Alternate forms reliability is a concept in statistics and psychology which means that two different versions of the same questionnaire/form/test should produce similar responses from the individuals being tested.
Example:
Suppose that a person is asked to fill out a questionnaire in order to conduct a personality test. Suppose that a month later, the person is once again given the same form to fill out with some of the questions rephrased. If the responses of the person vary wildly we conclude that the questionnaire does not have alternate forms reliability.
This means that there is some mistake in the way the questionnaire was designed by the researcher. For the purposes of the statistical study, it is not appropriate that the same questionnaire produce such wildly different responses from the same individual. The researcher must redesign the questionnaire to obtain reliable results.
How do you measure alternative forms reliability?
- Give the first version of the test to a selected group of people and convert the results into numerical form.
- After the passage of some time, give the second version of the test to the same group of people and convert the results into numerical form.
- Calculate the correlation coefficient between the results obtained in the two tests.
- If the correlation coefficient has a high positive value we conclude that the test has alternate forms of reliability. This is because the results of the two tests are highly correlated and hence somewhat similar to each other.
Why are alternate forms necessary?
Alternate Forms are necessary because if the subject is asked to fill the same kind of form again and again then he will develop a familiarity with the format of the questions. By rearranging and rewording the questions we can ensure that the responses of the subject are fresh and spontaneous.
What kind of changes can be made in the alternate form?
We can construct another version of the form by:
- Reordering the questions.
- Rephrasing the questions without changing the meaning.
- Adding questions that ask for the same kind of data as in the previous version of the form.
