Wednesday, July 11, 2007

Wed 2 - Qual sampling & Retention

Sampling, Recruitment and Retention -- Qual part by Deborah Padgett

Hey y'all, we've had some quantitative presentations that didn't apply to my work as much now, so I didn't take notes, but now, we've got Deborah Padgett talking, and she wrote a couple of books about qualitative methods and I'm taking notes so let me take them here.

She starts by reminding us that most all research is with a non-probability sampling.

Sample size
Vigorous discussion here about how you prestate sample size when the goal is to get saturation or redundancy, which is an iterative phenomena you can't predict beforehand. You need to include a sample size table in your grant (but I don't see that in the directions, I'll ask more later). And you need to tell your IRB sample size. And I've heard unofficially from an AJPH editor that they really hoped any incoming qualitative studies had at least 25 folk. Deborah says it's understood that yr sample might be larger if you are doing grounded theory, and less if you are doing phenomenological theory, & if your journal doesn't like your sample size, switch, because a sample of 10 can really be valuable when conducted well. But consensus is to project sample size, even if you then undershoot. In the immortal words of one of the co-chairs here... "You always have to say a number, then we never hit it, but you say it."

Main types of Purposive sampling
  • extreme or deviant case sampling (seeking the outliers)
  • intensity sampling (similar but doesn't go to the extreme outliers)
  • maximum variation sampling (sampling for heterogeneity)
  • homogeneous sampling (opposite of above)
  • typical case sampling (getting individuals in the middle of the bell curve)
  • criterion case sampling (setting criteria and using folk who meet it, also with Nominations as a subset where folk refer others who meet criteria then you pick, often picking the folk with most nominations for the criteria, best with a positive valence)
  • snowball sampling (when first enrollees refer others, NIDA has a protocol for snowball with IDUs)
  • RDS and PDR for CBPR (respondent driven or participant driven sampling)
Retention issues
Many folk have gotten grants, set sample sizes then fallen way short... now NIH is talking about cutting off funding for unmet enrollment goals (Quote from our OBSSR rep: "It's beyond talk")

So plan for your attrition in yr sample plan. (note from another meeting on this point - best way to keep your folk enrolled and in contact is to plan for routine contact points with them, don't leave big gaps, otherwise you lose folk).

Retention tips (from a large homeless study that had amazing retention rates)
  • keep a toll free project number and distribute business cards
  • use incentives, cash & metro cards (cash is best and you should argue for it)
  • monthly check-in phone interviews
  • Refreshments for all gatherings
  • easy access location, near subway hub
  • incentives mailed out quickly
  • letters and holiday cards sent to participants
  • holiday party (the idea is to create a community of folks)
  • interviewers become familiar with participants life and use "rapport talk"
  • use respectful and formal language (not first names unless indicated)
Now we want to talk more, but it's time for our next speaker, so half of us will probably corner her in the hall later.

Hey, thanks to Rupaleem, another participant, she's telling us all about this good sampling website from RWJF with cites for the different types of sampling....

No comments: