Table 2
Latent variable correlations
| M | S | B | D | O |
|---|
| Motivating | 1.00 | | | | |
| Situations | 0.91 | 1.00 | | | |
| (0.02) | | | | |
| 41.29 | | | | |
| Behaving | 0.87 | 0.95 | 1.00 | | |
| (0.02) | (0.02) | | | |
| 36.03 | 52.79 | | | |
| Directions | 0.84 | 0.94 | 0.96 | 1.00 | |
| (0.03) | (0.02) | (0.02) | | |
| 32.25 | 50.09 | 56.77 | | |
| Organizing | 0.87 | 0.95 | 0.98 | 0.96 | 1.00 |
| (0.02) | (0.02) | (0.02) | (0.02) | |
| 35.61 | 53.67 | 63.20 | 57.26 | |
6.4 External validation
6.4.1 Concurrent validities
To stress the consistency of the results across methods of scoring, we show r-based
validity coefficients for the sample-specific data and 2-PL rescaled r-based validity
coefficients for the combined sample data.
Supervisors' ratings. Concurrent validities for the rank-order correlation scoring
of the ESJI are shown in Table 3 for the U.S. sample, the Spanish sample, and both
samples combined. For the U.S. sample, validity coefficients ranged from .22 to .46
with a median of .36. All (8 of 12) correlations were statistically significant at the .05
level. For the Spanish sample, validity coefficients ranged from -.10 to .21 with a
median of .14. Only three of the correlations were statistically significant. For the total
sample, validity coefficients based on the 2-PL model ranged from .09 to .32 with a
median of .20. Ten of the 12 correlations were statistically significant. Correlations
were approximately the same for the d2 scores and actually slightly better with the 1-PL model.
|