I'm seeking some advice about a project that I am working on.
In this project, my colleagues and I are planning on collecting a sample that is intended to be representative of the national adult population of Australia. We are collecting data through a panel company which can approximate, but not exactly match the demographics of the broader population. They have advised us that some groups are likely to be underrepresented, such as 18-25s. However, we want to estimate some descriptive statistics from the collected data, such as estimated mean scores and percentages, which reflect the characteristics of the larger population.
We have discussed several methods of doing this, such as some kind of demographic weighting (e.g., rim weighting). Another alternative we discussed was to treat this as a missing data problem, and use multiple imputation to generate values for underrepresented groups. (That is, we would insert blank cases into the dataset containing only the demographic information of underrepresented groups, and then impute data for these cases.)
I would be keen to hear the advice and suggestions of the group about how this issue could be dealt with.