Introduction

For all of our clients who have developed a customized target profile for one or more positions in your company, our report includes a number called the "Similarity Index" in the heading section of the 16-element Specific Personality Factors graph. This is a single number which is intended to reflect the overall degree of match between the candidate's test results and the specific target profile for that position. The maximum value for this number is 100, which represents a (very rare) perfect match. From a practical perspective, numbers in the 90's tend to be very good matches, candidates with scores in the 80's are likely appropriate for the position, and scores in the 70's warrant some caution.

How is the Similarity Index calculated?

Scientifically, the most powerful data included in a TCP report are the scores on the 16 specific personality trait scales. These are the basic building blocks of human personality, and Cattell's 16-factor theory, on which our testing is based, has been the focus of thousands of published research papers over the past five decades. Indeed, very recent research, to be published this year, demonstrates that the 16-factor theory is more accurate than all other major personality theories in predicting "real-life" behaviors such as job performance. Thus, when we construct a target profile, we utilize all 16 primary personality traits.

Briefly, the client company identifies the "peak performers" who will constitute the target sample. These should be the best in their position and should represent an ideal model for hiring new people. These 10-15 individuals complete the testing process, and we use their scores to calculate the average value (mean) for each of the 16 scales as well as the degree of variability of scores around the mean (standard deviation), adjusting for outliers. The most job-relevant scales are those with the smallest variation. For example, in one of our computer-programmer samples, the peak performers all had problem-solving ability scores within 1 or 2 points of each other; this indicates that problem-solving ability is highly relevant for this position. On the other hand, scales with a high degree of variation are much less relevant for the position. Our computer programmers exhibited a wide range of scores on the "Sensitivity" scale, indicating that this factor is much less relevant to performance in this position (it should be noted that Sensitivity is highly relevant in other positions).

To figure the Similarity Index, we take a candidate's 16 trait scores and compare them to the 16 trait scores of the target profile. In essence, we start with a value of 100, and then, for each of the 16 traits, compare the candidate's score to the target score, and subtract points from 100 to reflect any discrepancy. How much is subtracted from 100 depends on how different the candidate's score is from the mean, and, very importantly, how much variation in the target group occurred on that scale. For example, take the case where the mean value for the target group on a given scale is 7.5, and the candidate produced a 5.5, an absolute difference of 2. The candidate would be penalized much more heavily if the scale in question had very low variability in the target group than if it had high variability. It costs you more to be different on scales that matter as opposed to scales that don't matter as much.

As indicated above, our current formula for the Similarity Index produces values in a range that seem to make good sense to most of our clients. We almost never see a 100 or even a 99, but top candidates do tend to have scores in the 90's. However, it is important to note that many well-rated performers will have Similarity Index scores in the 80's as well. As scores approach the low 80's, and certainly the 70's, questions arise about the candidate's psychological similarity to the peak performer group.

Recommended use of the Similarity Index

Just as psychological testing itself should be seen as only one part of a much broader hiring process, the Similarity Index should be seen as just one component of the report, not the "bottom line." Statistical approaches such as this are most appropriate at the aggregate level, when dealing with large numbers of candidates over extended periods of time. For example, one of our clients, a large national financial services firm, has about 50-100 internal candidates interested in attending management training programs that can accommodate only 16 per session; they use the Similarity Index to sort the applicant pool, reducing a very large number to a much smaller, more manageable number of applicants, each of whom then receives individual attention. We have demonstrated that over the course of a year, use of the Similarity Index in this manner has resulted in a much higher proportion of successful participants in the training program.

Use of the Similarity Index will certainly be effective on a global level, such as reducing turnover, increasing divisional sales totals, and so forth, depending of course upon the criteria used in forming the target group. However, it is vulnerable to misinterpretation and misuse at the individual level. As noted above, we have general guidelines that are proving useful to most clients. Scores in the 90's are very good, scores in the 80's are likely good matches, and scores in the 70's and lower warrant a closer look and some caution. Thus, if a candidate earns a high score, it is generally safe to assume that they are very similar to your peak performers with regard to key psychological characteristics. Lower scores are much harder to interpret. If a candidate receives a Similarity Index of 78, for example, that should not immediately exclude them from consideration. The next step in analysis is to determine what has lowered their score. Look at the 16-factor graph, take the factors one at a time, focus on the blue highlighted scales, and consider where the candidate's score differs from the target range. Are they slightly off on most scales, or are there one or two scales on which they are substantially off, with good fits on the others? A thoughtful analysis of the actual data, using common sense and a knowledge of the job requirements and environment, may certainly override a low Similarity Index on many occasions.

Conclusions

The Similarity Index is a very powerful piece of information when used appropriately, but one which because of its power is subject to misuse. Take it as an hypothesis to be explored through more careful analysis of the 16-factor graph, a first step in the process of interpreting and using the report.