship between Wave 1 parental attitude toward deviance and self-reported delinquency at Wave 5. The target chain was both significant and significantly stronger than the reverse chain and there was no evidence that age, race, or sex moderated this mediated relationship. These results support the presence of a chaining process in which proximal social, perceptual, and cognitive events link to distal behavioral outcomes like delinquency.

Results Preliminary analyses Descriptive statistics for and correlations between the four control variables (age, race, sex, DEL-1), one independent variable (APAD-1), four mediating variables (PPAD-2, YAD-2, PPAD-3, YAD-3), and one dependent variable (DEL-5) can be found in Table 1. Intercorrelations between the 9 predictor variables (control variables, independent variable, mediator variables) revealed no evidence of multicollinearity (Tolerance = .477–.973, Variance Inflation Factor [VIF] = 1.028–2.098). Table 1. Descriptive statistics and correlations for the ten variables included in this study. Variable N/n M SD Range 2 3 4 5 6 7 8 9 10 1. Age 1725 13.87 1.94 11–17 −.04 −.05 .06 .15* .18* .34* .27* .02 −.05 2. Race 1725 1.21 0.41 1–2 −.03 .06 .15* .08 −.07 −.14* .05 .00 3. Sex 1725 1.47 0.50 1–2 .01 −.15* −.17* −.17* −.17* −.12* −.20* 4. APAD-1 1683 10.47 2.42 9–21 .13* .10* .09* .08 .03 .05 5. PPAD-2 1654 12.98 3.40 9–35 .57* .40* .31* .05 .10* 6. PPAD-3 1625 13.31 3.44 9–32 .38* .48* .10* .11* 7. YAD-2 1655 14.32 4.35 9–36 .67* .14* .20* 8. YAD-3 1626 15.27 4.42 9–33 .14* .21* 9. DEL-1 1725 14.08 6.33 12–108 .14* 10. DEL-5 1494 13.07 3.08 12–64 Note. Age = chronological age in years; Race = 1 (White) and 2 (Nonwhite); Gender = 1 (male) and 2 (female); APAD-1 = actual parental attitude toward deviance at Wave 1; PPAD-2 = perceived parental attitude toward deviance at Wave 2; PPAD-3 = perceived parental attitude toward deviance at Wave 3; YAD-2 = youth attitude toward deviance at Wave 2; YAD-3 = youth attitude toward deviance at Wave 3; DEL-1 = self-reported delinquency at Wave 1; DEL-5 = self-reported delinquency at Wave 5; N/n = number of participants with non-missing data; M = mean, SD = standard deviation; Range = range of scores in current sample. *p < .0011 (Bonferroni-corrected alpha level; .05/45 individual comparisons). Table options Mediation analyses Moderation by age, race, and sex was tested by including a series of interactions between age, race, and sex and the independent and mediator variables in a preliminary path analysis. Only one interaction effect achieved significance in this analysis and it did not involve a relationship between an independent/mediator variable and dependent variable (the only significant effect was YAD-3 on sex x YAD-2). These results indicate that age, race, and sex failed to moderate the PPAD/YAD mediated relationship between APAD-1 and DEL-5. Table 2 lists the results of a two-stage cross-lagged mediation analysis of the parental attitude toward deviance–offspring delinquency relationship via four mediators (PPAD-2, PPAD-3, YAD-2, YAD-3) without the aforementioned interaction effects. Overall model fit ranged from fair (TLI = .91) to good (CFI = .99, RMSEA = .06 [.05–.08]). Consistent with the main hypothesis, the target pathway (APAD-1 → PPAD-2 → YAD-3 → DEL-5) was significant, the alternate or reverse pathway (APAD-1 → YAD-2 → PPAD-3 → DEL-5) was not significant, and the difference between the two indirect effects was significant using a non-zero bootstrapped 95% confidence interval as the criterion for significance (see Table 3). Table 2. Two-stage cross-lagged mediation analysis of parental attitude toward deviance as a predictor of delinquency via perceived parental and youth attitude toward deviance. Variable b (95% CI) β t p APAD-1 on Age 0.081 (0.023, 0.140) 0.065 2.75 .006 Race 0.350 (0.049, 0.652) 0.059 2.27 .023 Sex 0.085 (–0.154, 0.306) 0.018 0.72 .470 PPAD-2 on APAD-1 0.161 (0.091, 0.233) 0.115 4.46 .000 Age 0.254 (0.172, 0.334) 0.145 6.12 .000 Race 1.238 (0.796, 1.680) 0.149 5.46 .000 Sex −0.949 (−1.244, −0.628) −0.139 −6.00 .000 YAD-2 on APAD-1 0.137 (0.051, 0.222) 0.077 3.18 .001 Age 0.721 (0.624, 0.721) 0.323 14.94 .000 Race −0.690 (−1.144, −0.221) −0.065 −2.90 .004 Sex −1.372 (−1.759, −1.003) −0.158 −7.08 .000 PPAD-3 on YAD-2 0.121 (0.082, 0.164) 0.153 5.82 .000 PPAD-2 0.493 (0.434, 0.549) 0.488 16.75 .000 APAD-1 0.036 (−0.020, 0.098) 0.026 1.20 .229 Age 0.083 (0.009, 0.152) 0.047 2.26 .024 Race 0.135 (−0.239, 0.533) 0.016 0.68 .494 Sex −0.529 (−0.805, −0.263) −0.077 −3.84 .000 YAD-3 ON PPAD-2 0.081 (0.027, 0.138) 0.062 2.91 .004 YAD-2 0.619 (0.568, 0.668) 0.608 24.45 .000 APAD-1 0.036 (−0.028, 0.110) 0.020 1.02 .309 Age 0.092 (0.009, 0.176) 0.041 2.17 .030 Race −1.113(−1.534, −0.697) −0.103 −5.20 .000 Sex −0.574(−0.889, −0.251) −0.065 −3.52 .001 DEL-5 on PPAD-3 −0.010 (−0.065, 0.046) −0.012 −0.36 .716 YAD-3 0.094 (0.049, 0.139) 0.135 4.11 .000 PPAD-2 0.005 (−0.045, 0.048) 0.005 0.20 .839 YAD-2 0.082 (0.021, 0.156) 0.116 2.40 .016 APAD-1 0.050 (−0.007, 0.118) 0.040 1.60 .109 Age −0.216(−0.301, −0.149) −0.137 −5.66 .000 Race 0.036 (−0.338, 0.482) 0.005 0.17 .863 Sex −0.966 (−1.264, −0.707) −0.157 −6.85 .000 DEL-1 0.047 (0.012, 0.145) 0.096 1.26 .206 PPAD-2 with YAD-2 4.796 (4.165, 5.499) 0.367 14.08 .000 PPAD-3 with YAD-3 3.071 (2.531, 3.690) 0.345 10.58 .000 Note. APAD-1 on = regression equation with actual parental attitude toward deviance at Wave 1 as the outcome; PPAD-2 on = regression equation with perceived parental attitude toward deviance at Wave 2 as the outcome; YAD-2 on = regression equation with youth attitude toward deviance at Wave 2 as the outcome; PPAD-3 on = regression equation with perceived parental attitude toward deviance at Wave 3 as the outcome; YAD-3 on = regression equation with youth attitude toward deviance at Wave 3 as the outcome; DEL-5 on = regression equation with self-reported delinquency at Wave 5 as the outcome; PPAD-2 with YAD-2 = correlation between PPAD-2 and YAD-2; PPAD-3 with YAD-3 = correlation between PPAD-3 and YAD-3; Age = chronological age in years; Race = 1 (White) and 2 (Nonwhite); Sex = 1 (male) and 2 (female)); APAD-1 = actual parental attitude toward deviance at Wave 1; PPAD-2 = perceived parental attitude toward deviance at Wave 2; PPAD-3 = perceived parental attitude toward deviance at Wave 3; YAD-2 = youth attitude toward deviance at Wave 2; YAD-3 = youth attitude toward deviance at Wave 3; DEL-5 = self-reported delinquency at Wave 5; DEL-1 = self-reported delinquency at Wave 1; b(95% CI) = unstandardized coefficient and the lower and upper limits of the bootstrapped 95% confidence interval for the unstandardized coefficient (in parentheses); β = standardized coefficient; t = asymptotic t-test; p = significance level of the asymptotic t-test; N = 1725. Table options Table 3. Specific indirect effects and contrasts in the full sample and male and female subsamples. Standard Estimate BCBCI X-Adjusted Estimate BCBCI Total effect 0.074 (0.017, 0.143) 1.476 (0.346, 2.879) Direct effect 0.050 (−0.007, 0.118) 1.010 (−0.116, 2.384) Total indirect effect 0.023 (0.007, 0.043) 0.466 (0.158, 0.864) Specific indirect effects APAD-1 → PPAD-2 → YAD-3 → DEL-5 0.001 (0.000, 0.003) 0.024 (0.008, 0.061) APAD-1 → YAD-2 → PPAD-2 → DEL-5 0.000 (−0.001, 0.001) −0.003 (−0.027, 0.015) Preacher & Hayes Contrast Test 0.001 (0.000, 0.004) 0.028 (0.001, 0.071) Note. APAD-1 = actual parental attitude toward deviance at Wave 1; PPAD-2 = perceived parental attitude toward deviance at Wave 2; PPAD-3 = perceived parental attitude toward deviance at Wave 3; YAD-2 = youth attitude toward deviance at Wave 2; YAD-3 = youth attitude toward deviance at Wave 3; DEL-5 = self-reported delinquency at Wave 5; APAD-1 → PPAD-2 → YAD-3 → DEL-5 = predicted chain; APAD-1 → YAD-2 → PPAD-3 → DEL-5 = reverse or comparison chain; Preacher & Hayes Contrast Test = comparison between predicted and reverse chains; Standard = no adjustments made to X (APAD-1)-variable in an analysis using bootstrap resampling with replacement (b = 5000); X-Adjusted = X-variable divided by 20 in an analysis using bootstrap resampling with replacement (b = 10,000); Estimate = point estimate; BCBCI = bias-corrected bootstrapped 95% confidence interval (in parentheses). Table options It should be noted that because MPlus does not provide confidence interval data beyond the thousands decimal place the independent variable (APAD-1) had to be divided by 20 (X-adjusted) to determine whether the confidence interval included zero. The two primary criteria for chaining were satisfied in this analysis: 1. the indirect effect of the target pathway was significant and 2. the indirect effect of the target pathway was significantly stronger than the indirect effect of the reverse pathway. Because sensitivity testing can only be conducted on one mediator at a time, two separate sensitivity tests were performed using Kenny's (2013) “failsafe ef” procedure. Sensitivity testing on PPAD-2 as mediator of the APAD-1 → YAD-3 relationship revealed that an unobserved confounding covariate would need to correlate at least .28 with both the mediator (PPAD-2) and outcome (YAD-3) to completely eliminate the mediating effect of PPAD-2 on the APAD-1 → YAD-3 relationship. Sensitivity testing on YAD-3 as mediator of the PPAD-2 → DEL-5 relationship indicated that an unobserved confounding covariate would need to correlate at least .18 with both the mediator (YAD-3) and outcome (DEL-5) to completely eliminate the mediating effect of YAD-3 on the PPAD-2 → DEL-5 relationship. These results suggest that the first mediating effect was moderately robust to the effects of extraneous variables and the second mediating effect was modestly to moderately robust to the effects of extraneous variables, with the target pathway being represented by a dotted line in (Fig. 1). Standardized coefficients for the cross-lagged model in the full sample. Note. ... Fig. 1. Standardized coefficients for the cross-lagged model in the full sample. Note. Dashed lines represent the predicted path from APAD-1 to DEL-5; standardized coefficient for the PPAD-2 → PPAD-3 path = .49 and standardized coefficient for the YAD-2 → YAD-3 path = .61; paths involving control variables are not shown. *p < .05; **p < .01; ***p < .001. Figure options Comparing the target chain to a shortened chain Comparing chains of differing length can be problematic because of the loss of information that occurs each time a path is added to a chain. Nonetheless, given the fact that the PPAD-2 → YAD-3 coefficient, though significant, was smaller than the YAD-2 → PPAD-3 coefficient it could be argued that PPAD-2 added nothing to youth attitudes toward deviance in predicting Wave 5 offending. In order to evaluate this alternative hypothesis the APAD-1 → YAD-3 → DEL-5 specific indirect effect was computed. Although this shortened chain could not be compared to the full APAD-1 → PPAD-2 → YAD-3 → DEL-5 chain for reasons already stated, the shortened chain failed to achieve statistical significance in this study (X-adjusted estimate = 0.068[–0.049, 0.230]).