Why we categorize

This is part of the Data Stories Series.

I was looking through the data set, sifting through for any preliminary clues. The screening interviews were scheduled for the following week. I was writing the screener questionnaire. The classification work had begun.

The initial data mining had not revealed any obvious answers as to who should be recruited. Not the gender, not the ethnicity, not even the ZIP Code. With such wide parameters how, then, do you select screening criteria?

The answer, it turns out, is related to categorization. For a researcher, categories emerge as early as research design. When we think of who to recruit for our research, we are not thinking of individual entities. We think in terms of groups and of identifiable characteristics individuals within that group might share. This is the basis for stratification recruitment strategies.

The role of classification in this particular instance was to make a large data pool manageable. Groups are like shortcuts. We think in groups because it offers us a simplification. When they are well made, categories help us highlight the meaningful characteristics, and eliminate the noise. Not all characteristics are significant, after all — and judicious groupings filter accordingly.

Classifications also make it possible to compare groups across dimensions. And while individuals within a group will never be identical, they will be the same on the variables that matter. In that way, the good researcher is the one who can…