TABLE 4

Mean (SE) for Variables by Usage Groups, With F for ANOVA and Ps for 3 Contrasts From Multinomial Regression

VariableUsage GroupFContrast (P)
1. Nonuser2. E-Cigarettes Only3. Cigarettes Only4. Dual Use1 vs 22 vs 32 vs 4
Social-cognitive protective factors
 Parental support25.9 (0.20)23.3 (0.40)20.0 (0.97)21.8 (0.48)36.5.001.01.03
 Parental monitoring21.2 (0.10)20.0 (0.20)18.7 (0.50)19.0 (0.24)33.0.001.04.01
 Academic involvement18.3 (0.11)16.6 (0.22)15.5 (0.53)15.5 (0.26)46.7.001ns.002
 Grades past year4.3 (0.02)3.9 (0.05)3.8 (0.11)3.5 (0.06)65.1.001ns.001
 Behavioral self-control65.4 (0.32)61.2 (0.65)57.9 (1.58)57.0 (0.77)43.2.001ns.001
 Emotional self-control44.3 (0.28)40.4 (0.57)35.9 (1.42)38.9 (0.68)33.9.001nsns
 Social competence9.4 (0.08)9.7 (0.16)8.7 (0.39)9.7 (0.19)2.5#nsnsns
Social-cognitive risk factors
 Smoker prototype8.9 (0.10)9.4 (0.20)10.1 (0.50)10.5 (0.24)13.3.02ns.01
 Smoking expectancies8.8 (0.16)10.1 (0.31)13.3 (0.79)14.3 (0.38)68.6.001.002.001
 Behavioral dysregulation38.9 (0.27)43.6 (0.54)42.6 (1.33)45.7 (0.64)45.3.001ns.02
 Emotional dysregulation21.7 (0.22)24.7 (0.45)27.4 (1.11)26.7 (0.54)37.0.001.03.01
 Peer smoking0.8 (0.04)1.5 (0.08)1.9 (0.19)2.8 (0.09)135.6.001ns.001
 E-cigarettes healthier1.6 (0.01)1.8 (0.03)1.7 (0.06)1.8 (0.03)30.3.001nsns
Problem behavior risk factors
 Parent–adolescent conflict7.6 (0.09)8.7 (0.18)9.4 (0.46)9.5 (0.22)29.2.001ns.02
 Sensation seeking13.4 (0.12)15.8 (0.24)16.8 (0.58)17.8 (0.29)89.6.001ns.001
 Rebelliousness6.3 (0.09)8.4 (0.18)9.5 (0.46)10.9 (0.22)152.7.001ns.001
Collateral substance use
 Alcohol use0.5 (0.03)1.5 (0.06)1.9 (0.14)2.5 (0.07)296.4.001ns.001
 Marijuana use0.1 (0.03)0.6 (0.05)1.1 (0.13)2.6 (0.06)474.4.001ns.001
 Heavy drinking0.1 (0.02)0.3 (0.03)0.4 (0.08)0.9 (0.04)138.2.001ns.001
  • ns, nonsignificant. Usage group sizes are as follows: Nonuser group = 1319 cases; e-cigarette only group = 331 cases; cigarette-only group = 54 cases; and dual-use group = 228 cases. F for ANOVA has approximately 3/1910 degrees of freedom. Fs are significant at P < .001 unless otherwise noted; # indicates not significant. Regressions include control for gender, grade, family structure, ethnicity, and parental education and adjustment for any within-school clustering. For contrasts, 0.001 indicates effect is significant at P < .001; otherwise, exact probability indicates significance level of result.