3 PCA: Building an Asset Index

This section of the practical will take you through constructing a DHS style asset index of household wealth. We will also construct a concentration index to look at household wealth and HIV prevalence. Then, we will return to mediation to answer the question of whether the number of sexual partners mediates the relationship between educational attainment and HIV status.

Load the usual libraries:

Before proceeding, either run the entire code that you saved from session 10 or run the line below. If you need a refresher on how PCA works, take a quick flick through the Beers Example or refer to the lecture slides.

tz <- read.csv("./tz2.csv")

We are first going to extract measures of household assets. It is reasonable to think that ownership of some of these assets may encode some information about the socioeconomic position of that household within the wider community. Note that not only might this asset index be useful for characterising relative position (tacitly invoking a psychosocial theoretical model), but it may also be useful as a direct measure of deprivation (tacitly invoking a neomaterial theoretical model).