Next Generation of Personality Tests:
Predicting Performance not Describing Personality

In the beginning...

First generation personality testing is typically dated to the work of Raymond Cattell in 1949, when he developed the 16PF scales. He wanted to describe and classify groups of personal traits into clinically useful categories. As with most pioneers, he had the vision to see that it was possible to accomplish this task and expended the time and effort necessary to develop the instrument.

The Cattell model was conceptually simple:

  • Choose questions that accurately sample the traits you are seeking to measure
  • Group those clusters of traits into profiles with descriptive, and recognizable labels.
  • Ascribe certain behavioral patterns to people with similar profiles.

It took considerable genius to make the 16PF work. Thousands of questions had to be devised, tested and winnowed to a manageable number. Statistically, every question introduced a tiny amount of error. Grouping into profiles was laborious and introduced even more statistical error. Attributing behaviors to profiles continued to introduce more error. Even so, the instrument was considerably better than anything known before and was useful – especially to clinicians.

We are now in the sixth decade of personality testing and little has changed. We have gotten considerably more sophisticated in the way we perform all three steps of the Cattell Model; better questions, more precise profiles and more narrow behavioral patterns. Even so, the field has not gotten beyond describing behavior.

More important, we have not moved into the area of behavioral prediction. It is one thing to describe clusters of behavioral tendencies – it is another to predict future behavior. Cattell’s model was never designed to do this task.

Personality Testing and Performance

Most professionals assume that there is a relationship between personality and performance. People can differ in their ability to perform even though they have roughly the same basic skills, abilities and experience. Frequently, people with lesser skills outperform those with greater skills. Often these differences in performance are attributed to characteristics such as energy, motivation, drive, desire to succeed, aggressiveness, perseverance, sociability, or dozens of other personality traits.


Components of Performance
(Although the pie chart shows equal sections,
it is unlikely that they would be equal and without overlap)

While skills and knowledge are frequently measured objectively, personality has been dealt with as descriptive and subjective in nature. This subjective nature makes research difficult. It is this personality component of performance that the Rosner-Edwards Protocol focuses on.

Problems With Using Profile-based Personality Testing

Other professionals understood the importance of this personality component and began to use profile-based models to try to predict performance in a “round-about way”. After developing a profile-based test, they would administer it to the “successful performers” in an organization. They would develop an average profile of these successful performers and use this profile as the criterion measure. It rested on the assumption that “the closer a candidate matched the profile, the better the performance”.

The obvious problem with this assumption was that Human Resource professionals know that a) many people who “don’t fit the mold” but still turn out to be high performers, and b) many who do fit the mold are often low performers.

Experience has shown that the world is not as simple as “X amount of Aggressiveness and Y amount of Self-awareness, … = A successful performer”. Profiling is too limiting and too static. There undoubtedly are multiple constellations of traits that are associated with successful performance, so a single "best" profile is unlikely. Additionally, a high amount of trait A and a medium amount of trait B might mean one thing when trait C is high but another thing when trait C is low. Add 10 more traits and the inter-correlations become unmanageable.

The Rosner-Edwards Protocol: The Next Generation of Personality Tools

Dr. Phillip E. Rosner believed that the direct prediction of performance from personality test questions (without the intermediate step of profiling) was clearly the way to eliminate much of the testing error inherent in the Cattell model. The concept made sense, but was ruled “impossible” by authorities in test construction, statistics and programming. The state of statistics, he was told, did not have the tools to accomplish the task. For almost three decades that situation continued to exist.

Dr. Jack L. Edwards, a world recognized expert in Artificial Intelligence (AI), had developed the statistical tools which could make Dr. Rosner’s concept feasible. Working together, they developed the Rosner-Edwards Protocol (REP) (patent pending).

A Deceptively Simple Process:

Performance + Performers + REP = Prediction

In essence the REP is a way to predict the personality component of performance of any measurable behavior. The chart below illustrates how the process operates:


The Rosner-Edwards Protocol

1. Choose the Target Performance

This is the most complicated part of the process for the survey user because there are virtually no limits to what behaviors can be predicted. The general rule is “If it can be measured, it can be predicted”. Here are some examples:

  • For salespeople - sales revenue, profit generated from sales revenue, numbers of new clients accounts opened, longevity, or weighted combinations of them.
  • For college admissions – grade point average, retention (e.g.- freshmen returning for their sophomore year), graduation, or a weighted combination of them.
  • For seniors being recruited for companies – success in terms of ratings or ranking of on-the-job performance at the company, either in general or by occupation.

2. Choose the People for the Normative Group

The single largest source of error in any previous test was the difference between the normative group and the group to which the test is administered.

With the REP, those groups are identical. The normative group are your employees. You select a sample 25, 50 100, etc. of the highest performers in your company. You then select an equal number of the lowest (still employed) performers. Their performance data is sent to our servers and is entered in a separate database.

3. Develop the Unique Selection Survey (USS)

  • Each of the people in the normative group takes the Self-Assessment Survey (SAS). The test is taken on-line 24/7/365 and typically takes about an hour. All answers are entered into the same database containing their performance data.
  • Our proprietary artificial intelligence (AI) engine selects the 80 – 120 questions and answers which best differentiate between the high- and low-performers. That becomes the unique selection survey for the specific use selected and for your organization only (USS).

4. The USS becomes part of your selection system.

All candidates now take the USS along with whatever other testing and interviewing is typically done. The decision-makers now have a new, objective measure of probability of high-performance based on the candidates personality. They will gradually become more confident in its use as they see the results they achieve.

5. The USS is systematically reformulated (“gets more accurate”)

Periodically, at the organizations discretion, new performance data from each of the people who have taken the USS is transmitted to our servers. The more frequently the data is updated, the better – preferably once per quarter. The USS is re-calculated in order to become more and more accurate at assessing the probability of high performance.

Benefits of the Rosner-Edwards Protocol

Over the previous Generation of Selection Tools

Personality Tests - A Good Tool --- But Not Good Enough – Until Now!

Human Resources professionals have known the value of personality testing for decades. Having a relatively objective assessment of personality traits has always been seen as better than a non-professional’s personal guess at the “chemistry” of a person

These same professionals have always been uncomfortable with the underlying assumption that one profile was “right” for a particular job. They could see two rather obvious facts:

  1. People with different personality profiles could be successful at the same job, and
  2. People with basically the same personality profile could be either successes or failures.

Obviously, there is more to predicting success than personality profiles.

Now that the Rosner-Edwards protocol has been developed, HR professionals have:

  • A reliable, inexpensive way to assess the personality factor and directly relate it to performance in their organization.
  • A survey that gets even more accurate as it is used.
  • Convenience: The survey is administered on-line, 24/7/365 and scores are reported immediately.
  • The ability to have different personality surveys for different jobs, inexpensively.
  • More peace of mind: Because the normative group are the organization’s employees, there is never a question of appropriateness of the instrument.
  • A Low cost: Each administration is equal to or less than other non-dynamic tests.