The result comes back as SAE, or RIA, or one of the other combinations, and under it sits a list of job titles. Twenty of them, in a neat column. Two are jobs the reader has already done and disliked.
That column is the part everybody screenshots, and it is the part with the least support behind it.
26
studies in the meta-analysis of interest fit and job satisfaction
.17
the correlation it found
105
studies in the 2020 review that revisited the question
60
studies behind the finding that interests predict performance
What the code measures, precisely
An interest inventory asks which activities appeal. Not which ones somebody is good at, not which ones pay, not which ones a person values. Only what pulls.
John Holland's contribution was to argue that both people and work environments can be described with the same six categories, and that people look for environments where they can exercise the skills and express the attitudes they prefer. Congruence is the word for the match between the two, and Holland's theory predicted that congruent people would be more satisfied, more stable in their choices and higher performing.
Three predictions. They have not held up equally.
The satisfaction claim is the weakest one
Tsabari, Tziner and Meir pulled together 26 studies covering 53 samples and 6,557 respondents in 2005 and found a correlation of .17 between congruence and satisfaction, with a confidence interval wide enough to include zero at one end.
- Interest congruence3% · 3%
- Everything else97% · 97%
- Interest congruence: 3%
- Everything else: 97%
Three percent. Hoff and colleagues revisited the question in 2020 with 105 studies and 39,602 participants spanning research from 1949 to 2016, and reached the same shape of answer. Interest fit predicts job satisfaction significantly and modestly, and the relation is strongest for satisfaction with the choice of occupation and with the organisation rather than with the work itself.
The practical translation is unwelcome to everyone who sells one of these tests, this site included. Whether somebody enjoys their job on a Wednesday is mostly decided by things an interest inventory never asked about, starting with the manager and the people at the next desk. The trouble with following a passion covers the same finding from the other end, where the advice to chase what you love keeps producing people who did and are still unhappy.
The prediction that did hold is the one nobody markets
Here the literature turns in an unexpected direction.
Nye, Su, Rounds and Drasgow meta-analysed 60 studies and roughly 568 correlations in 2012 and found that interests do predict performance and persistence in work and academic settings. More surprisingly, congruence indices predicted performance more strongly than raw interest scores did, which is what Holland's theory said should happen and what the satisfaction literature had failed to deliver.
So a Holland code is better at forecasting whether somebody will do the work well and stay in it than at forecasting whether they will be happy doing it. That is close to the reverse of how a result page presents itself, and it is the single most useful thing to know about a code.
| The question | How well the code answers it |
|---|---|
| Will these activities interest me | Well, since that is the measurement |
| Will I stay in this field | Reasonably, interests predict persistence |
| Will I perform in it | Reasonably, and better through congruence than through raw scores |
| Will I be satisfied at this job | Weakly, at around three percent of the variance |
| Should I accept this specific offer | Not at all, since the offer is not the occupation |
Why the job list is the weakest artifact on the page
A job list is produced by matching a person's three letters against codes assigned to occupations, and the assignment is the problem. The United States Department of Labor maintains the reference version of those codes inside O*NET, where each occupation carries interest ratings derived from the people who hold it. That is an average across incumbents, which means the code describes who currently does the job rather than what the job is, and it arrives at a single code for a title that covers wildly different work.
Consider nursing, or software engineering, or teaching. Each has a code. Each also contains roles that share almost nothing on a daily basis, and the difference between an emergency department and a research ward is larger than the difference between two occupations elsewhere on the list.
The other half of the problem is what the list omits. An occupation whose code matches poorly can still contain the exact role somebody wants, because organisations are not uniform and the work a person actually does inside a title is negotiable. Reshaping an existing job around what it is missing has a better evidence base behind it than switching occupations does, and it never appears at the bottom of an interest result.
Who the hexagon was validated on
One more limitation, and it matters for anyone taking the test outside North America.
Holland arranged the six types in a circle, with adjacent types more similar than opposite ones, and that circular order is what makes a three-letter code interpretable at all. Rounds and Tracey tested the arrangement in 1996 against 76 international correlation matrices from 18 countries, 20 matrices from United States ethnic minority samples, and a benchmark of 73 United States matrices. The circular order held in the benchmark. It was not supported cross-culturally, with only a handful of countries reaching acceptable fit.
That does not make the six categories meaningless elsewhere. It does mean the geometry, and every interpretation that depends on which letters sit next to each other, was established on a specific population and has not generalised cleanly.
The site's walkthrough of using the model to pick a first job covers the constructive version of all this for somebody starting out. The caution belongs next to it, because a list of twenty titles carries an authority that a correlation of .17 does not earn, and the people most likely to act on the list are the ones with the least experience to check it against.
Sources
- Tsabari, O., Tziner, A., Meir, E. I. (2005). Updated meta-analysis on the relationship between congruence and satisfaction. Journal of Career Assessment, 13(2).
- Hoff, K. A., Song, Q. C., Wee, C. J. M., Phan, W. M. J., Rounds, J. (2020). Interest fit and job satisfaction: a systematic review and meta-analysis. Journal of Vocational Behavior, 123.
- Nye, C. D., Su, R., Rounds, J., Drasgow, F. (2012). Vocational interests and performance: a quantitative summary of over 60 years of research. Perspectives on Psychological Science, 7(4).
- Rounds, J., Tracey, T. J. (1996). Cross-cultural structural equivalence of RIASEC models and measures. Journal of Counseling Psychology, 43(3).
- Su, R., Rounds, J., Armstrong, P. I. (2009). Men and things, women and people: a meta-analysis of sex differences in interests. Psychological Bulletin, 135(6).
- National Center for O*NET Development. O*NET Interest Profiler.