Survey research is based on questioning and respondents typically answer to questions by picking one of several response options. These response options are labelled verbally with terms such as ‘very happy’ or ‘fairly happy’. Response scales differ in the number and wording of response options and this could affect the degree of happiness denoted by such words. If so, scores on differently worded questions on the same topic cannot be compared and this reduces the opportunities for research synthesis greatly. Several methods for transformation have been developed for dealing with that problem, among which the recently proposed ‘scale interval method’ in which judges rate the interval denoted by verbal response options on a continuous 0–10 scale. This method allows a view on the size of the problem. Application of the scale interval method to commonly used survey questions on happiness in Dutch language reveals considerable differences. The implications of this for research synthesis are discussed.