Addi van Bergen

Construction of a multidimensional measure for social exclusion 53 3 Construction of the measurement instruments In the four health questionnaires, we identified 11 items that matched the SCP operationalisation of the dimension limited social participation. All of these items belong to the loneliness scale of De Jong Gierveld [15]. In addition, in the Utrecht questionnaire 3 items were available on the frequency of social contacts. For the measurement of the dimension ‘material deprivation’ 2 items were available in each of the cities. These items relate to the financial situation of the household and difficulties in making end meets. We found no items to measure ‘inadequate access to basic social rights’ in the Amsterdam sample. In the questionnaires of The Hague, Rotterdam and Utrecht we found 18 items that matched the operationalisation of this dimension by the SCP, including 5 items on neighbourhood cohesion, 2 items on satisfaction with housing and living environment and 11 items on environmental and nuisance problems in the neighbourhood. The questionnaires from The Hague and Rotterdam included 2 additional items on feeling unsafe during the day or night. From the Utrecht questionnaire 26 additional items were selected that related to the presence of moisture or mold in the home, to the need for information or assistance with health problems and to the need for facilities in the neighbourhood. Items for the dimension ‘lack of normative integration’ were not available in any of the questionnaires. With the aforementioned 62 items, three indices were constructed: Index1 was based on the items from the Amsterdam questionnaire, Index2 on the items from the Rotterdam and The Hague questionnaires combined and Index3 on the items from the Utrecht questionnaire. In Index1, 8 of the 13 items were retained, in Index2 14 of the 33 items and in Index3 17 of the 57 items. With one exception, items were removed because of low component loadings or low weights. The item on ‘mold and moisture in the home’ from the Utrecht questionnaire was removed because of a reverse association with the other items. The centroid plots generated by the OVERALS analyses are given in Figures S1, S2 and S3. Table 3 shows the selected items per index and per dimension. From the 14 items that were present in two or more datasets, 10 were included in all relevant indices and 4 items were included in some indices but not in others. For example, the item ‘There are enough people I feel close to’ was incorporated in the indices 2 and 3 but not in Index3. Instead, Index3 contained the item ‘I miss having people around’, which was absent in the indices 1 and 2.