Perhaps the most commonly used form of scaling is attributed to Rensis Likert (1932). Likert introduced the ever-popular 5-point scale that we talked about in chapter 9, on questionnaire construction. Recall that a typical question might read as follows:
Please consider the following statements carefully. After each statement, check the answer that most reflects your opinion. Would you say you agree a lot with the statement, agree a little, are neutral, disagree a little, or disagree a lot with each statement? Ok, here’s the first statement:
When I need credit to bring my bananas to market, I can just go the agricultural bank in Ralundat and they give it to me.
- ? Agree a lot
- ? Agree
- ? Neutral
- ? Disagree a little
- ? Disagree a lot
The 5-point scale might become 3 points or 7 points, and the Agree-Disagree scale may become Approve-Disapprove, Favor-Oppose, or Excellent-Bad, but the principle is the same. These are all Likert-type scales.
I say ‘‘Likert-type scales’’ rather than just ‘‘Likert scales’’ because Likert did more than just introduce a format. He was interested in measuring internal states of people (attitudes, emotions, orientations) and he realized that most internal states are multidimensional. You hear a lot of talk these days about conservatives and liberals, but the concept of political orientation is very complex. A person who is liberal on matters of domestic policy—favoring government-supported health care, for example—may be conservative on matters of foreign political policy—against involvement in any foreign military actions. Someone who is liberal on matters of foreign economic policy—favoring economic aid for all democracies that ask for it—may be conservative on matters of personal behavior—against same-sex marriage, for example.
The liberal-conservative dimension on matters of personal behavior is also complicated. There’s no way to assign people to a category of this variable by asking one question. People can have live-and-let-live attitudes about sexual preference and extramarital sex and be against a woman’s right to an abortion on demand.
Of course, there are packaging effects. People who are conservative on one dimension of political orientation are likely to be conservative on other dimensions, and people who are liberal on one kind of personal behavior are likely to be liberal on others. Still, no single question lets you scale people in general on a variable as complex as ‘‘attitude toward personal behavior,’’ let alone ‘‘political orientation.’’ That’s why we need composite scales.