Hi everyone,
I am in the process of conducting a survey of work teams; the original idea was to survey everyone in each team (each team is composed of 4 members), and aggregate their scores to come up with group-level scores for our constructs of interest. We are not interested in any individual level characteristics; everything is at the group level or higher.
We just found out, however, that team membership is not independent; in other words, we thought each team member belonged to one and only one team. But in some teams, certain members are shared by more than one team. So the data would look something like this:
Member Team
Mary 1
Sue 1
Bob 1
Anne 2
Mary 2
Tom 2
Tom 3
Chris 1
Alex 1
So Mary is part of both teams 1&2 and Tom is part of teams 2 &3. So how do you handle them in the survey? Do you:
1. Ask them to respond with respect only to the team they spend the most time on?
2. Ask them to complete a separate survey for each team they work with?
3. Exclude their responses from the team's data?
4. Exclude that entire team?
5. Use a key informant approach, and only survey one team-member per team?
6. Something else entirely?
Any clever ideas would be appreciated.
Thank you!
My dear
I suggest to include shared membership as control variable or exclude the entire team.
If you are really not interested in individual level data, I suggest you have each group complete the survey as a group. In other words, they have to reach consensus on each survey question and jointly complete one questionaire.
Some good suggestions above, also consider:
1) Depending on the goals of your research, you may want to "control" for multiple team participants as it may skew your results. For example only include members from team 1, or survey candidates who work with one team and disregard members who work in more than one.
2) You could analyze the data of the entire sample and then control for the variable mentioned in suggestion 1). Compare the results.
3) Know your variables and design your survey in a way that isolates the information you are researching and excludes irrelevant info.
The thing that I have learned about research is, the conditions are often at the discretion of the researcher. So long as you can rationalize each decision and explain it to other academics then you should be in good shape.
Thank you all, for your thoughtful suggestions. I think after discussing these and other options with the team, we're going to collect data from all team members, and ask the "double dippers" to complete the survey with respect to the team they work in the most. That way we could do some comparisons with and without those teams included, or with and without those team members included. Thank you!