The objective of Study 1 is to identify the psychometric assessment and reporting practices of authors who have analyzed data from the Kessler Abbreviated Psychological Distress Scale (K6), collected in Canadian population surveys. The goal of Study 2 is to compare the performance of six reliability coefficients estimated from K6 data.

In Study 1, 71 publications using the K6 were reviewed and synthesized using an analysis grid. In Study 2, analyses were performed to compare the performance of the alpha coefficient to five other reliability coefficients using data from the 2009-2010 and 2013-2014 Canadian Community Health Survey (CCHS). Specifically, we estimated all six coefficient values, as well as their confidence intervals, regarding all respondents and respondent subgroups.

Out of the 71 publications identified in Study 1, only nine reported a reliability coefficient drawn from their own sample. Even though no condition essential to use of the alpha coefficient was mentioned, it was the only coefficient presented. In Study 2, the values of all the other coefficients were found to be higher than those of the alpha coefficient. Significant variations were found in some respondent subgroups.

Existing recommendations for the use of reliability coefficients are poorly implemented. It behooves authors to provide more information in their manuscripts, thereby enabling assessment of the psychometric qualities of the K6. The presentation of reliability coefficients for relevant subgroups and confidence intervals must also become standard practice, so that results can be more precisely interpreted.