- Personality tests ask about many aspects of your personality and compare your results to those of others.
- The results’ trustworthiness depends on how likely you are to get the same result again.
- The results should be similar if you retake the same test, take a different test, or have your friend rate your personality.
Many people take personality tests to learn about themselves. Of course, the tests only mirror what we think about ourselves, but the results can give a unique perspective on our traits. First, a good test asks about many aspects of our personalities, so its results provide a more comprehensive picture of us than any casual reflection. Second, the results are calculated by comparing our trait scores with those of many others.
For example, suppose you complete a Big Five personality test and get feedback on where you stand on each of the five traits compared to other people. Say you are high in extraversion (in the top third among other people), low in neuroticism (in the bottom third among other people), and medium on the other three traits.
But how much can you trust this feedback?
Test Reliability and Validity
Psychometricians who develop and deploy psychological tests often describe test quality with the terms reliability and validity.
Reliability refers to the consistency of test scores from one measurement to another, assuming the trait doesn't change. Inconsistencies may result from people changing their mood, not being quite sure about themselves, or just being distracted, for example.
Psychometricians can assess a test's reliability by asking many people to take it twice over a short period. The correlation between the scores from the two testing occasions is usually between 0.80 and 0.90. Psychometricians interpret like this: 80 percent to 90 percent of the variance in people's test scores is reliable. But what does the 90 percent reliable variance means for your score?
Validity refers to the degree to which a test measures what it's supposed to measure. Because we currently have few ways to measure personality traits without personality tests, there are not many independent yardsticks for verifying the measurements. So, validity is assessed indirectly.
For example, we can compare people's test scores to:
- Scores of other tests that presumably measure the same trait.
- Scores are based on ratings of other people, such as friends or partners.
The more similar people's scores are, regardless of which specific test they take or whether they or someone else rate their traits, the more we can believe that the scores say something valid about people's traits.
Of course, you should not expect your friend or partner to share the same information about you. But validating your own opinion against someone else's is often among the most straightforward ways to reality-check it. Besides, others are sometimes better positioned to rate our personality, seeing us in a broader context and not necessarily sharing our desire to overclaim desirable traits.
Scores of different personality tests designed to measure the same traits typically correlate between 0.60 and 0.80. Self-report scores usually correlate about 0.50 with scores based on ratings of other people.
But again, what does this say about the validity of your test feedback?
Here's What It Means for You
When unsure about a doctor's diagnosis, you are encouraged to seek a second opinion. Likewise, to think about the accuracy of your test feedback, ask this: how likely is it that I get a similar second opinion?
That second opinion can come from yourself on another day, from another test you complete about yourself, or from another person.
- Let's assume that scores of Big Five traits taken twice a week or two apart correlate about 0.90. This means about 75 percent probability that you will get the same feedback for any given Big Five trait twice, scoring either low, medium, or high on both occasions. It is rare that a high score changes to low or the other way, though—most changes involve hopping in or out of the medium category.
- Assuming that Big Five scores of different tests correlate around .70, there is about 60 percent probability that you will get the same feedback for any given Big Five trait from two separate tests of the same trait.
- Let's assume that the Big Five scores based on self-reports and ratings by good friends correlate about 0.50. Then, it is about 50 percent probability that you get the same feedback for any given Big Five trait, regardless of whether it was you or your good friend who rated the trait.
So, it is not uncommon to get a different opinion on your personality traits, either from yourself on another day, another test, or from other people who know you very well.
To get a more accurate picture of your traits, it is a good idea to take a test twice or take multiple tests, and see where the results agree. If you keep getting the same result, it probably is trustworthy. Also, if you are prepared to accept and learn from different views on your personality, it may be useful to convince your friend or partner to complete a test about you.
Let’s Not Make the Perfect the Enemy of the Good
The accuracy of personality testing is not comparable to measurement accuracy in hard science. This is understandable because the human mind and behavior are very, very complex, and personality traits only summarize broad patterns in them—as these appear to people themselves or others.
And yet tests can offer us a valuable perspective on our personalities by thoroughly summarizing a range of traits and showing how we stand in these compared to other people. This gives the results some objectivity compared to just casually mulling our personalities—one characteristic at a time and without exactly knowing where most other people stand in it.
Taking a test more than once, taking multiple tests, and having someone else complete a test about you may be especially helpful.
Mõttus, R. (2022). What does a correlation say about me? A tutorial on translating correlational research findings to their implications for individual people. https://doi.org/10.31234/osf.io/bpm9y