STATISTICAL RELIABILITY AND VALIDITY

There are 3 major indicators for assessing the validity and reliability of any personality assessment. Below you can find their explanation and see how HIGH5 scores with respect to each one of them.

Introduction

While a solid and thorough conceptual framework is essential for the development and reliability of any psychometric instrument, it is equally important that the instrument correctly measures what is supposed to, thus being empirically and statistically valid.

There are 3 major indicators for assessing the validity and reliability of any personality assessment. Below you can find their explanation and see how HIGH5 scores with respect to each one of them.

DISCRIMINANT VALIDITY

The discriminant validity analysis evaluates whether scales measured by the instrument are indeed distinct, independent and uncorrelated. In other words, the analysis should answer the question whether the test measures unique strengths or an undefined mix of unidentified factors. Are we certain that when HIGH5 evaluates your Deliverer strength, it does not unknowingly evaluate half of your Empathizer strength?

The maximum accepted value for this indicator is 0.7-0.8. Higher scores mean that scales overlap to such an extent that they become indistinguishable. In practice, it would imply that there is not too much difference between two strengths evaluated.

The maximum value of this indicator for HIGH5 strengths is 0.29. It is significantly lower than the accepted maximum for such assessments. It means that all 20 strengths evaluated by HIGH5 test do not influence each other and are not correlated to the extent that would question the discriminant validity of the assessment.

INTERNAL CONSISTENCY

With the discriminant validity present, the second step is to check the internal consistency of scales. It makes sure that questions asked for evaluating a specific strength indeed evaluate this strength. Are we certain that questions A and B are both measuring the same strength of yours?

A coefficient called Cronbach’s Alpha is designed to measure exactly that. The minimum acceptable value for this indicator is 0.7. Any lower values imply that questions might not measure the intended strength.

The minimum value of Cronbach’s Alpha for HIGH5 questions and strengths is 0.87. It is significantly higher than the accepted minimum for such assessments. It means that all questions asked during the HIGH5 assessment measure the intended strengths and not others.

TEST-RETEST RELIABILITY

If the psychometric instrument is precise today, are we sure that its results will remain after some time? This is what the final indicator of the quality of the assessment looks at – the correlation between the test results today and the results of retaking it after a significant period of time.

As personality and individual strengths are subject to development depending on changing circumstances, it is important to choose the right time horizon to analyze this indicator. Taking periods of time of more than a year would assume that the test taker’s environment and mindset remain unchanged for one year, which is close to impossible in real life.

The best practice time-horizon across personality assessments is to look at 6-month periods and less, with a minimum acceptable correlation of 0.7. The higher the value is, the more consistent the test results are across time.

The minimum value of this indicator for HIGH5 test for the 6-month retesting window is 0.81. It is statistically significantly higher than the minimum acceptable value. In practice, this means that if you were to retake the HIGH5 test in 6 months or less, 81% percent of your results will remain the same, compared to your test results today.

CONCLUSION

The data on 3 major indicators for HIGH5 confirms that:

HIGH5 evaluates 20 distinct and independent strengths following its conceptual framework;
All evaluated strengths are internally consistent;
HIGH5 results remain stable even after 6 months.

All presented indicators are presented for the English version of the HIGH5 test and are based on 312,000 observations.

If you have additional ideas on how HIGH5 can increase any of the above-mentioned indicators and thus increase its precision, reliability and validity, please feel free to submit them here.