Skip to main content
. Author manuscript; available in PMC: 2019 Feb 1.
Published in final edited form as: Addiction. 2017 Oct 12;113(2):299–312. doi: 10.1111/add.13964

Table 3.

Multilevel Analysis of the Moderating Effects of Baseline Depressive Symptoms on Relations between 1) Stress and Lapse Risk, 2) Stress and Affect, and 3) Affect and Lapse Relationship (Short Timeframe: 2-12 hours).

a. The direct effect of stressful event and lapse risk within 12 hours (c path) and the effect of negative and positive affect on lapse risk (b path)
Level-1, Report-Level Coefficient Odds Ratio 95% CI p-value
Level-2, Individual-Level & Cross-Level Interaction Coefficient
Model of t1 Lapse Risk 15 minutes to 12 hours after Index Report (c path)
Random Intercept: Mean Person-level Probability of Smoking at t1 a 0.064 (0.037, 0.112) 0.000**
Fagerström Test for Cigarette Dependence 1.198 (0.955, 1502) 0.118
High Depressive Symptoms (vs. Low) 3.682 (1.338, 10.13) 0.012*
Proportion of Stressful Event Reports 1.019 (0.038, 27.50) 0.268
t0 (Index Report) Stressful Event (Y/N) 2.231 (1.240, 4.014) 0.007**
t0 Stressful Event X High Depressive Symptoms (vs. Low) 0.836 (0.345, 2.027) 0.692
t-1 Negative Affect 15-min. to 12 hrs. pre-index 0.927 (0.782, 1.099) 0.382
t-1 Positive Affect 15-min. to 12 hrs. pre-index 0.930 (0.780, 1.108) 0.417
t0 (Index Report) Recent Smoking (within 2hrs. pre-index) Random Slope (Y/N) 4.081 (2.629, 6.336) 0.000**
t-1 Previous Smoking Status (Y/N) 1.526 (1.144, 2.036) 0.004**
Minutes between t0 and t1 Reports Random Slope 1.002 (1.001, 1.003) 0.000**
b path - entered into c path model aboveb
t0 (Index Report) Negative Affect - main effect 1.264 (1.050, 1.522) 0.013*
t0 Negative Affect X High Depressive Symptoms (vs. Low) 1.175 (0.824, 1.674) 0.373
t0 (Index Report) Positive Affect - main effect 1.139 (0.937, 1.370) 0.168
t0 Positive Affect X High Depressive Symptoms (vs. Low) 0.932 (0.644, 1.348) 0.707
b. The effect of stressful event on negative affect within 2 hours (a path)
Level-1, Report-Level Coefficient Estimate Standard Error p-value
Level-2, Individual-Level & Cross-Level Interaction Coefficient
Model of t0 (Index Report) Negative Affect (a path)
Random Intercept: Mean Negative Affect at t0 (Index Report)a -0.056 0.057 0.323
High Depressive Symptoms (vs. Low) 0.285 0.113 0.012*
Proportion of Stressful Event Reports 0.012 0.004 0.003**
t0 (Index Report) Stressful Event (Y/N) 0.809 0.049 0.000**
t0 Stressful Event X High Depressive Symptoms (vs. Low) -0.117 0.079 0.137
t1 Negative Affect 15-min. to 12 hrs. pre-index 0.332 0.014 0.000**
t0 (Index Report) Recent Smoking (within 2hrs. pre-index) Random Slope (Y/N) 0.074 0.043 0.085
t1 Previous Smoking Status Random Slope (Y/N) -0.003 0.040 0.934
c. The effect of stressful event on positive affect within 2 hours (a path)
Level-1, Report-Level Coefficient Estimate Standard Error p-value
Level-2, Individual-Level & Cross-Level Interaction Coefficient
Model of t0 (Index Report) Positive Affect (a path)
Random Intercept: Mean Positive Affect at t0 (Index Report)a 0.095 0.081 0.254
High Depressive Symptoms (vs. Low) -0.266 0.160 0.096
Proportion of Stressful Event Reports -0.004 0.006 0.428
t0 (Index Report) Stressful Event (Y/N) -0.406 0.051 0.000**
t0 Stressful Event X High Depressive Symptoms (vs. Low) 0.020 0.081 0.805
t-1 Positive Affect 15-min. to 12 hrs. pre-index 0.204 0.015 0.000**
t0 (Index Report) Recent Smoking (within 2hrs. pre-index) (Y/N) -0.017 0.032 0.601
t-1 Previous Smoking Status (Y/N) -0.032 0.230 0.249

Note:

The variables that are left aligned are Level-1 (report: within-subject) variables and those that are right aligned are Level-2 (individual: between-subject) variables and cross-level interaction terms. Proportion of Stressful Event.

Reports: this variable was multiplied by 100 before being entered into the model in order to interpret the changes in estimate or odds ratio per 1% increment in the proportion of the stressful event reports. Radom Slope: Specified as a random coefficient. All other predictors were treated as fixed to facilitate model convergence.

t-1 = Report 15 minutes and 12 hours preceding index report

t0 = Index report

t1 = Report 15 minutes to 12 hours after index report

a

Mean affect when all variables in the model are at zero (for the continuous variables, zero reflects the overall mean).

b

In order to test the b paths, negative affect and positive affect at index were first entered into the model (presented as main effects) and then cross-level interactions between depressive symptoms (level 2) and affect (level 1) were entered into the model. For simplicity, the other variables in the model are not presented.

*

p<.05,

**

p <.001