Constructing a Personality Inventory

The Minnesota Multiphasic Personality Inventory (MMPI) is one of the most widely used personality tests in existence. It consists of 550 statements, each of which the respondent is to identify as either "true as it applies to me" or "false as it applies to me." The MMPI was developed according to the empirical method of test construction. The first step in development was to prepare a large set of questions that psychologists and psychiatrists thought might discriminate between normal people and people with abnormal behavior problems. Then, these questions were tried out with normal subjects and subjects diagnosed as having various forms of mental illness (for example, depression, schizophrenia, and hypochondriasis). If normal subjects responded consistently to an item in one way and abnormal subjects responded consistently to the same item in a different way, that item was considered a good discriminator of behavior and was retained on the test. In this way, response patterns for normal versus abnormal subjects emerged. (The items that did not discriminate were dropped from the inventory, since they added nothing to the test's ability to diagnose mental illness.)

This empirical method of test construction is a highly technical and complex process. Sophisticated statistical techniques are used to estimate how much each question is contributing to the predictive power of the inventory. Although the development and documentation of a reliable and valid inventory is beyond the scope of an introduction to assessment and individual differences, you can use the same basic method used to develop the MMPI to create an instrument that will discriminate between two different groups of people. This experiment will show you how to develop a personality inventory that will differentiate between men and women.

Experimental Procedure

To do this experiment, you will need to get ten of your friends to volunteer to serve as subjects. Five should be women and five should be men. You can tell them the experiment will take less than 10 minutes to complete. You will also need to make 10 copies of the Personality Inventory provided here. Do the following steps to collect the data for this experiment.

STEP 1 Examine the questions on the Personality Inventory. These questions seem likely to be answered differently by men and by women.

STEP 2 Give each subject a copy of the Personality Inventory. Make sure the subject understands the instructions. Encourage all subjects to answer all items as best they can. In some cases, they will probably have trouble deciding which of the two choices is better. Tell them to make the choice that best describes themselves.

STEP 3 Have each subject answer the questions.

STEP 4 Thank each subject for participating and collect the questionnaires. Be sure to mark the sex of the subject on the Personality Inventory form if it is not marked in Question 21.

Analyzing the Results

STEP 1 Divide the completed Personality Inventories into two groups-those answered by male subjects and those answered by female subjects.

STEP 2 Number the questionnaires in each group from 1 to 5

STEP 3 Score each questionnaire according to the following scale. Record the scores in the appropriate columns of the Scoring Table on page 281.

For Male Subjects

1. Questions 2, 3, 5, 7, 10, 15, 16, 17, 18, 20: Write a 1 beside each of these question numbers of the man answered T; write a 0 beside the question number if he answered F.

2. Questions 1, 4, 6, 8, 9, 11, 12, 13, 14, 19: Write a 1 beside each of these question numbers if the man answered F; write a 0 beside the question number if he answered T.

For Female Subjects

1. Questions 1, 4, 6, 8, 9, 11, 12, 13, 14, 19: Write a 1 beside each of these question numbers if the woman answered T; write a 0 beside the question number if she answered F.

2. Questions 2, 3, 5, 7, 10, 15, 16, 17, 18, 20: Write a 1 beside each of these question numbers if the woman answered F; write a 0 beside the question number if she answered T.

Each box in the Scoring Table should now contain either a 1 or a 0.

STEP 4 Add up the scores for each question. For example, for Question 1, add up the scores for Male #1, #2, #3, #4, #5, and for Female #1, #2, #3, #4, #5. You must get a total between 0 and 10 for each question. Record this score in the Total column.

STEP 5 Circle the total scores on the Scoring Table that exceed 6, the number that indicates that more than half of your subjects have responded as expected according to their gender. These are the questions that are good discriminators between men's and women's attitudes and behaviors.

Discussion

Look at the actual questions in the Personality Inventory. Did the questions differentiate the way you expected they would? Were there any surprises? Do you think you might get different results if you gave this questionnaire to people in a different age group? Why or why not?

Do you see any problems with this type of testing? Do you think your subjects answered truthfully? Lying or faking is a big problem with self-report tests like this one and the MMPI. The MMPI has three validity scales-the L scale, F scale, and K scale-that are used to determine if a respondent is faking. In addition, published tests are evaluated many times before they are used in diagnosis. They are tested and retested in many situations to determine if they are reliable and valid. Finally, normative data are collected so that test administrators have average scores to which they can compare an individual's results.

Perhaps now you understand better the construction of tests that are developed according to the empirical method, like the MMPI. If so, you might like to try your hand at writing questions that would discriminate between people under thirty and people over thirty, for instance. What might some of these questions be?