I’ve been experimenting on myself and collecting all sorts of data about my mood, who I’m with, when I’m working, etc. It’s a useful way to understand yourself better. I even made a website called N = One for some of the data I collected.
There’s only one problem; the data looks weird. Here’s an example:
According to my data, I spend most of my time either with nobody or strangers. The data was collected using the Reporter App. Reporter beeps you at random points during the day and asks you to fill out a short survey.
This works well for collecting data, especially over an extended period of time.
But only if you actually fill out the survey. What I realized when I looked at my data is that I was less likely to answer the beeps to fill out the survey when I was working or with friends. When I was riding the subway or in a cab, I would answer the survey straightaway.
So what’s the solution? I could become more disciplined and always answer the call of the survey. I could weight certain answers knowing that some answers are more likely to be ignored than others. For example, each time I answer that I’m with friends, I’ve probably ignored the survey on three other occasions. Therefore, one “I’m with friends” answer is actually worth three.
If you’re tracking yourself and finding weird patterns in your data, check whether your collection method has a bias toward answering. The boring moments—riding the subway, waiting in line—might be overrepresented, while the interesting ones slip away unrecorded.