Results of generalised estimating equation models indicating the change in HWL effectiveness between waves
| Wave 1 % (95% CI) | Wave 2 % (95% CI) | Wave 3 % (95% CI) | Wave 2–Wave 1 | Wave 3–Wave 2 | Wave 3–Wave 1 | Test* | |||||
| Outcome measure | PP (95% CI) | P value | PP (95% CI) | P value | PP (95% CI) | P value | χ2 | P value | |||
| All respondents | |||||||||||
| Aware of warning labels on SLT packaging (n=8381, obs=16 166) | 73.1 (67.2 to 78.2) | 73.4 (67.4 to 78.6) | 71.1 (66.9 to 75.0) | 0.3 (−5.7 to 6.4) | 0.909 | −2.3 (−9.0 to 4.4) | 0.496 | −1.9 (−8.7 to 4.9) | 0.570 | 0.51 | 0.775 |
| Noticed warning labels ‘often/whenever use’ SLT (n=8368, obs=16 113) | 37.2 (30.7 to 44.1) | 32.1 (25.6 to 39.4) | 35.8 (31.2 to 40.5) | −5.1 (−13.1 to 3.0) | 0.213 | 3.6 (−4.5 to 11.7) | 0.373 | −1.4 (−9.9 to 7.0) | 0.733 | 1.67 | 0.434 |
| Among respondents who noticed HWLs ‘often/whenever use’ SLT | |||||||||||
| Read warning labels ‘often/regularly’ (n=4253, obs=5641) | 50.8 (43.5 to 58.0) | 44.6 (36.0 to 53.5) | 32.8 (27.7 to 38.5) | −6.2 (−18.4 to 6.0) | 0.313 | −11.7 (−22.0 to –1.5) | 0.026† | −17.9 (−27.4 to –8.4) | <0.001 | 15.92 | <0.001 |
| Warning labels make you think about risks ‘a lot’ (n=4229, obs=5605) | 13.8 (10.8 to 17.6) | 17.9 (11.9 to 26.0) | 33.6 (30.1 to 37.3) | 4.1 (−3.7 to 11.8) | 0.296 | 15.7 (7.4 to 24.0) | <0.001 | 19.8 (14.8 to 24.8) | <0.001 | 52.09 | <0.001 |
| Warning labels make you ‘a lot’ more likely to quit (n=4216, obs=5573) | 14.4 (10.7 to 19.1) | 18.9 (12.8 to 27.0) | 36.5 (32.2 to 41.0) | 4.5 (−3.6 to 12.6) | 0.274 | 17.6 (8.8 to 26.4) | <0.001 | 22.1 (15.1 to 29.0) | <0.001 | 35.45 | <0.001 |
| Avoided looking at warning labels (n=4246, obs=5632) | 7.6 (5.0 to 11.5) | 10.1 (7.0 to 14.5) | 40.2 (31.8 to 49.2) | 2.5 (−1.7 to 6.8) | 0.238 | 30.1 (21.1 to 39.0) | <0.001 | 32.6 (22.5 to 42.6) | <0.001 | 58.96 | <0.001 |
| Gave up SLT ‘at least a couple of times’ because of warning labels (n=4237, obs=5618) | 29 (21.9 to 37.4) | 36.6 (27.5 to 46.8) | 35.2 (28.9 to 42.1) | 7.6 (−3.5 to 18.6) | 0.175 | −1.4 (−14.1 to 11.3) | 0.829 | 6.2 (−4.7 to 17.1) | 0.260 | 2.30 | 0.317 |
Estimates in columns ‘Wave 1’, ‘Wave 2’ and ‘Wave 3’ are predicted marginal estimates from the logistic model. Estimates in columns ‘Wave 2–Wave 1’, ‘Wave 3–Wave 2’ and ‘Wave 3–Wave 1’ are average marginal effects, or the percentage point difference between waves. Test.
*Test shows results of χ2 testing comparing results across all three waves. Tests do not account for multiple testing (although all significant results will remain significant if a Bonferroni correction is applied, except for the cell marked with a ‘†’). Covariates in logistic GEE models were survey wave, sex, age group, living in an urban area, marital status, socioeconomic status, past-year attempts to quit SLT (at least one vs none), intentions to quit SLT and Tobacco (SLT) Dependence Index.
GEE, generalised estimated equation; SLT, smokeless tobacco.