Table 3. Relationship between time and evidence dissemination.
| Tweet variable | Code | Time 1 | Time 2 | All | Standardised residuals (z scores)* | Overall significance |
|---|---|---|---|---|---|---|
| Evidence mentioned, n = 1038 | Yes | 101 | 39 | 140 | z = ±3.8 | χ2 = 33.4, df = 1, p<0.001 |
| No | 412 | 486 | 898 | z = ±1.5 | ||
| URL linked, n = 1038 |
Yes | 373 | 353 | 726 | z = ±0.7 | χ2 = 5.1, df = 2, p = 0.078 |
| No | 128 | 150 | 278 | z = ±0.8 | ||
| No document access | 12 | 22 | 34 | z = ±1.2 | ||
| Quality of evidence cited in URL, n = 726 | Cites peer-reviewed journal article(s) | 250 | 70 | 320 | z = ±6.7 | χ2 = 168.7, df = 2, p<0.001 |
| Refers to non-peer-reviewed research or evidence | 65 | 185 | 250 | z = ±5.6 | ||
| Does not refer to research or evidence | 58 | 98 | 156 | z = ±2.5 | ||
| URL author sector, n = 726 | Health sector | 110 | 18 | 128 | z = ±5.5 | χ2 = 76.5, df = 2, p<0.001 |
| Tobacco industry-linked | 27 | 23 | 50 | z = ±0.3 | ||
| Neither | 236 | 312 | 548 | z = ±2.7 |
* Categories which significantly contribute to the overall chi squared statistic have z scores outside ±1.96 (significant at p<0.05), outside ±2.58 (significant at p<0.01), and outside ±3.29 (significant at p<0.001). All significant scores are highlighted in bold.