Does Personality Change? On the Stability of Personality Assessment Scores
Does personality change? This is a question we receive regularly from our clients, along with a lot of hypotheses about when and why scores shift. Answering this seemingly straightforward question actually requires addressing three related questions:
- How often do scores on assessments change?
- When scores on assessments change, how large are those changes?
- Why do scores on assessments sometimes change?
How often do scores on assessments change?
Personality assessments—like the ones we create at Hogan—measure patterns of behavior. Decades of research have demonstrated that personality assessments predict future behavior, including workplace performance. A major reason why personality assessments work so well at predicting future behavior is because personality is quite stable; that is, people do not change very much. For example, in one study elementary school teacher ratings of students’ personalities predicted how those students behaved as adults 40 years later! The best method for quantifying personality stability is the test-retest correlation: you take a test now and we see how well it predicts your scores on the same test in the future. The short term (14-21 day) test-retest correlations for the Hogan Personality Inventory scales range from .69-.87. The long term (8-year) test-retest correlations range from .30-.73. A meta-analysis of 3,217 (7-year) test-retest correlations ranged from .30-.70. The point here is this: personality test scores are highly stable. Thus, most of the time, a person retaking a personality assessment will get very similar results.
When scores on assessments change, how large are those changes?
If you are a careful reader, you will note that “very similar results” is not the same thing as “identical results,” or that the test-retest correlations just described are not perfect. Indeed, even over short intervals, test scores do fluctuate. This is true for all tests, including cognitive ability and even medical tests (e.g., blood pressure, glucose tests, etc.). For Hogan, reassessments typically fall within two raw points of the original assessment. This indicates a small degree of fluctuation, and typically does not change the interpretation of the overall profile.
Why do scores on assessments sometimes change?
Consider the following situation: Your doctor measures your cholesterol levels. After some time has passed, your doctor measures your cholesterol again to discover the result has changed. In this scenario, there are two broad (and obvious) reasons the result changed: (a) something about the measurement process changed and/or (b) your actual cholesterol levels changed. Likewise, for psychological assessments, changes in scores may occur because of measurement-related and/or individual/psychological reasons.
Changes Related to the Measure
Sometimes aspects of the testing and measurement process itself result in score changes upon reassessment. There are three major methodological reasons scores on a psychological assessment might change: (1) imprecision in measurement, (2) changes in forms, and (3) changes in norms.
Imprecision in Measurement. When a doctor measures your cholesterol, he or she is not literally measuring all of the cholesterol in your bloodstream. Only a small sample of your blood is taken, and this sample is assumed to represent all of your bloodstream. Cholesterol in one sample may vary from cholesterol in another sample. Thus, one reason your cholesterol levels might change is because the instruments and procedures employed are imperfect. These imperfections are referred to as measurement errors. On a personality assessment, most people will always mark the item “I like to bend the rules every now and then” as either True or False no matter how many times they are asked. However, for a small group of people, whether they mark True or False to this item may depend on something that has happened to them recently (e.g., watched a movie about prisoners; sped through an intersection). Such imprecision in measurement affects the person’s score on the assessment. The good news is that these measurement errors occur randomly, meaning that the tests are unbiased. Such imperfections are present in all assessments, including medical tests. In developing our assessments, we work to reduce measurement error to the minimum possible. The short-term test-retest correlations reported above indicate that measurement error is quite low for our assessments.
Changes in forms. At Hogan, we believe in Kaizen Psychometrics. This means we work to continually improve our assessments. In doing so, we regularly update our testing instruments with new, better, forms when they are available. Because the new forms are designed to be better, they are not identical to the previous forms. Thus, a person completing an assessment on an earlier form and later taking an assessment on a new form, may receive slightly different scores.
Changes in norms. Because raw scales score can be hard to interpret (e.g., what does a 7 out of 12 mean?), we report normed (percentile) scores for our clients. Our norms are calculated from a stratified sample of millions of assessments completed by test-takers from nearly every job in every sector all over the world. However, in keeping with our commitment to Kaizen Psychometrics, we continually work to update our norms as better or more representative samples become available. As a result, a person completing an assessment scored on a previous norm may receive slightly different scores on a more recent assessment using updated norms.
Changes Related to the Individual
Beyond methodological reasons for change, there are also psychological reasons scores on assessments can change over time related to (a) maturity, (b) major life events, and (c) feedback and coaching.
Maturity. Although personality is relatively stable, personality does change across the lifespan. People become more self-confident, agreeable, conscientious, and more emotionally stable as they age. This pattern of personality development is typical for most people and reflects maturation into adulthood. As a result, such changes in personality are most dramatic for teenagers and young adults (early 20s), with personality becoming more stable with age. Thus, assessments taken over a shorter time span and assessments taken by older adults are less likely to show change.
Major Life Events. Personality can also change due to life events or personal experiences. For example, personality does seem to change in the wake of major life events (e.g., unemployment, marriage, divorce). Likewise, there is some evidence that experience in the military can result in personality change. However, people tend to adapt and return towards their baseline scores shortly after, even when these events are traumatic (such as bereavement) or conversely, positive (such as winning the lottery). Large changes due to major life events are rarely permanent, though small changes may be more lasting. Assessments taken recently after a major life event may show dramatically different results from assessments taken under more normal circumstances. However, in our experience, these results are still accurate at the time of assessment. For example, if a person experiencing a personal trauma scores lower on a scale measuring stress tolerance compared to their baseline, this is often a real and interpretable result. In other words, even though the change in scores may be temporary, they should not be discarded as “inaccurate.”
Assessment Feedback and Coaching. Finally, there is evidence that personality can change as a result of intentional practice and/or expert feedback and coaching. Likewise, effective coaching in business contexts appears to affect personal and organizational outcomes. However, in such cases the change in personality assessment scores tends to be relatively small. Thus, while coaching can be effective, we would expect any changes to manifest in 360° or performance feedback rather than in personality test results. When changes do occur, it is impossible to discern whether this is due to actual developmental growth or due to increased awareness about one’s assessment results (e.g., the person becomes aware of tendency to be arrogant and un- or sub-consciously manipulates assessment responses). These changes can and do occur in both directions: (1) the score becomes more exaggerated or (2) decreases in strength due to heightened awareness.
Personality is quite stable overall. Changes in scores on re-assessments of personality are rare, usually small, and often due to methodological reasons related to the assessment rather than meaningful psychological reasons specific to the individual. Thus, when a reassessment looks very different from the original result, it is best to verify whether these differences are driven by methodological reasons first (changes in forms or norms). Although less common, assessment results may change due to psychological factors (maturation, life events, or intentional efforts). However, such changes usually are not very large or meaningful, and are difficult to interpret.
*This post was co-authored by Hogan’s Chief Science Officer, Ryne Sherman, and Hogan’s Director of Global Learning, Jackie V. Sahm.
*Originally posted on Hogan assessment