Table 2. Quality variables by cluster membership and predicted latent class.
| Quality variables | Full sample | Cluster Membership | Latent Class Membership | ||||
|---|---|---|---|---|---|---|---|
| Other* | 1 | 2 | 1 | 2 | 3 | ||
| 1. Minutes to complete questionnaire (median) | 21 | 21 | 14 | 935 | 20 | 15 | 935 |
| 2. Correlation of synonyms (mean) within-person correlation of items with strongest overall correlation |
0.4 | 0.39 | 0.58 | 0.33 | 0.28 | 0.59 | 0.33 |
| 3. Even-odd item correlation (mean) Correlation of even/odd items on unidimensional scales |
0.62 | 0.61 | 0.74 | 0.66 | 0.60 | 0.64 | 0.62 |
| 4. Distance from average (mean) Euclidean distance between participant’s responses and population average responses |
15.64 | 15.82 | 12.85 | 15.33 | 17.40 | 13.45 | 15.18 |
| 5. Runs of identical responses (mean) Average over pages with many items |
0.43 | 0.44 | 0.47 | 0.27 | 0.43 | 0.49 | 0.27 |
| 6. Inconsistent state/climate selection | 15.0% | 12.9% | 16.9% | 61.2% | 6.8% | 13.8% | 52.7% |
| 7. Non-US phone | 16.0% | 15.3% | 13.9% | 36.7% | 13.1% | 8.4% | 52.0% |
| 8. Wrong phone number | 13.0% | 13.8% | 9.2% | 2.0% | 8.9% | 22.4% | 1.3% |
| 9. Obviously fake name | 6.8% | 6.8% | 12.3% | 0.0% | 0.5% | 16.3% | 3.3% |
| 10. Nonsensical feedback e.g. “dsadasdasd” |
2.8% | 2.0% | 20.0% | 0.0% | 0.3% | 7.1% | 0.0% |
| 11. Discrepancies within questionnaire | 4.3% | 3.8% | 16.9% | 0.0% | 2.3% | 7.7% | 2.0% |
| 12. Other | 9.3% | 9.7% | 6.2% | 4.1% | 7.4% | 12.5% | 7.3% |
Note: Shading emphasizes unusual behavior patterns as measured by the quality indicators
*“Other” denotes membership in one of the clusters not identified as usually large/similar