Dear Dr. Pete...

Dear Dr. Pete:
Which is the best scale to use in a survey?

~ Do Remi

Dr. Pete replies...

Dear Do:
This is tricky because there are a lot of scales, and they each have pros and cons. You may be surprised to know that scales actually were invented by real people with real thought behind them.

Let’s start with one of the most common. The Likert Scale is named after Rinsus Likert, who invented this scale for his university thesis in 1932. The scale is bi-directional and usually has 5 points on it with a neutral position in the middle. Typical Likert scales are “Agree Strongly, Agree Slightly, Neither Agree nor Disagree, Disagree Slightly, and Disagree Strongly.” Likert scales can have 7 or 9 points or can force choices with 4, 6, or 8 points (the neutral point having been removed).

Although it is common to assign numeric values to the points on the scale, this should not be done. One cannot assume respondents perceive the difference between adjacent levels as equidistant. The beauty of a Likert Scale is that it gives both a strength and direction to a given attitude measurement.

The Guttmann Scale was invented by Louis Guttmann in 1944. This scale measures unidimensionality – there is a single trait underlying the scale and the respondent chooses the highest one that applies. It is assumed that all those below apply and all those above do not apply.

These types of scales are extremely difficult to write well. Consider the statement: “I think immigration should be reduced.” This would not be chosen by people who are in favor of open immigration or by those who favor no immigration at all.

However, the following Guttmann Scale does work:

1. Are you willing to permit immigrants to live in your country?
2. Are you willing to permit immigrants to live in your community?
3. Are you willing to permit immigrants to live in your neighborhood?
4. Are you willing to permit immigrants to live next door to you?
5. Would you permit your child to marry an immigrant?

Agreement with option 3 would imply agreement with options 1 and 2.

The Thurstone Scale was developed in 1928 by Louis Leon Thurstone. This scale measures attitudes along a single dimension by asking respondents to indicate whether they agree or disagree with each of a large set of statements (e.g., 100). The statements are designed to be parallel in construction, with some toward one end of the scale and some toward the other end, each trying to indicate the attitude in a slightly different way.

In addition to developing the statements, they must be validated. For instance, expert judges (or pre-testing subjects) would rate each of the statements on how they reflect the extremes of the attitude being measured. Thurstone scales are hard work!

Charles Egerton Osgood invented the Semantic Differential Scale. This has opposites (e.g., fair/unfair and bitter/sweet) at the ends of the scale and respondents place themselves somewhere on the scale. The scale might also consist of words (e.g., excellent, good, adequate, poor, and inadequate).

Of course, we can’t forget the classic rating scale 1 – X, always anchored with words. (This one doesn’t seem to have an inventor.) You might think you can manipulate the results of such scales mathematically, producing averages, etc. However, while these are ordinal scales, there is nothing to say that the attitudinal “distance” is the same between 1 and 2 as it is between 9 and 10. This doesn’t stop researchers from doing exactly that but, hey-ho.

Oh, to answer your question, I don’t know; no one does. It really depends on your study objectives and which scale you prefer.

Cheers!
Dr. Pete