ORIGINAL RESEARCH ARTICLE
Accounting for measurement error in human life history trade-offs
using structural equation modeling
Section of Ecology, Department of
Biology, University of Turku, 20014,
Department of Biology, University of
Turku, 20014, Finland
Samuli Helle, Section of Ecology,
Department of Biology, University of
Turku, 20014 Turku, Finland.
o, Grant/Award Numbers:
085253 and 086809
Objectives: Revealing causal effects from correlative data is very challenging and a
contemporary problem in human life history research owing to the lack of experimen-
tal approach. Problems with causal inference arising from measurement error in
independent variables, whether related either to inaccurate measurement technique or
validity of measurements, seem not well-known in this field. The aim of this study is
to show how structural equation modeling (SEM) with latent variables can be applied
to account for measurement error in independent variables when the researcher has
recorded several indicators of a hypothesized latent construct.
Methods: As a simple example of this approach, measurement error in lifetime allo-
cation of resources to reproduction in Finnish preindustrial women is modelled in the
context of the survival cost of reproduction. In humans, lifetime energetic resources
allocated in reproduction are almost impossible to quantify with precision and, thus,
typically used measures of lifetime reproductive effort (e.g., lifetime reproductive suc-
cess and parity) are likely to be plagued by measurement error. These results are
contrasted with those obtained from a traditional regression approach where the single
best proxy of lifetime reproductive effort available in the data is used for inference.
Results: As expected, the inability to account for measurement error in women’slife-
time reproductive effort resulted in the underestimation of its underlying effect size
on post-reproductive survival.
Conclusions: This article emphasizes the advantages that the SEM framework can
provide in handling measurement error via multiple-indicator latent variables in
human life history studies.
It is a commonly held view in evolutionary biology that
causal inference is tied to experimental approach and that
correlative data cannot be used to draw causal conclusions
(Roff, 2002; Stearns, 1992). This problem concerns particu-
larly human life history studies, because the manipulation of
life history traits in humans is not feasible owing to ethical
reasons. Most researchers in our field recognize that such
inferential problems are due to the nonrandom selection of
study subjects or to missing confounding variables not
included in the analysis (Gagnon et al., 2009; Sear, 2007).
But few have seemed to recognize that measurement error in
independent variables, defined as the difference between a
value measured and the true value of scientific interest, has
also detrimental consequences on causal inference (Antona-
kis, Bendahan, Jacquart, & Lalive, 2010; Antonakis, Benda-
han, Jacquart, & Lalive, 2014; Pearl, 2009). Measurement
error, non-random selection, and omitted variables all under-
mine causal inference because they all introduce a correlation
between independent variables and the model errors (e.g., of
measurement error, please see the Appendix), thus violating
a key assumption for any regression modeling (Antonakis
et al., 2010, 2014; Pearl, 2009). Of these, the problem of
missing variables is, however, the most severe in human evo-
lutionary research as it is almost impossible to include all
Am J Hum Biol. 2018;30:e23075.
2017 Wiley Periodicals, Inc.
Received: 14 March 2017
Revised: 22 August 2017
Accepted: 18 October 2017
American Journal of Human Biology