Determining minimally important score differences in scales of the Copenhagen Psychosocial Questionnaire.

AIM To determine minimally important differences (MIDs) for scales in the first version of the Copenhagen Psychosocial Questionnaire (COPSOQ). METHODS Data were taken from two separate studies: a national population survey (N = 1062), and an intervention study at 14 workplaces (N = 1505). On the basis of the population survey, the MID for each COPSOQ scale was calculated as one-half of a standard deviation (0.5 SD). For the core COPSOQ scales on ''Quantitative demands'', ''Influence at work'', ''Predictability'', ''Social support (from colleagues and supervisors, respectively)'', and ''Job satisfaction'', the MIDs were evaluated in the intervention study, where score differences for the scales were linked to the respondents' global self-evaluation of the impact of the interventions. The scales were scored from 0 to 100 in both studies. RESULTS The MIDs calculated as 0.5 SD were, on average, 9.2 (range 6.8-14.9) for the long version scales, and 10.8 (range 7.6-14.9) for the medium-length version scales. The analysis of the self-evaluated changes on the scale scores for the core COPSOQ scales showed that the anchor-based estimates of MID were generally lower than 0.5 SD. CONCLUSIONS We recommend the following MID values for the COPSOQ scales: ''Quantitative demands'', 0.3 SD; ''Influence'', 0.2 SD; ''Predictability'', 0.3 SD; ''Social support from colleagues'', 0.3 SD; ''Social support from supervisor'', 0.7 SD; and ''Job satisfaction'', 0.4 SD. For all other COPSOQ scales, where we do not have anchor-based results, we recommend the conventional MID value of 0.5 SD.