Addi van Bergen

Chapter 4 82 Trichotomisation The 85 th and 95 th percentile scores of the index and dimension scales were calculated in the weighted total sample (Figure 2). This resulted in corollary prevalence rates between 5.0 and 5.2 percent “moderate to strong” exclusion and between 8.6 and 11.8 percent “some” exclusion on the general index and the dimensions scales. Prevalence rates in the development and validation samples were very similar. Figure 2. Distribution of SEI-HS scores. Each dot represents 1% of the weighted study population. The pink square marks the 85 percentile. The red triangle marks the 95 percentile. Validation of the SEI-HS 1. Content validity The data in Table 3 show that the SEI-HS items covered all the aspects of SE that form part of the SCP index. All four dimensions of SE were measured with three or more items. Only one item had a low component loading i.e. ‘didn’t receive medical or dental treatment’ (component loading 0.27); and one item had a low weight i.e. ‘I have enough money to heat my home’ (weight 0.09). The eigenvalues of the dimension scales ranged from 0.43 for Normative Integration tot 0.54 for Social Participation and Social Rights, which is largely consistent with the eigenvalues of the SCP dimension scales. As expected, the scores on the SEI-HS were right-skewed (Figure 2) with mean 0 and standard deviation 1, i.e. similar to the SCP Index. 2. Internal consistency The SEI-HS has a sufficient canonical correlation (0.33). This is somewhat lower than the correlation found for the SCP Index (0.38). Cronbach’s alpha for the dimension