Question

Topic: Research/Metrics

Combining Mixed Mode Survey Data

Posted by Anonymous on10/28/2009 at 9:06 PM ET 250 Points
我工作在一个纵向研究uses both phone and web survey methods (respondents are roughly 50/50 phone/web). Are there any rules of thumb or advice you can offer regarding combining the data from these two sources together... I plan on weighting the data 50/50 phone web to reduce the mode biases. Any other advice? What has been your experience when combining data from mixed modes? I am already aware of mode differences... I'm looking for advice on best ways to account for mode differences.
To continue reading this question and the solution, sign up ... it's free!
SUBSCRIBE
或,SIGN IN

RESPONSES

  • Posted bykoen.h.pauwelson10/28/2009 at 9:50 PM Accepted
    If your sample sizes are sufficient, it pays to keep the data seperate for each mode. This will reduce noise and thus increase the internal validity of your study. Next, you can do a pooling test to investigate whether your studied relationships hold up across modes, and whether you can combine both modes without problem
  • Posted on10/29/2009 at 2:21 PM Accepted
    Typically, in projects with mixed-modes, I have not really seen any weighting because the idea is that one mode is compensating for the bias in the other approach. I would suggest looking at the demographic profiles of the two groups, and proportions of responses to key questions and developing a weighting system based on those variables. (If you didn't collect the demo information, you can use the key questions or any behavioral responses).

    Also, the following article may be helpful (it's pretty outdated - pre-Internet age, but may give some overall guidance):

    https://www.quirks.com/articles/a1991/19911105.aspx?searchID=4782206&sort=5...
BACK TO TOP

Post a Comment

Related Discussions

Community Info

Top 25 Experts
(Research/Metrics)