Two approaches to purposive sampling 1 theoretical a

This preview shows page 6 - 8 out of 15 pages.

two approaches to purposive sampling 1. theoretical: a. start with maybe a small number of respondents b. through your data collection, analysis, and learning process, you might start to define other selection criteria, c. find additional respondents.
d. continuously sampling during your study duration (not always easy for program evaluation) 2. need a structured budget and work plan for our evaluation activities a. a priori define the characteristics and the structure of our qualitative sample, b. link it very clearly with our evaluation question and purpose. c. decide on a number of respondents in each sort of category of characteristic d. if additional sampling units needed or don't need as many, be able to build that in there different types of techniques within a purposive sampling strategy sampling from the extremes create homogeneous sample Or every possible level of heterogeneity is included among your respondents Often in program evaluation, concerned about finding the typical case. want to know on average how the program is functioning. opportunistic sampling: just get the respondents that we can snowball sampling: start with a pool of respondents and they help you identify who else you should talk to get the information you're looking for Will always be asked about sample size we don't care that much about sample size in qualitative research. What we're concerned about is the completeness of the data. want to make sure that our sample is representative of all the different variations o ex. we want to know how well an information system has been implemented and it has components down from the health post level all the way up to the central level, then we want to make sure that we get representation from all the different types of users of the information system o sample size is more determined by the completeness of the data and when we feel we've reached saturation with the knowledge that we've gained from our respondents o as evaluator usually need a budget and work plan: that's what number you should use for minimum sample or perhaps a maximum sample that's related to your research purpose o need to always document how you are selecting units and respondents in your sample o If there's any change in sampling, need to always document the justification typical case is either an illustrative case or it's your actual unit of analysis o need to first have discussions with stakeholders during the scoping, key informants in the beginning, to actually understand what constitutes typical o define those characteristics and make your selection appropriately o Snowball sampling can be really useful when you're not familiar with the field. if you happen to need to investigate something as part of your program evaluation, where respondents are actually hard to find, snowball sampling might be the only way that you can do that.

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture