Wednesday, January 21, 2015

Food Preferences Survey

Towards the very beginning of the class, way back when the weather was nice and we could survey people outside, I collaborated with two other stats students to conduct a statistical study on food preferences. By providing a real life experience of going out and collecting data, this project gave me the bedrock of real world knowledge of the challenges statisticians face and the many ways a study can go wrong before you even get to the point of having data to analyze.

As we started the project, we learned about the challenges one can run into in terms of choosing good questions that people will find relevant and which, as much as possible, can’t be tainted by lurking variables. If you have an excess amount of potential lurking variables in your question, your data could be almost impossible to make sense of even if you collect and analyze it well. So we went back and forth with different ideas for questions for some time.

In many ways, even the questions we had settled on and the setting for collection we had chosen to conduct the survey weren’t airtight, even after much deliberation. There were certainly flaws in the data, but I think we did pretty well overall, and it was an exhilarating experience to gather our own data. We also had fun making a graphic which you can see here:

In conclusion, this project, within the context of the lessons we were learning at the time, was very effective in teaching me the bedrock of statistics, and the extreme caution one has to approach even well-collected data with. There are far too many ways to mess with statistics and for statistics to go wrong for one to be able to take anything with a grain of salt.

No comments:

Post a Comment