Academia.eduAcademia.edu

Figure from:Constructing socio-economic status indices: how to use principal components analysis

See full PDFdownloadDownload figure
Table 2. Mean socio-economic score by quintile

Table 2 Mean socio-economic score by quintile

Related Figures (5)

Figure 1. Representation of two sequential components in PCA. Source: [http://www.ucl.ac.uk/oncology/MicroCore/ HTML _resource/PCA_1.htm], accessed 7 September 2005
Figure 1. Representation of two sequential components in PCA. Source: [http://www.ucl.ac.uk/oncology/MicroCore/ HTML _resource/PCA_1.htm], accessed 7 September 2005
Table 1. Results from principal components analysis the representation of households from urban areas into the richer groups, and subsequently increased inequality. An explanatory analysis should consider an index without direct determinants of the outcome of interest. However, exclusion of variables may make it more difficult to divide households, particularly when considering similar groups, for example in a rural community. In some studies, this has been found to be closely associated with the choice of variables included in the index. For example, Houweling et al. (2003) compared the relative economic position of households using either durable assets, infrastructure, housing characteristics or a combination of all variables to derive four different PCA-based measures. However, Filmer and Pritchett (2001), in their analysis, concluded the categorization of households was robust to the measure used.
Table 1. Results from principal components analysis the representation of households from urban areas into the richer groups, and subsequently increased inequality. An explanatory analysis should consider an index without direct determinants of the outcome of interest. However, exclusion of variables may make it more difficult to divide households, particularly when considering similar groups, for example in a rural community. In some studies, this has been found to be closely associated with the choice of variables included in the index. For example, Houweling et al. (2003) compared the relative economic position of households using either durable assets, infrastructure, housing characteristics or a combination of all variables to derive four different PCA-based measures. However, Filmer and Pritchett (2001), in their analysis, concluded the categorization of households was robust to the measure used.
Table 3. Ownership of durable assets and housing characteristics by SES quintile
Table 3. Ownership of durable assets and housing characteristics by SES quintile
Figure 2. Distribution of socio-economic scores 4. Discussion
Figure 2. Distribution of socio-economic scores 4. Discussion
Table 4. Proportion of households in low, medium and high socio-economic group for entire sample rural or regional communities, and constructing an index at community level increases the risk of clumping and truncation. If the analysis is to be undertaken for a rural community, Houweling et al. (2003) advise including items associated with SES for that location. Planning surveys before hand, and using local knowledge to pick out variables that could discriminate households into groups, could help to determine such a list of indicators. However, there will continue to be a trade-off in terms of the additional expense of obtaining more specialized data for a particular setting, and the simplicity of using asset-based measures.
Table 4. Proportion of households in low, medium and high socio-economic group for entire sample rural or regional communities, and constructing an index at community level increases the risk of clumping and truncation. If the analysis is to be undertaken for a rural community, Houweling et al. (2003) advise including items associated with SES for that location. Planning surveys before hand, and using local knowledge to pick out variables that could discriminate households into groups, could help to determine such a list of indicators. However, there will continue to be a trade-off in terms of the additional expense of obtaining more specialized data for a particular setting, and the simplicity of using asset-based measures.