The research design for generating personality profiles for 739 different careers involved obtaining rating profiles for three sources: (1) regression equations that produced estimated personality test scores, (2) an SME (Subject Matter Expert) study in which personality scores were derived from expert judgments of occupational analysts, and (3) ongoing field research to develop rating profiles directly from respondents in various careers (a.k.a. career “incumbents”). Each of the methods has its advantages. Incumbents, for example, are able to provide personal and evidentiary sample scores, whereas analysts are able to discern relative differences across all occupations and adjust for sampling errors.

Personality career profiles are then used to in constructing the CareerMatch™ system (discussed in the next chapter), which allows users to understand how they compare to people in various careers in terms of personality.

The regression study met with reasonable success, yielding equations that accurately estimated scores for six of the eight personality dimensions, and somewhat accurately for the other two dimensions. Specifically, the procedure involved (1) obtaining a development sample from people in a representative subset of all careers, (2) regressing those scores on variables that measured other characteristics of people in careers, and (3) evaluating the degree of relationship between the predicted and actual scores. The development sample came from the main study (see previous chapter) which produced 48 “anchor profiles” (or statistically significant personality profiles for 48 careers.) Predictor variables used in developing the regression equations included data and ratings on worker interests, general educational development, temperaments, demographics, knowledge, skills and aptitudes. In total, 43 input variables were considered.

The first step was to reduce the number of variables while retaining as much of the variance present in the scores as possible. A principal components analysis of the 43 variables yielded a 9-component solution.

The next step was to regress each of the eight personality dimensions on the 9 principal components scores. Regression modeling involved an iterative process of splitting the development sample, building models, testing and analyzing coefficients of determination (R²). During each iteration, 65% the development sample was randomly drawn for building each model, while the remaining 35% was set aside for testing the resulting models. A regression equation was finally calculated for each of the eight personality dimensions. The mean R² values for the final eight models were respectable, with six of the eight personality types being predicted quite well.

Four occupational analysts were assembled to rate all 739 careers according to each occupation’s likely average personality preference. The analysts were trained in Jung’s personality theory and archetypal model, and each was given a scored copy of the final form of the Career Personality Test for review. Next, the analysts were given reports on 48 “anchor profiles,” comprising reference profiles for 48 careers, developed from the main study. Finally, the analysts were instructed to score each career by assigning 10 points within each of the four personality scales (e.g. 5 points to ‘Extrovert’ and 5 points to ‘Introvert’, 3 points to ‘Thinker’ and 7 points to ‘Feeler’, etc.) This exercise was spread out over 8 days. By the end, each analyst had independently rated 739 careers.

Outlier ratings were first examined and discussed before being included or discarded. After a thorough review of any statistical anomies, profiles were then developed from the mean scores of the four analyst ratings, and submitted for further review by statisticians. Interrater reliabilities were calculated for the mean of the analyst rating on each construct. The reliabilities for each construct mean were all above .71 with an overall mean reliability of .89, indicating strong internal consistency.

Data collection from career incumbents is an extensive and ongoing research effort being conducted virtually every day online as users take the MyPlan.com Career Personality Test. Data from users who volunteer their careers is anonymously compiled and periodically analyzed. Incorporating this incumbent data allows us to continuously refine career profiles, as well as be aware of any potentially evolving career profiles that may result from population changes over time.