Posted on

Victor Lange and Thor Grünbaum – “Measurement Scepticism, Construct Validation, and Methodology of Well-Being Theorising”

A young pregnant woman is holding a small balance for weighing gold. In front of her is a jewelry box and a mirror; on her right, a painting of the last judgment.
“Woman Holding a Balance” (c. 1664) Johannes Vermeer

In this post, Victor Lange and Thor Grünbaum discuss the article they recently published in Ergo. The full-length version of their article can be found here.

Many of us think that decisions and actions are justified, at least partially, in relation to how they affect the well-being of the involved individuals. Consider how politicians and lawmakers often justify, implicitly or explicitly, their policy decisions and acts by reference to the well-being of citizens. In more radical terms, one might be an ethical consequentialist and claim that well-being is the ultimate justification of any decision or action.

It would therefore be wonderful if we could precisely measure the well-being of individuals. Contemporary psychology and social science contain a wide variety of scales for this purpose. Most often, these scales measure well-being by self-reports. For examples, subjects rate the degree to which they judge or feel satisfied with their own lives or they report the ratio of positive to negative emotions. Yet, even though such scales have been widely adopted, many researchers express scepticism about whether they actually measure well-being at all. In our paper, we label this view measurement scepticism about well-being. 

Our aim is not to develop or motivate measurement scepticism. Instead, we consider a recent and interesting reply to such scepticism, put forward by Anna Alexandrova (2017; see also Alexandrova and Haybron, 2016). According to Alexandrova, we can build an argument against measurement scepticism by employing a standard procedure of scientific psychology called construct validation. 

Construct validation is a psychometric procedure. Researchers use the procedure to assess the degree to which a scale actually measures its intended target phenomenon. If psychologists and social scientists have a reliable procedure to assess the degree to which a scale really measures what it is intended to measure, it seems obvious that we should use it to test well-being measurements. For the present purpose, let us highlight two key aspects of the procedure. 

First, construct validation utilises convergent and discriminant correlational patterns between the scores of various scales as a source of evidence. Convergent correlations concern the relation between scores on the target scale (intended to measure well-being) and scores on other scales (assumed to measure either well-being or some closely related phenomenon, such as wealth or physical health). Discriminant correlations concern non-significant relations between scores on the target scale and scores on scales that we expect to measure phenomena unrelated to well-being (e.g., scales measuring perceptual acuity). When assessing the construct validity of a scale, researchers evaluate a scale by considering whether it exhibits attractive convergent correlations (whether subjects with high scores on the target well-being scale also score high on physical health, for example) and discriminant correlations (e.g., whether subjects’ scores on the target well-being scale have significant correlations with perceptual acuity).

Second, the examination of correlational patterns depends on theory. Initially, we need a theory to build our scale (for instance, a theory of how well-being is expressed in the target population). Moreover, we need a theory to tell us what correlations we should expect (i.e. how answers on our scale should correlate with other scales). This means that, when engaging in construct validation, researchers test a scale and its underlying theory holistically. That is, the construct validation of the target scale involves testing both the scale and the theory of well-being that underlies it. Consequently, the procedure of construct validation requires that researchers remain open to revising their underlying theory if they persistently observe the wrong correlational patterns. Given this holistic nature of the procedure, correlational patterns might lead to revisions of one’s theory of well-being, perhaps even to abandoning it. 

The question now is this: Does the procedure of construct validation provide a good answer to measurement scepticism about well-being? While we acknowledge that for many psychological phenomena (e.g., intelligence) the procedures of construct validation might provide a satisfying reply to various forms of measurement scepticism, things are complicated with well-being. Here the normative nature of well-being rears its philosophical head. We argue that an acceptable answer to the question depends on the basic assumptions about the methodology of well-being theorising. Let us clarify by distinguishing between two methodological approaches.

First, methodological naturalism about well-being theorising claims that we should theorise about well-being in the same way we investigate any other natural phenomenon, namely, by ordinary inductive procedures of scientific investigation. Consequently, our theory of well-being should be open to revision on empirical grounds. Second, methodological non-naturalism claims that theorising about well-being should be limited to the methods known from traditional (moral) philosophy. The question of well-being is a question about what essentially and non-derivatively makes a person’s life go best. Well-being has an ineliminative normative or moral nature. Hence, the question of what well-being is, is a question only for philosophical analysis.  

The reader might see the problem now. Since construct validation requires openness to theory revision by correlational considerations, it is a procedure that only a methodological naturalist can accept. Consequently, if measurement scepticism is motivated by a form of non-naturalism, we cannot reject it by using construct validation. Non-naturalists will not accept that theorising about well-being can be a scientific and empirical project. This result is all the more important because many proponents of measurement scepticism seem to be methodological non-naturalists.  

In conclusion, if justifying an action or a social policy over another often requires assessing consequences for well-being, then scepticism about measurement of well-being becomes an important obstacle. We cannot address this scepticism head-on with the procedures of construct validation. Such procedures assume something the sceptic might not accept, namely, that our theory of well-being should be open to empirical revisions. Instead, we need to start by making our methodological commitments explicit. 

Want more?

Read the full article at


  • Alexandrova, Anna (2017). A Philosophy for the Science of Well-Being. Oxford University Press. 
  • Alexandrova, Anna and Daniel M. Haybron (2016). “Is Construct Validation Valid?” Philosophy of Science, 83(5), 1098–109. 

About the authors

Victor Lange is a PhD-fellow at the Section for Philosophy and a member of the CoInAct group at the Department of Psychology, University of Copenhagen. His research focuses upon attention, meditation, psychotherapy, action control, mental action, and psychedelic assisted therapy. He is a part of the platform Regnfang that publishes podcasts about the sciences of the mind.

Thor Grünbaum is an associate professor at the Section for Philosophy and the Department of Psychology, University of Copenhagen. He is head of the CoInAct research group. His research interests are in philosophy of action (planning, control, and knowledge), philosophy of psychology (explanation, underdetermination, methodology), and cognitive science (sense of agency, prospective memory, action control).