How to create a representative sample from a population

[Nicolas Orgeira]

Hi,

I have a dataset of property characteristics (access to water, surface area, type of wall/roof, etc.) for all properties in a given city. I would like to sample a given number of properties while ensuring that the sample is representative, and proportionally sample by area in the city. Because I have more than 30 property characteristics, I am not able to do a stratified sampling, and a random sample wouldn't be representative (or with probability nearing zero). I looked into using the repsample command but unfortunately it displays the message "Not enough cases to provide the requested sample".


Any suggestions on how to proceed would be greatly appreciated.

Thank you

[Rich Goldstein]

The stat lit is replete with differing definitions of "representative" in w.r.t. sampling so you need to clarify what you mean; if you want to look into this, an old but still valuable set of papers on this can be found:
1. Kruskal, W and Mosteller, F (1979), "Representative sampling I", International Statistical Review, 47(1): 13-24
2. Kruskal, W and Mosteller, F (1979), "Representative sampling II", International Statistical Review, 47(2): 111-127
3. Kruskal, W and Mosteller, F (1979), "Representative sampling III", International Statistical Review, 47(3): 245-265
4. Kruskal, W and Mosteller, F (1980), "Representative sampling IV", International Statistical Review, 48(2): 169-195