Before we turn to the topic of measurement, I'll briefly clarify the terms variable and operationalization. Earlier, I referred to variables as operationalized constructs. But the term variable can also refer to a representation of a construct that is still somewhat abstract. We use the term operationalization if we want to explicitly indicate we're talking about a specific, concrete measure to measure or manipulate a construct. Say I'm interested in the construct, political power. I could represent this construct with the variable, influence in parliament. This representation is more specific. But we could still come up with very different procedures to operationalize or measure the variable, influence in parliament. For example, I could count the number of bills that a member of Parliament or Congress got passed. Or I could count the number of years someone has been in Parliament. Or I could ask political analysts to rate members of Parliament in terms of their influence. So, operationalization means selection or creation of a specific procedure to measure or manipulate the construct of interest. An operationalization makes it possible to assign people an actual score on the variable of interest. Suppose I want to operationalize the construct, love of animals. I can observe a person interacting with a cat and count how often the person pets or strokes the cat. I could also decide on a different operationalization, or operational definition, by creating a questionnaire with statements like, I usually avoid other people's pets, and I love animals. What about independent variables that are manipulated? Suppose I want to know if exposure to animals increases love of animals. I can operationalize the variable, exposure to animals, by creating two levels of exposure. I could randomly assign people to take care of a cat from a shelter for a month, or assign them to a control condition with no cat. Another operationalization would be to take one-half of a school class to a petting zoo and the other half to an abstract art museum. Or I could assign participants to watch an animal documentary or a train documentary. As you can see, the possibilities for operationalization are endless. Of course, some operationalizations are better than others. An operationalization doesn't necessarily capture or represent the construct in its entirety. As our constructs get fuzzier or more complex, there's a greater chance that we measure or manipulate only a part of the construct. For example, if we measure love of animals with a self report questionnaire, we measure feelings and attitudes, which might give a more positive image. If we measure love of animals by placing cameras in people's homes and observing behavior, we might find lower scores. We might find that, compared to their self-reported love of animals, people show a lot less love when their cat wakes them up at 5 a.m., or blocks the TV when they're watching their favorite show. So it's important to keep in mind what aspect of the construct the operationalization actually measures or manipulates, especially once we use the data to draw conclusions about the entire construct in our hypothesis. Our conclusion may apply only to a limited aspect of the construct.
