Table 3.
Results of Weighted Logistic Regression - Risk Perceptions by Tobacco User Groups (Non-user, Single-product User and Dual/Multiple-product User), (n = 3907; N = 461,069)
| Non-user (1) 89.5% (n = 3574; N = 412,677) |
Single-product (2) 6.56% (n = 206; N = 30,238) |
Dual/Multiple-product (3) 3.94% (n = 127; N = 18,153) |
|||
|---|---|---|---|---|---|
|
% reporting ____ not at all harmful to healtha |
%c (95% CI) |
% (95% CI) |
% (95% CI) |
p-value | Cohen’s dd |
| E-cigarettes | 14.22 % (12.92%–15.52%) |
43.27 % A (34.41%–52.13%) |
53.26 % A (41.05%–65.47%) |
<.001 | 1 vs 2: 0.71 vs 3: 0.98 vs 3: 0.27 |
| Cigarettes | 6.57 % A (5.34%–7.80%) |
6.27 % A (2.39%–10.15%) |
21.04 % (7.86%–34.22%) |
<.001 | 1 vs 2: 0.10 vs 3: 0.45 vs 3: 0.67 |
| Hookah | 10.44 % (9.19%–11.70%) |
21.72 % (12.47%–30.97%) |
42.41 % (29.40%–55.41%) |
<.001 | 1 vs 2: 0.42 vs 3: 0.81 vs 3: 0.61 |
| Cigars | 6.73 % A (5.48%–7.99%) |
11.89 % AB (4.28%–19.51%) |
23.95 % B (11.32%–36.59%) |
<.001 | 1 vs 2: 0.15 vs 3: 0.62 vs 3: 0.32 |
| Smokeless | 7.06 % (5.61%–8.50%) |
14.85 % A (8.15%–21.54%) |
26.48 % A (12.98%–39.97%) |
<.001 | 1 vs 2: 0.19 vs 3: 0.64 vs 3: 0.39 |
|
% reporting ____ not at all addictiveb |
|||||
| E-cigarettes | 54.29 % (49.87%–58.70%) |
71.75 % A (62.35%–81.15%) |
70.66% A (60.33%–81.00%) |
<.001 | 1 vs 2: 0.32 vs 3: 0.31 vs 3: 0.01 |
| Cigarettes | 46.61 % A (42.36%–50.86%) |
59.25 % B (48.49%–70.02%) |
57.58 % AB (43.46%–71.70%) |
<.001 | 1 vs 2: 0.17 vs 3: 0.15 vs 3: 0.02 |
| Hookah | 54.29 % A (49.68%–58.90%) |
72.77 % B (63.62%–81.91%) |
64.07 % AB (52.60%–75.54%) |
<.001 | 1 vs 2: 0.32 vs 3: 0.15 vs 3: 0.20 |
| Cigars | 49.00 % A (44.46%–53.56%) |
31.53 % B (22.00%–41.06%) |
35.84 % AB (23.87%–47.81%) |
<.001 | 1 vs 2: 0.27 vs 3: 0.20 vs 3: 0.09 |
| Smokeless | 51.93 % (47.83%–56.04%) |
72.60 % A (64.40%–80.80%) |
69.39 % A (58.52%–80.27%) |
<.001 | 1 vs 2: 0.35 vs 3: 0.30 vs 3: 0.07 |
Note.
All analyses adjusted for sex, grade, race/ethnicity and family income.
Measured on a 1–4 scale from “not at all harmful” to “definitely harmful.” Responses were collapsed into 2 categories: “Not at all” harmful versus “some” harm
Measured on a 1–3 scale from “not at all addictive” to “very addictive.” Responses were collapsed into 2 categories: “Not at all” addictive versus “some” addictiveness.
Adjusted proportions were calculated post-estimation. Regression coefficients are available on request. Proportions in cells sharing a letter are not significantly different at the 5% level; p-value adjusted with Bonferroni correction
Effect sizes are classified as small (d=0.20), medium (d=0.50) and large (d≥0.80)