Table 2.
Replication Attempt 1: Total, direct, and indirect associations of age at first use on alcohol use and alcohol-related problems comparing original study, replication attempt, and extended attempt
|
Palmer et al. (2010) |
Replication attempt (restricted sample) |
Extension attempt (full sample) |
||||
| Variable | β | [95% CI] | β | [95% CI] | β | [95% CI] |
| Direct effects | ||||||
| Age at first use → PBS use | 0.26 | – | 0.15 | [0.06, 0.23] | 0.10 | [0.01, 0.20] |
| PBS use → alcohol use | -0.34 | – | -0.29 | [-0.37, -0.20] | -0.29 | [-0.37, -0.20] |
| PBS use → alcohol-related problems | -0.34 | – | -0.21 | [-0.29, -0.14] | -0.21 | [-0.29, -0.14] |
| Age at first use → alcohol use | -0.38 | – | -0.10 | [-0.04, -0.00] | -0.01 | [-0.13, 0.01] |
| Age at first use → alcohol-related problems | -0.26 | – | -0.05 | [-0.31, 0.06] | 0.02 | [-0.10, 0.17] |
| Indirect effects | ||||||
| Age at first use → PBS use alcohol use | z = 3.86 | – | -0.04 | [-0.01, -0.00] | -0.03 | [-0.01, -0.00] |
| Age at first use → PBS use → alcohol-related problems | z = 3.74 | – | -0.03 | [-0.14, -0.03] | -0.02 | [-0.08, -0.01] |
Notes: Significant associations are in bold typeface for emphasis and within the present study were determined by a 95% bias-corrected bootstrapped confidence interval that does not contain zero. z = Sobel’s z test for mediation. The restricted sample consisted of 561 college student drinkers with age at first use greater than 8 (i.e., replication attempt). The full sample consisted of 582 college student drinkers (i.e., extension attempt). For the replication and extension attempts, two independent path models were conducted in which age at onset was modeled as a predictor of alcohol outcomes (i.e., consumption and problems) via protective behavioral strategy use while age, gender, and impulsivity were controlled for. Although not shown for clarity, age at onset had a significant total association on alcohol consumption (β = -.15, 95% CI [-.05, -.01]) within the restricted sample but not for the full sample (β = -.02, 95% CI [-.02, .01]). Further, age at onset had a significant total association on alcohol-related problems (β = -.08, 95% CI [-.38, -.01]) within the restricted sample but not for the full sample (β = -.00, 95% CI [-.14, .12]). Significant associations of the covariates are available from the authors on request. CI = confidence interval; PBS = protective behavioral strategy.