Monday, 20 March 2017

Measurement Scale in Research

Measurement Scale in Mass Media Research


A scale represents a composite measure of a variable; it is based on more than one item. Scales are generally used with complex variables that do not easily lend themselves to single-item or single-indicator measurements. Some items, such as age, newspaper circulation, or number of radios in the house, can be adequately measured without scaling techniques. Measurement of other variables, such as attitude toward TV news or gratification received from going to a movie theater, generally requires the use of scales.
Several scaling techniques have been developed over the years. This section discusses only the better-known methods. Search the Internet for additional information about all types of measurement scales.

Simple Rating Scales


Rating scales are common in mass media research. Researchers frequently ask respondents to rate a list of items such as a list of programming elements that can be included in a radio station’s weekday morning show, or to rate how much respondents like radio or TV on-air personalities.

The researcher’s decision is to decide which type of scale to use: 1 to 3? 1 to 5? 1 to 7? 1 to 10? 1 to 100? Or even a 0 to 9 scale, which is commonly used by researchers who don’t have computer software to accept double-digit numbers (like 10). Selecting a type of scale is largely a matter of personal preference, but there are a few things to consider:

1. A scale with more points rather than fewer points allows for greater differentiation on the item or items being rated. For example, assume we are rating the importance of programming elements contained in a radio station’s weekday morning show. Let’s say the respondents are told, “The higher the number, the more important the element is to you.” Will a 1–3 scale or 1–10 scale provide more information? Obviously, the 1–10 scale provides the broadest differentiation. Broad differentiation in opinions, perceptions, and feelings is important because it gives the researcher more information. Artificially restricting the range of ratings is called factor fusion, which means that opinions, perceptions, and feelings are squeezed into a smaller space. It’s better for the respondents and the researcher to have more rating points than fewer rating points. Restricting respondents’ responses by using too few scale points always hides the range of potential responses and restricts the potential of any research study.

2. Our experience shows that males and females of all age groups, and all races and nationalities like to use a 1–10 scale. This is true because the 1–10 scale is universally used, particularly in sporting events like the Olympics. Virtually everyone understands the 1–10 rating scale. A 10 is best or perfect, a 1 is worst or imperfect. Our experience also shows that researchers should not use a 0–9 or 1–9 rating scale because, quite frankly, respondents do not associate well with a 9 as the highest number.

3. When using simple rating scales, it is best to tell respondents that “The higher the number, the more you agree,” or “The higher the number, the more you like.” Over thousands of research studies, we have found this approach better than telling respondents, “Use a scale of 1 to 10, where ‘1’ means Dislike and ‘10’ means Like a lot.”

 Transforming Scales

On occasion, a researcher will conduct a study using one scale and then later want to compare those data to other data using a different rating scale. For example, let’s say that a researcher uses a 1–7 rating scale and wants to convert the results to a 1–100 scale.
What can be done? The procedure is always the same: Divide the smaller rating scale into the larger to produce a multiplier to transform the scale. For the transformation of 1–7 to 1–100, first divide
100 by 7, which is 14.2857, and then multiply this number times each of the 1–7 elements to compute the converted 1–100 scale numbers
The new, transformed (rounded) ratings are:
1 5 14
2 5 29
3 5 43
4 5 57
5 5 71
6 5 86
7 5 100
What about transforming a 5-point scale to a 7-point scale? The procedure is the same:
Divide 7 by 5, which produce a multiplier of 1.4. This number is multiplied times each of the numbers in the 5-point scale to produce a transformed scale:
1 5 1.4
2 5 2.8
3 5 4.2
4 5 5.6
5 5 7.0

SPECIALIZED RATING SCALES

Thurstone Scales

Thurstone Scales
Measurement Scale in Research -  Thurstone Scales

Thurstone scales are also called equal appearing interval scales because of the technique used to develop them and are typically used to measure the attitude toward a given concept or construct. To develop a Thurstone scale, a researcher first collects a large number of statements (Thurstone recommends at least 100) that relate to the concept or construct to be measured. Next, judges rate these statements along an 11-category scale in which each category expresses a different degree of favorableness toward the concept. The items are then ranked according to the mean or median ratings assigned by the judges and are used to construct a questionnaire of 20 to 30 items that are chosen more or less evenly from across the range of ratings. The statements are worded so that a person can agree or disagree with them. The scale is then administered to a sample of respondents whose scores are determined by computing the mean or median value of the items agreed with. A person who disagrees with all the items has a score of zero.

One advantage of the Thurstone method is that it is an interval measurement scale. On the downside, this method is time consuming and labor intensive. Thurstone scales are not often used in mass media research, but they are common in psychology and education research.

Guttman Scaling

Guttman Scale
Measurement Scale in Research - Guttman Scale

Guttman scaling, also called scalogram analysis, is based on the idea that items can be arranged along a continuum in such a way that a person who agrees with an item or finds an item acceptable will also agree with or find acceptable all other items expressing a less extreme position. For example, here is a hypothetical four-item


Guttman scale:
1. Indecent programming on TV is harmful to society.
2. Children should not be allowed to watch indecent TV shows.
3. Television station managers should not allow indecent programs on their stations.
4. The government should ban indecent programming from TV.

Presumably, a person who agrees with Statement 4 will also agree with Statements 1–3. Furthermore, if we assume the scale is valid, a person who agrees with Statement 2 will also agree with Statement 1 but will not necessarily agree with Statements 3 and 4. Because each score represents a unique set of responses, the number of items a person agrees with is the person’s total score on a Guttman scale.
A Guttman scale requires a great deal of time and energy to develop. Although they do not appear often in mass media research, Guttman scales are common in political science, sociology, public opinion research, and anthropology.

Likert Scales

 Likert Scales
Measurement Scale in Research - Likert Scales

Perhaps the most commonly used scale in mass media research is the Likert scale, also called the summated rating approach. A number of statements are developed with respect to a topic, and respondents can strongly agree, agree, be neutral, disagree, or strongly disagree with the statements (see Figure 2.1). Each response option is weighted, and each subject’s responses are added to produce a single score on the topic. 

This is the basic procedure for developing a Likert scale:

1. Compile a large number of statements that relate to a specific dimension. Some statements are positively worded; some are negatively worded.
2. Administer the scale to a randomly selected sample of respondents.
3. Code the responses consistently so that high scores indicate stronger agreement with the attitude in question.
4. Analyze the responses and select for the final scale those statements that most clearly differentiate the highest from the lowest scorers

Semantic Differential Scales

Another commonly used scaling procedure is the semantic differential technique. As originally conceived by Osgood, Suci, and Tannenbaum (1957), this technique is used to measure the meaning an item has for an individual. Researches indicated that three general factors—activity, potency, and evaluation—were measured by the semantic differential. Communication researchers were quick to adapt the evaluative dimension of the semantic differential for use as a measure of attitude. To use the technique, a name or a concept is placed at the top of a series of seven-point scales anchored by bipolar attitudes. Figure 2.2 shows an example ofthis technique as used to measure attitudes toward Time magazine.

The bipolar adjectives that typically “anchor” such evaluative scales are pleasant/ unpleasant, valuable/worthless, honest/ dishonest, nice/awful, clean/dirty, fair/unfair, and good/bad. However, we recommend that a unique set of anchoring adjectives be developed for each particular measurement situation.

Strictly speaking, the semantic differential technique attempts to place a concept in semantic space using an advanced multivariate statistical procedure called factor analysis. When researchers borrow parts of the technique to measure attitudes, or images or perceptions of objects, people, or concepts, they are not using the technique as originally developed. Consequently, perhaps a more appropriate name for this technique is bipolar rating scales.