Adjustment for confounding factors: multivariate linear regression analysis
The approach taken to the analysis of the study has been to first examine the actual observed changes in pollution, exposure and fuel costs ('univariate' analysis), and the to examine whether confounding factors may have played a part in either over or under-estimating the observed effects of the interventions. In this Annex, we examine the effect of taking account of confounding factors. Overall, these appear to make little difference to the conclusions we drew from the initial analysis. Given this conclusion, and some of the complexity in carrying out and interpreting the adjusted analyses given the nature of the data and the relatively small numbers, the results of this stage are reported in this Annex.
Annex 19 to Smoke, health and household energy, Volume 1: Participatory methods for design, installation, monitoring and assessment of smoke alleviation technologies, 10 pp.