Thursday, June 3, 2010

Keeping Your Survey Reliable

For future students who will be doing this project my biggest piece of advice would be to overestimate everything. Overestimate how much time it will take to get the surveys back, how many surveys you will need to get out, and how long it will take to count the numbers. I would suggest doubling all estimates. If you think it will take a week to collect all your data, plan on two weeks. If you think you need to pass out 100 surveys to get a reasonable amount of numbers to work with, pass out 200.

I would also suggest testing your survey on several people before passing it out. We were required to test the survey on at least two people, but I think that testing it on closer to ten would be a better number. From there it is also important to consider the answers testers gave, not just questions they had. It is crucial to check that they’re giving the answers you want. They may answer a question in one way and not think there’s a problem with it when in fact there is. If you come across this problem you should ask them about it and try to figure out why they answered it the way they did. Assuming you know why they answered a question the way they did could end up making the survey even less reliable.

No comments:

Post a Comment