Radiance Blog

Forbes.com made no friends with America’s mayors recently when they decided to rank “America’s Most Sinful Cities.” Ranking each city in the United States on the classic Seven Deadly Sins (Lust, Pride, Avarice, Envy, Gluttony, Sloth, and Wrath), they created a snazzy-looking interactive graphic that shows which cites are the best examples of each sin. The Mile High City, home of Corona Research, was proudly represented, finishing fourth in Avarice and an astounding first in Lust!

But the interesting thing about this project was exactly how Forbes chose to measure sinfulness. There’s no Federal Bureau of Inquisitions to generate and track statistics on the seven deadly sins, so Forbes needed to find a proxy, or another variable that can stand in for the variable you really care about. This process of moving from the concept you really care about to the variable you can actually measure is called operationalizing your variable, and how you choose to operationalize variables can have dramatic impacts on your results.

For example, to measure Lust, Forbes used numbers from ACNielsen regarding the per capita sales of condoms and other contraceptives in grocery and drug stores. Does this truly measure “Lust”? You could argue that it better measures responsible lust, optimism in perceived future sexual behavior, the effects of public-health campaigns, or the number of young professionals in the metro area (to be fair, Forbes does fess up to the likelihood of the last two alternate explanations).

So what would be a better measure of lust? A survey is probably out of the question, because this is one topic where people will likely lie (although there are detailed—and expensive– methods that can be used to lower this tendency). The number of nightclubs per capita? The number of craigslist personal ads per capita per week? STDs per capita? Prostitution arrests? Attractiveness of the population? Or how about a set of extensive neurological measurements of sexual response? All of these have positives and drawbacks, and honestly I would lean toward creating an index that combines several relevant concepts to minimize the errors inherent in any one measure. But regardless of the measure chosen (or the way a question is worded) to operationalize a variable, when interpreting research you need to be very clear about what you want to measure, how you actually measure it, and the distance between the two.