Mate Grader as a test originated as an experimental research tool in a private practice addressing relationship issues. Later, it became a beta-research project at a Mid-West University in a Psychological Test and Measurement course examining test construction, reliability and validity. Overall, this project investigated if this simple test could predict divorce in the general public. Although the test was able to predict divorce statistically above chance; results suggest that predicting the strength of relationship had greater practical implications. Therefore, results were re-scaled to fit a standard grading scale (i.e. A, B, C, D, and F) because this traditional measure is universally understood.
For clarification, many participants (>27%) openly admitted to being involved in a bad/terrible relationship, but because of personal, financial, family or religious reasons would not end the relationship or divorce. In short, Mate Grader is a predictive validity test assessing the potential long-term relationship strength. The average participant in this data pool reported being in a relationship (married or not married) for more than eight years (N = 832).
The test-retest reliability coefficient is .98 for self-evaluation and .91 for rating their mate (N = 241, Interval: 35 days). The overall test reliability is .94.
The established hit-ratio is: 76% when successfully predicting a reported bad relationship; and, 81% when successfully predicting a reported good relationship. Therefore, although the instrument is statistically valid, that is, measuring what it is intended to measure – it does not predict with 100% accuracy.
Mate Grader was not designed as a face-valid test. With a face valid test it is obvious what the test is measuring. Rather, this is an empirically valid test, that is, the items, although on the surface do not obviously look as though they are related to relationship strength; nevertheless, will statistically predict relationship strength. For clarification, an empirically valid test relies on outcomes and predictive data to establish validity estimates.