Utveckling och psychometrisk testning av "Pain Impact Inventory" - en jämförelse mellan två olika ansatser
Psychometric evaluationStatisticsPsychometric testingPIINon-parametricStatisticsOperations researchProgrammingActuarial mathematicsStatistikOperationsanalysProgrammeringAktuariematematikMathematics and Statistics
The aim of this study was to develop and psychometrically evaluate an instrument for measuring the impact from chronic pain on daily life. Traditionally psychometric evaluation is done with parametric methods even if the items are ordinal data. In this study the instrument to be developed is solely based on ordinal data, and the psychometrically evaluation is done with both traditional methods (i.e. parametric statistics) as well as with non-parametric methods. These two different ways of psychometrically evaluation are then compared with each other. The instrument called ?Pain Impact Inventory? (PII) was evaluated in means of face validity, content validity, construct validity, criterion validity, and two measures for reliability. Reliability was measured trough test-retest (2-week interval) and internal consistency measured with Cronbach´s alpha. The results showed that PII had acceptable validity as well as reliability. The factor analysis indicated that the instrument had five sub-scales: Physical impact, Psychological impact, Impact on social life, Social support, and Control. However, the PII had in all 20 items which can be seen as too many, and a short form of PII may be desired. When comparing the two different approaches to do the psychometric evaluation some differences were found. Above all the main difference was that the non-parametric way was lacking of methods to analyse the results giving that some types of validity and reliability measures could not be calculated with this approach. It is, however, possible to do psychometric evaluations with only non-parametric methods even if some measures could not be calculated. Previous research have not shown that traditional psychometric evaluation is equal or more appropriate than non-parametric evaluation, and this study do not give any answer either to which approach that is most appropriate and/or efficient. However, a mix of these two approach may be the solution there the researcher mainly use the non-parametric methods but do some simple calculation as a complement to expand the list of methods available.