(Note: (The title of this comment was edited shortly after posting for better clarity) Since posting the blog I've had quite a few emails with some very relevant questions. I thought it might be helpful to summarize the replies here in case others are wondering the same things. I would also encourage everyone to post questions and comments directly here on the blog, so that everyone can benefit, especially if I haven’t been completely clear about some of the technical points. Firstly: how much could the statistics be affected by large outlying values from the rich households? There are some households on Nakai who are doing quite well, many of whom were already well off before the project. Are their consumption levels biasing the statistics upwards? The answer to this is no, not really. The statistics I have used here, and will try to use as far as possible in future blogs are median not mean values. To get the median we basically line up all the households in the sample (or subsample) in order of consumption (or whatever variable is under consideration). Then we pick the one in the middle- the median household, and list that household’s consumption level. This is affected by the richer households- if we removed them, the middle person would be lower down in the ranking. But the middle household itself, the one that we list, isn’t going to be one of these rich households.