If there is one college course I wish I could take a second time it would be statistics.
Sure, I still remember the differences between mean, median and mode and when and why you use them. I recall calculating standard deviation and thanks to Excel it’s a snap. And I vaguely remember that for one reason or another having a p-value? closer to 1.0 on a graph is a good thing…
But thanks to not paying attention in statistics class and treating it as I did calculus and organic chemistry (as in, seriously, will I ever use this again?) I have an embarrassing problem:
I have a hard time making the case to the scientists I work with that “sample size” and “statistical significance” doesn’t factor in to the kind of user research I conduct; especially if a survey is involved.
- Will you take, “Just because,” as an answer?
- How about: “This is soft science.”
- Or more honestly: “Nope, it’s not a statistically significant sample size.”
At one point in my career, having statistically significant sample sizes WAS an important consideration. Why isn’t it important to the work I do now?
Finally, I found a decent explanation and discussion from the folks at Survey Gizmo.
Read Statistical Significance in Market Research by Dr. Ed Halteman.
Suggestions for further reading are welcomed!
And if Dr. Ed is totally wrong, please tell me. After all, this article on the Internet so it MUST be true, right?
P.S. If Zoomerang isn’t meeting your needs, give SurveyGizmo a try.