Last update:
12-Nov-2004

The purpose of this study was to implement documented statistical tests of data quality, scaling assumptions and reliability for the Greek version of the well-known health survey, SF-36, which is used widely for selfreporting of Health Related Quality of Life (HRQoL). This procedure assists in enhancing confidence in using the instrument in non English-speaking settings. The Greek SF-36 questionnaire was administered via interview to a representative sample (n=126) of inhabitants of the Greek island Ios, corresponding to approximately 10% of the total island population. Statistical tests of data quality and completeness, response frequencies, item means, internal item consistency, discriminant validity and scale reliability were conducted. The results were, to a great extent, comparable with those obtained from similar studies in other countries. For all items, all response choices were used and patterns of item means appeared as expected. Thirty four (of 36) items correlated (Pearson’s r>0.40) with the hypothesized scale, satisfying the internal consistency criterion, and all item-scale correlations were statistically significant at the 0.01 level. Two hundred and eighty discriminant validity tests were conducted, and in only 8 cases did an item correlate higher with a competing scale, than with its own. These items belonged to three scales, General Health, Vitality and Social Functioning. This finding is attributed to the high standard error associated with the relatively small sample size. The minimum scale reliability level of 0.70 (Cronbach’s alpha) suggested for group level comparisons was satisfied for six of the scales, with the Social Functioning and General Health scales scoring 0.69 and 0.64 respectively. These preliminary results are encouraging as far as using the SF-36 in Greek studies is concerned, but in order to validate any conclusions, the procedures must be repeated with larger samples more representative of the Greek population.

Key words: Quality of life, Reliability, SF-36, Validity.