Dissatisfaction or Satisfaction

 ref: debate on front page

 One of the assumptions of research in satisfaction is that a bipolar scale (satisfied-dissatisfied) is sufficient to analyse the drivers to both.  But is this true? Or are respondents anchored by asking for their scores on satisfaction such that drivers to dissatisfaction are never picked up? Or indeed is it that most things are done in an OK fashion such that dissatisfaction is never usually in the data until something goes wrong i.e., like many things it is latent.   If we can actually demonstrate that satisfaction measures are really about measuring reductions in dissatisfaction then there are significant issues in the way it is currently measured and used.  Certainly on the face of it this would seem to be a reasonable hypothesis bearing in mind Prospect Theory (Kahneman and Tversky) (losses are felt more than gains); gains represent something diffuse and value-add whereas negatives are more pointed and direct (Frederickson) and that regressions may be by consequence less linear and more curvilinear (sensitive to the negative but not the positive). If this holds true then constant attempts to raise satisfaction levels are misdirected; working maybe in the short-term but holding little relevance to behaviour (unlike dissatisfaction). One way to test this is to actually ask directly the dissatisfaction question as well as the satisfaction.  If there is anchoring on satisfaction then there may also be anchoring on dissatisfaction. Forcing variance in this way may determine current hidden drivers to dissatisfaction – even if the latent issue remains unresolved: unless you are using conjoint techniques which have other issues. This is exactly what we found in reviewing a dissatisfaction dataset in Insurance.  Here call centre was found to drive both satisfaction and dissatisfaction: an effect hidden on a bipolar scale. The conclusion is that satisfaction acts as a flagging system for blandness and dissatisfaction should be used – although sparingly if we consider the problem of latents. Incentivising for increases in satisfaction actually depends on the level of dissatisfaction: fine for improving scores up to a point where the relationship to value creation starts to breakdown through its non-linear relationship.