Home / History of testing
A history of personality testing, 1917 to today
Personality assessment and career assessment developed as two separate traditions, for two different purposes, largely without reference to each other. That separation explains most of what’s wrong with the tests you’ve taken.
One thread was trying to find out what people are like, and starts in 1917 with a military screening questionnaire. The other was trying to work out what they should do, and starts a little earlier, in 1909. They only converge at the very end.
Thread one
What are people like?
The personality thread begins with a military screening problem and ends, seventy years later, with six factors that keep reappearing in every language anyone tests.
1917
The Woodworth Personal Data Sheet
The first widely used personality questionnaire, built to screen U.S. Army recruits for susceptibility to shell shock. It arrived too late to be used at scale, but it established the format the field still uses: self-descriptive statements, answered yes or no, scored by counting.
1936
The lexical hypothesis
Allport and Odbert extracted around 18,000 personality-descriptive words from an English dictionary. The premise: differences between people that matter get encoded in language, so the structure of the vocabulary should reveal the structure of personality. Nearly every model since is a compression of that list.
1943
The MMPI, and the clinical branch
The Minnesota Multiphasic Personality Inventory is built empirically — items are kept because they distinguish clinical groups, regardless of whether they appear to make sense. It becomes the dominant clinical instrument and permanently separates clinical assessment from normal-range personality measurement.
1944
The Myers-Briggs Type Indicator
Katharine Cook Briggs and Isabel Briggs Myers adapt Jung’s theory of psychological types into a practical questionnaire, working largely outside academic psychology. It becomes the most widely administered assessment in the world and, along the way, teaches the public to think about personality in types rather than degrees. Full discussion here.
1949
Sixteen factors
Cattell applies factor analysis to the lexical data and condenses thousands of trait words into sixteen underlying descriptors. The 16PF is the first serious attempt to derive personality structure from data rather than from theory, and it sets the method everything after it follows.
1961
Five factors, found twice
Tupes and Christal, working with U.S. Air Force data, find that Cattell’s sixteen reduce further to five recurring dimensions. The paper is published in a technical report and largely ignored for two decades — the result was right and arrived before the field was ready for it.
1984–1992
The Big Five consolidates
Through the 1980s the five-factor structure is rediscovered repeatedly by independent groups, in different languages and different data sets. Goldberg names it the Big Five; Costa and McCrae publish the NEO-PI-R. Openness, conscientiousness, extraversion, agreeableness and neuroticism become the working consensus — not because anyone designed them, but because they kept turning up.
2004
HEXACO adds a sixth
Ashton and Lee run the lexical studies across a wider set of languages and find a sixth factor the Big Five had folded into agreeableness: honesty-humility. It turns out to be the factor most strongly associated with counterproductive behavior at work — arguably the most consequential addition to trait psychology since, and the one almost nobody outside the field has heard of. Full explanation here.
Thread two
What should people do?
The vocational thread starts from a practical question, not a theoretical one, and it has always been more concerned with describing jobs than describing people.
1909
Parsons and the matching model
Frank Parsons, working with young people in Boston, sets out the idea that vocational guidance requires three things: understanding yourself, understanding the requirements of different lines of work, and reasoning about the relationship between the two. Every career assessment since is an implementation of that sentence.
1927
Strong Vocational Interest Blank
E. K. Strong builds the first widely used interest inventory on an empirical principle: compare a respondent’s likes and dislikes to those of people already satisfied in an occupation. It works, and it establishes interest — rather than aptitude — as the thing worth measuring for career fit.
1939
The Dictionary of Occupational Titles
The U.S. Department of Labor publishes structured descriptions of thousands of jobs. It is the first serious attempt to describe the work rather than the worker, and it is the ancestor of the database that makes modern occupational matching possible.
1959
Holland Codes (RIASEC)
John Holland proposes that people and work environments can be described with the same six dimensions — Realistic, Investigative, Artistic, Social, Enterprising, Conventional. Because person and job are coded in the same terms, fit becomes calculable. It may be the most useful single idea in the vocational literature.
1970
The Self-Directed Search
Holland makes RIASEC self-administered and self-scored, removing the counselor from the loop. Career assessment becomes something a person can do alone, which is how it reaches millions of people — and how it eventually reaches the internet.
1998
O*NET
The Department of Labor replaces the Dictionary of Occupational Titles with a relational database: tasks, work activities, work styles, work contexts, skills, and RIASEC interest codes for over 900 occupations, updated on a rolling schedule and published openly. It remains the most useful public dataset in the field, and it’s free.
Why the two threads stayed apart
Personality psychology was an academic discipline pursuing a structural question: how many dimensions are there, and what are they. Vocational psychology was an applied field serving counselors and schools, judged on whether its recommendations helped. Different journals, different conferences, different incentives.
The consequence is visible in the products. Personality instruments describe you well and say nothing credible about work. Career instruments produce occupational lists with a thin model of the person behind them. And the most popular product in either category — the four-letter type — comes from neither tradition, having been built in the 1940s outside both.
There was also a technical obstacle. Combining them means reading things people write, at volume, and turning unstructured language into structured scores. Until recently that required human scorers, which put it out of reach at any consumer price.
Present day
Where Haliscope sits
Haliscope’s eleven traits are its own, built as an evolution of the HEXACO and RIASEC models rather than taken from either, with an augmented assessment of morality and tolerance. It integrates job data directly from O*NET, and expands data collection from multiple-choice to open text, uploads, rankings, and other dynamic exercises.
Eleven traits are measured, but results focus on personal insights rather than categorizing respondents into types — and every report concludes with a roadmap that ties the insight to action.
Read the full methodology, limitations included →The two models this page ends on are set out in full on HEXACO and RIASEC: the models behind Haliscope.
A hundred years of research, in twenty minutes
Free preview of your results; $10.00 for the full report.
Start the assessment