Morse, J. M. (2009). Sampling in Grounded Theory. The SAGE Handbook of Grounded Theory, 229-244. doi:10.4135/9781848607941.n11
By Janice M. Morse
Q: Would determining the interviewees in advance contradict GT?
Principles of Sampling for Quantitative Inquiry
Principle 1: Excellent research skills are essential for obtaining good data
- The better the interview, the less interviews you’ll need
- Quickly gain trust of the interviewees to get accurate information (data) from them
Principle 2: It is necessary to locate “excellent” participants to obtain excellent data
- Excellent interviewees are: experienced, articulate, willing, available
- Strategize sampling based on conceptual information needs of the study; make it valuable for the interviewee
Principle 3: Sampling techniques must be targeted and efficient
- Too much data is hard to analyze and sift through; leads to “conceptual blindness”
- Excellent qualitative inquiry is biased – but this doesn’t necessarily mean the research is bad
- You have to stick to analyzing the best examples
- Theoretical saturation cannot be reached in randomized sampling
Principles of Sampling in GT
Main types of sampling:
- Convenience Sampling
- Finding people available to scope the boundaries of phenomenon to explore
- Leads to snowball sampling
- Purposeful Sampling
- Looking for participants in a certain stage (which was identified in convenience sampling)
- Confirms trajectory of research
- Biased process, but it’s ok – leads to saturation
- “Shadowed data” = participants speaking for others
- You sample the next participant based on concepts in your current data. If person A talks about the influence of x types of people, you want to interview people with x characteristics
- Theoretical Sampling
- Emerging categories and developing theories guide the sampling
- How should I articulate the way I will be changing the sampling of interviews in my proposal?
- Theoretical Group Interviews
- Fill in missing pieces of puzzle
- Researcher presents ongoing analysis and asks to fill in the “thin” areas
Terminating Data Collection
- Occurs when characteristics of instances are constant
- Ex. Modesty as a cultural value for Fiji-Indians interferes with maternal and infant health
Sampling Data
- Researchers may disregard some data that’s not helpful or irrelevant
- Not all data will be the same – some will have better descriptions than others
- The better examples will be used to draw categories and theories from
- This is not biased because not all data is equal