Home | Business News | Browse by Publication | J | Journal of Information Systems

Alternative personality measurements: commentary on accounting information systems research opportunities using personality type theory and the Myers-Briggs Type Indicator.

Article Excerpt
I. INTRODUCTION

Four primary themes are presented in "Accounting Information Systems Research Opportunities Using Personality Type Theory and the Myers-Briggs Type Indicator": (Wheeler et al. 2004) (1) the need for personality-based research to supplement cognitive science research in AIS, (2) the Myers-Briggs Type Indicator (MBTI) as a personality measurement instrument, (3) a summary of MBTI-based accounting research, and (4) potential uses of the MBTI as an AIS research tool. Because most of the paper either discusses the MBTI directly or its potential use in future AIS research, I shall refer to it as the MBTI paper. I appreciate and applaud the authors for the discussions and admonishments in all of the areas. This response compares the MBTI with two alternative personality instruments in order to provide AIS researchers with the option of selecting the instrument that best fits research needs. The first section of the response discusses evaluation of personality measurement instruments, the second section describes the Millon Inventory of Personality Styles (MIPS), and is followed by discussion of the Big Five Traits of Personality.

II. EVALUATING PERSONALITY MEASUREMENT INSTRUMENTS

AIS researchers have much to gain by extending common research approaches (e.g., experiments involving system users and/or developers, surveys of users and/or installers, modeling of systems, and education for AIS) to include the measurement and impact of personality types. The focus of this response, however, is that many different personality measurement instruments exist and some alternatives to the MBTI might better accomplish AIS research objectives in given situations. Each researcher including personality measurement in a project should first determine why the impact of personality type is important and evaluate which instrument is best to accomplish the research objective.

Before discussing the alternative personality measurement instruments and comparing them with the MBTI, different measures of test reliability and validity need to be considered. In order to illustrate different levels of reliability and validity, I will use a continuing example of an AIS researcher attempting to measure the relative speed at which AIS students complete macro programming tasks.

1. Reliability--internal consistency measures: If ten questions in an instrument are intended to measure students' relative programming speed and many of the participants scoring high on five of the questions also score low on the other five, the test is not considered internally reliable. Common measures of internal consistency are Cronbach's alpha coefficient and split-half procedures.

2. Reliability--test/retest comparisons: If student participants retake the same speed test some time period later and there is low correlation between the two test scores obtained for the measured constructs, then the instrument is not considered reliable.

3. Reliability--participant alertness: Are responses meaningful? Some students, when responding to multiple pages of Likert-scaled questions might respond haphazardly to finish quickly or circle different scales so as to generate a dot matrix appearance of a crude message to the researcher. Internal comparison questions such as "I finish most programming projects much more quickly than other students" in the beginning of an instrument and "I take much longer than most other students to complete programming assignments" can be used to measure participant alertness. A positive response to an internal check question such as "I recently wrote a 25,000-line program in one hour" can also indicate a lack of reliability in the participant's responses. After including reasonable questions for student alertness and attitude in responding to an instrument, a cut-off point needs to be established for how many inappropriate responses indicate that a student's responses are not reliable.

4. Face validity: A question of "I can run farther and faster than most other students" has little to do with measuring relative rates of programming speed. If an experiment or survey instrument does not, on the face of it, appear to measure what it is intended to measure, appears to bias participants' responses, or is based on assumed scenarios that differ from reality, then it is deemed to lack face validity. If an editor, two or three blind reviewers, and several other experts in the field of study do not believe the experiment or instrument is not properly designed, then these opinions provide a reasonable indicator of inadequate face validity.

5. Construct validity: An instrument designed for application to AIS students may be intended to measure one or more additional constructs beyond relative programming speed. For example, an AIS researcher might want to measure intelligence as well as programming speed. Because intelligence is an abstract construct, comparison of results to an absolute measure is not feasible. Therefore, measures of construct validity are based on judgment. If the construct is associated with a clear and generally accepted theory, and if the questions in the measurement instrument also appear consistent with the known theory, then the construct measurement is judged to be valid.

6. Congruent validity: If the measure of relative programming speed is highly correlated with other associated measures it is said to have congruent (convergent) validity. Continuing the AIS student example, if the end resulting measure of relative programming speed is not highly correlated with other associated measures such as scores on programming projects or timed programming exams, the measure would not be considered to have high congruent validity.

7. Discriminant validity: In contrast to congruent validity, test measurements should also be found to have low correlation with other measures that are theoretically unrelated. There is little reason to posit that a student's age, height, and weight are associated with relative programming speed. Low correlations with these variables indicate that discriminant validity further supports the positive construct validity of the test.

8. Behavioral validation: We would expect that students who score high on relative programming speed measures based on responses to a reliable and valid test instrument should follow through and subsequently complete and turn in programming projects more quickly than those with low test score measures. If high correlations with expected behaviors are not found, the behavioral validity of the construct measure is questionable.

9. Social desirability tendencies: Assume an AIS researcher is developing new questions to measure another desired construct to correlate with programming speed. A personality-based construct example might be aggressiveness. Researchers need to be concerned that test takers may attempt to present themselves in a favorable light--intelligent, friendly, and/or other socially desirable characteristics. Subjects may attempt to "fake" what is considered a desirable response in order to appear as more aggressive or more laid-back than they actually perceive themselves. Two frequently used measures of social desirability are the Edwards Social Desirability Scale (Edwards 1957) and Marlow-Crowne Social Desirability...

View this article FREE - Now for a Limited Time, try Goliath Business News
Free for 3 Days!