Thursday, March 8, 2012

CEDO 540 Week 4: Survey Crunch

Num - Number - Numbest

When our group decided on a topic to do a survey on, I volunteered to create the form using Jotform. After I got started, I realized that our method of trying to get meaningful results was flawed. Our original idea was parallel to asking this to get some kind of ranking of importance of items.

On a scale of 1-10, rate the following:
1 = no no never never uh uh uuuhhh, 10 = super duper double scooper.

  1. Do you like chocolate ice cream?
  2. Do you like chocolate cake?
Well, it will be hard to get anything meaningful to compare the two items using this method. Those that like chocolate will most likely rank both very high and those that don't will rank them both low, giving us a meaningless comparison.

So, I thought, how can we rank and compare a list of items, many of which most people will rank high. Then I thought of a couple of surveys I've taken where they ask you to choose between 2 or 3 items, sometimes where you really have a hard time choosing because you think all the answers are almost equal, but you have to choose. So it became more like this, "Would you rather have chocolate ice cream or chocolate cake?" Now we'll really see which people like better. Even those that don't really like chocolate would still have to choose the lesser of two evils. 

Image CC By Jorge Franganillo

I figured out how to do it and then did it, letting my group know I had made some changes. They didn't really get it at first. I didn't really know how I was going to analyze the results either and I certainly didn't expect my classmates to figure out how to crunch the numbers of something they didn't really design. I dug the hole and now I had to climb out. So I ended up learning more about spreadsheets again. Copying formulas works this way but not that way. Named ranges and cells are an enormous help with copying formulas. Taking the lazy route and doing all of the calculations and charts on the same sheet as the raw data makes for a messy sheet but saves you from having to reference the other sheet in a thousand times.

I learned a lot about creating surveys. I can see how getting reliable data from a survey really does take a lot of careful planning. In real life a survey would have to be piloted first to work out any kinks. I found out that my matrix of questions had a minor flaw, but flaw that wouldn't cut it in scientific journal. I don't think I want to be in the survey business or the research using statistics business but knowing more about it now will be good when someone tells me, "Oh, and we need a survey by Thursday when the newsletter goes out."

The survey is here. If it is still live, feel free to take it. The report is here.

No comments:

Post a Comment