New Measures to Model Decision Making

Working Paper: NBER ID: w30839

Authors: Ingvild Almas; Orazio Attanasio; Pamela Jervis

Abstract: Most empirical work in economics has considered only a narrow set of measures as meaningful and useful to characterize individual behavior, a restriction justified by the difficulties in collecting a wider set. However, this approach often forces the use of strong assumptions to estimate the parameters that inform individual behavior and identify causal links. In this paper, we argue that a more flexible and broader approach to measurement could be extremely useful and allow the estimation of richer and more realistic models that rest on weaker identifying assumptions. We argue that the design of measurement tools should interact with, and depend on, the models economists use. Measurement is not a substitute for rigorous theory, it is an important complement to it, and should be developed in parallel to it. We illustrate these arguments with a model of parental behavior estimated on pilot data that combines conventional measures with novel ones.

Keywords: measurement; decision making; economic behavior; causal links

JEL Codes: A1; D1


Causal Claims Network Graph

Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.


Causal Claims

CauseEffect
traditional measurement approaches (C52)limited ability to identify causal relationships (C30)
narrow set of measures (C52)limited ability to identify causal relationships (C30)
broader approach to measurement (C52)more realistic models (E17)
integrating novel measures alongside conventional ones (C52)better capture complexities of individual decision-making (D91)
better capture complexities of individual decision-making (D91)improve identification of causal links (C32)
new measures (C29)identify causal links without relying on restrictive assumptions (C32)
design of measurement tools (C90)better understanding of causal links between variables (C29)
model of parental behavior (J12)improve identification of causal links (C32)

Back to index