Investigating subjective aspects of lighting, such as preference and acceptance, involves evaluations and ratings by human subjects. The data from the subjects is gathered with questionnaire techniques. This work reviews the literature published on lighting field on methodology for subjective lighting studies. The review continues to questionnaire research in general, and discusses several factors which may influence the results or the quality of the data, such as the form of response options, labelling the response alternatives, the number of response option categories, whether or not there should be an option for a neutral answer. Response styles are addressed by a literature review, and a discussion on how they may affect a lighting acceptance or preference study. The work reviews also a number persisting controversies related to statistical analysis of questionnaire data: null hypothesis significance testing, adjusting the significance criterion for multiple comparisons, and analysing ordinal data with parametric statistical methods. A review of arguments from both sides for each controversy is presented, and the reasons why these controversies persist are discussed. Twenty subjective lighting studies published since 2005 are analysed for the methodology used and the way the methodology and results are presented. The finding was that about half of the studies considered did not report their methods, analysis or results in sufficient detail. The statistical power is also often quite low. Two experiments were conducted to see how varying the response format affects the results gained from a lighting experiment, and to evaluate if the choice of parametric or non-parametric statistical method effects the conclusion. The findings were, that reversing the response option order did not affect the conclusion or characteristics of the data, but providing labels for every response option resulted into a slightly higher scores than when only the end-points were labelled. Both parametric and non-parametric statistical methods led to the same conclusion, and thus this work does not support the view that parametric methods are not appropriate for questionnaire data.
|Translated title of the contribution||Metodologisia kysymyksiä valaistuksen hyväksymistä ja siihen liittyviä mieltymyksiä tutkittaessa|
|Publication status||Published - 2015|
|MoE publication type||G4 Doctoral dissertation (monograph)|
- response styles
- level of measurement