Mean reliability estimates by age group and item type in the PATH-RV Study
| Age group/item type | No of Items | Mean kappa/ correlation | Mean generalised kappa | Mean weighted kappa | Mean GDR | Mean generalised GDR |
| Adult (n=407) | ||||||
| Dichotomous | 252 | 0.67 | – | – | 0.06 | – |
| Categorical (>2 categories) | 37 | 0.62 | 0.66 | 0.67 | 0.16 | 0.06 |
| Ordinal scale | 137 | 0.47 | 0.75 | 0.64 | 0.31 | 0.06 |
| Numerical response | 21 | 0.75 | 0.68 | – | 0.43 | 0.24 |
| Youth (n=117) | ||||||
| Dichotomous | 106 | 0.69 | – | – | 0.07 | – |
| Categorical (>2 categories) | 15 | 0.47 | 0.66 | 0.63 | 0.32 | 0.12 |
| Ordinal scale | 103 | 0.41 | 0.62 | 0.53 | 0.3 | 0.07 |
Seven dichotomous items for the adults and 11 for the youth were excluded from the calculation of the mean kappa because of zero cells. For the numerical items, the mean Pearson correlation is displayed as a measure of reliability. Generalised kappa and generalised GDR are calculated based on approximate agreement, which treats answers within one category of the earlier response as agreeing. The weighted kappa takes the size of the discrepancy into account.
GDR, gross discrepancy rate; PATH-RV, Population Assessment of Tobacco and Health reliability and validity.