Survey Data Analysis


I am interpreting some data that was calculated by someone else and initially I was confused how this person got these numbers. So I emailed the office that oversees the analysis of this information and their explanation still left me a little confused. Hopefully someone can clarify this for me.

The data is from pre and post surveys. Part of the survey asks about dietary intake. So here is an example of a question asked:

In the past week, how many servings of fruit did you eat per day?

0 times per day, less than 1 time per day, 1 time per day, 2 times per day, 3 times per day, 4 times per day, 5 times per day, 6 times per day.

So after all the survey data is collected, it was reported that this group was eating an average of 1.9 pieces of fruit per day.

Now I emailed this office because I didn't know how one could obtain that information from what was provided. What they said is that they used 0.5 for the "less than 1 per day" category. First off, I know that this survey isn't exhaustive which was problem with it. Next, the way it is setup seems to me like it is interval data which as far as I know you couldn't possible spit a number like 1.9 pieces of fruit per day. My biggest question is using 0.5 for that "less than one per day" category appropriate to do? And is this a common practice? I just want to make sense of this all I can move on. Any inputs are welcome. Thanks!


Well-Known Member
Yes. It is common practice to use the midpoint of a bin as a proxy for everything in that bin - the 0 to 1 bin in this case - if you don't have a better idea. Sometimes you just have to do the best you can with what you've got. To get the average they probably found the sum of the proportion in each bin x the bin value.
One question you can ask is how much the results would be affected if you used a value other than 0.5.