Scoring of the NPA test
© A.M. Benis, Sc.D., M.D.
Evaluation of any kind of test is rarely straightforward, as one would ideally have to consider two separate aspects of testing that are sometimes antagonistic to each other: the sensitivity and specificity of the test -- for each of the major character types. Sensitivity is a measure of the frequency of false negatives, while specificity is a measure of false positives. (For example, if you are an N type and the test states that you are not, that is a false negative. If the test says that you are an N type but you are not, that is a false positive). In the bar graph showing your probable character type you may see a single type that is highly probable, but there may be relatively high scores for one or two of the other types as well.
The test was scored by the averaging of five different statistical methods employing "square scores," variances, correlation coefficients, and nineteen different standard tests. For example, a "square score" for a given NPA type was determined from the sum of squares of the differences of the subject's responses from the "expected values" of the standard test for that type. The "square scores" for each type were then transformed to relative probability values with the use of the reciprocal function. The effect of the squaring technique was to amplify responses concordant with the most probable NPA type. The probabilities of NPA character type and of the three traits are displayed on the bar graphs.
The S score was calculated from questions emphasizing submissiveness, social anxiety and depression. The S-score was determined from the S category of questions as the sum of squares divided by the maximum possible score, times 100. Four categories of S-score are considered: (1) less than 20 (dominant types), (2) between 20-30 (transition range), (3) between 30-60 (moderately elevated), and (4) greater than 60 (high).
The temperament T score was calculated from questions that emphasized dominant behavior in N, NA, NPA and A character types: individuals who answer the selected questions will tend to score higher on the T scale. The S and T scores are mathematically independent, the two scores being calculated from separate subsets of the list of fifty questions.
The focus F score was calculated from questions that emphasized behavior in the NP and PA character types. The basis of the F score was a set of questions separate from those used in the computation of the S and T scores.
The N, A, R and C scores were calculated from correlation coefficients. For example, the N score evaluated the subject's agreement to nineteen different standard tests, compared to the response of an N type to the nineteen tests
The bar graph giving the probabilities of the three NPA traits is most applicable to the six dominant types. If the S score is elevated (>20) then the estimate of the probability of traits N, P and A may not be accurate, since the traits N and/or A may have been suppressed by genetic or environmental factors. Values of probability generated by the statistical techniques are inherently dependent on (1) the particular questions that were chosen for the test, and (2) the varied ways that the questions will be interpreted and responded to by various subjects. Hence, the probability values finally produced are relative ones only. The values are relevant only to this particular test, and they do not have any other significance in a quantitative sense.
Further details of our statistical procedures may be found on "Sheet 2" of the English version of the NPA test.