Likert scales involve a list of statements related to the attitude in question. Instead of checking only those statements with which they agree, however, respondents are asked to indicate the degree of agreement or disagreement or disagreements with each of the statements. Each degree of agreement is given a numerical score and the respondent’s total score is computed by summing these scores from all the statements. The following examples on attitudes towards hot cereal will illustrate. Respondents were given a card on which the following was shown:
1. Agree very strongly
2. Agree fairly strongly
6. Disagree fairly strongly
7. Disagree very strongly
8. Don’t Know
The respondents were then given a list of statements about hot cereal and asked to select one of the above answers for each statement. Some of the statements were as follows:
1. Hot cereal has a taste I like
2. Hot cereal is a mess to make
3. Hot cereal sticks to your ribs
4. Hot cereal is expensive
5. Hot cereal is high in vitamins.
The responses were scored from one to seven, as indicated above, except that the values were reversed for statements unfavorable to hot cereals, for examples, statements 2 and 4.The sum of scores from all the statements made that total score of the respondent. In this example, a person with a very favorable attitude towards hot cereal could receive a score of 5 (one on each of the five statements). While someone with a very unfavorable attitude could receive a score as high as 35 (seven on each statement times five statements).
Likert scales are developed considering the following:
1. Collect a large number of statements relevant to the attitude in question that can be clearly identified as either favorable or unfavorable.
2. Select a series of responses that represent various degrees of agreement or disagreement. Five variations, such as agree, strongly agree, undecided, disagree, and disagree strongly are often used, but various other numbers can be used. It is doubtful that respondents can make meaningful distinctions among more than seven or nine variations. An odd number is usually used in order to have middle, or zero point, but this is not necessary. If respondents are forced to chosen between a favorable or unfavorable response when they in fact, have no opinion, they should pick the mildest degree of agreement and disagreement about an equal number of times.
3. Administer the collected statements to a group reasonably representative of the universe to be studied and have them check one of the classes of agreement or disagreement for each statement.
4. Compute each individual’s score by summing the scores of the responses to each question. The responses must be scored so that the most favorable response has the same ratings for each statement. As some statements will be favorable and some unfavorable, this will mean reversing the ratings of the respondents for one of the groups of statements.
5. Eliminate those statements that do not discriminate between the high and low scorers on the total test. This can be done by selecting the high and low quartiles of respondents according to total score. Determine the average score on each statement among those in the high quartile and similarly among those in the low quartile. Those statements on which these averages differ by the largest amount are the most discriminating.
Likert scales are of the ordinal type: they enable one to rank attitudes, but not to measure the difference between attitudes. They take about the some amount of effort to create and are considered more discriminating and reliable because of the larger range of responses typically given in Likert scales. They have the same advantages as the Thurstone scales – similar scores can be achieved through varying combinations of responses. In this connection, the same arguments as were presented in the discussion of Thurstone scales is pertinent to Likert scales.
Comparison of Thurstone and Likert Scaling Methods: It is obvious that the Thurstone and Likert scaling methods have much in common. They have been two of the most widely used methods; yet there have been relatively few tests of the comparative, validity, reliability and efficiency of the two. The evidence to date indicates that the Likert scale is probably more reliable than the Thurstone method, but there is no sound evidence on the comparative validity.