Scale norming is where respondents use qualitatively different scales to answer the same question across survey waves. It makes responses challenging to compare intertemporally or interpersonally. This paper develops a formal model of the cognitive process that could give rise to scale norming in year on year responses to life satisfaction scale questions. It then uses this model to conceptually differentiate scale norming from adaptation and changes in reference points. Scale norming could make life satisfaction responses misleading with regards to the changing welfare of individuals. In particular, individuals who would say that their life is "improving" or "going well" might nonetheless give the same scale response year after year. This has negative implications for the use of scales in cost-benefit analysis and other welfarist applications. While there is already substantial empirical evidence for the existence of scale norming, its implications for welfare analysis are sometimes understated on the grounds that this evidence might simply be the product of errors of memory. The paper presents new empirical evidence for scale norming from two surveys (N1 = 278; N2 = 1050) designed such that errors of memory are an unconvincing explanation for the results.