Tag Archives: development

Questioning the Role of Menarche Timing in Life History Models of Sexual Development

Life history models of sexual development are popular for understanding female sexual development as it relates to early experiences in the rearing environment. Broadly speaking, life history models suppose that early stressful environments will lead to accelerated maturation, manifesting as earlier timing of menarche, which subsequently leads to earlier age at first sex, and more unrestricted (and risky) sexual behavior.

Although there are subtle differences between the three primary life history developmental models (see Figure 1 for a conceptual depiction of predictions across the three models), they all have in common the prediction that menarche timing is, in part, regulated by early environmental stress, such that stress and father absence – a particular variable of interest within the literature – predict earlier menarche timing.

Figure 1. Conceptual model of the theoretical predictions derived from various life history models of development. Solid lines indicate predicted association. PA = psychosocial acceleration theory; CD = child development theory; PI = paternal investment theory.

The effect of father absence on menarche timing has been enjoying a rich discussion in recent years, with criticisms ranging from genetic confounding (which my colleagues and I outline here) to WEIRD effects not found in cross-cultural data. Other criticisms have focused on the issue of broader confounding factors such as socioeconomic status and body mass index. Although each of these criticisms and alternative explanations for menarche timing have been investigated independently in previous research, no previous study has attempted to comprehensively evaluate a broad array of proposed effects of life history antecedents on menarche timing (that are derived from life history developmental models) with a genetically informative research design.

In our recent research, we (co-authors, George Richardson, Joe Nedelec, & Hexuan Lui) sought to test a comprehensive model of theoretically relevant life history predictors on menarche timing and subsequent sexual behavior in girls using a twin subsample from Add Health. We took a two-stage approach to test the associations between father absence, specifically, and environmental stress, broadly, with age at menarche and sexual debut, and to identify potential genetic confounding of those associations.

In the first stage we tested and refined a measurement model specifying a latent environmental stress factor, and then built and tested a comprehensive structural equation model to test our life history hypotheses (see Figure 2). What is most striking about these results is what effects are not detected. There are no significant effects of environmental stress or father absence on age at menarche; only BMI is a significant predictor (a well-known predictor in the pediatrics literature.) There were, however, significant effects of early environmental stress on sexual debut, consistent with recent work in a Canadian sample.

Figure 2. Final SEM model. Only significant effects show.

In the next stage, we used univariate and multivariate behavior genetic models to identify whether significant associations between BMI and age at menarche, and between age at sexual debut and number of sexual partners and risky sexual behavior, were confounded by shared genetic covariation. Results here did not provide evidence of a nonshared environmental association (i.e., an association that survived control of genetic and shared environmental factors), suggesting the effect of BMI on age at menarche in our nongenetically informative model is likely spurious. Findings also did not provide evidence of a nonshared environmental association between age at sexual debut and risky sexual behavior. However, the association between age at sexual debut and number of sexual partners did survive control of genetic and shared environmental factors, suggesting it may reflect a causal effect of the former on the latter.

This study tested many other predictions in addition to the key findings presented above. Below in Table 1 is the tl;dr version of our primary aims and results summary.

Table 1. tl;dr summary of research aims and results.

What do all these results mean for life history models of sexual development? First, and in my view, most importantly, is that the null association between early environmental stress and father absence with menarche timing (the latest null result in a growing list) calls into question each of the three life history developmental models. Why? Because parental investment theory, psychosocial acceleration theory, and child development theory all rely on menarche timing as a key regulatory mechanism linking early environments to adolescent sexual behavior. (In our paper we discuss at great length the implications for each theory on preprint pp. 20-23).

Our behavior genetic models offer some interesting insights to life history models, too – just not for age at menarche. The evidence associated with age at sexual debut does suggest sexual behavior may be, in part, responsive to early environmental stress. Unlike age at menarche, age at sexual debut does reflect the shared environment in all the models we tested. Again, however, life history developmental models will have to reconcile the fact that the key regulatory mechanism of menarche timing is missing. Additionally, these results also do not provide evidence for my own hypothesis that the association between father absence and age at menarche is genetically confounded – because there appears to be no such association to confound!

This paper, currently under peer-review, contributes to the ongoing critical discussions in the life history literature particularly regarding developmental models, facultative adjustment, and genetic confounding. We look forward to hearing your comments!

Preprint (19 June 2020)

OSF project and pre-registration

Does Spanking Cause Negative Developmental Outcomes? Yes, But Our New Research Suggests That Effects Are Probably Smaller Than What We Think.

The debate over whether it is moral to spank children is well-regarded as ‘over’ in that there is a strong consensus both publicly and scientifically that it is immoral to spank children. Spanking, defined here as hitting a child on their buttocks or extremities using an open-hand with the intention of modifying undesirable child behavior, is banned in a large number of developed countries with global movements dedicated to banning the behavior. Journalists covering the research on spanking and its effects on child development are equally strong in their conclusions with a number of publications stating that yes, “The spanking debate is over.”

What is behind this consensus view? Mountains of research across family studies, developmental psychology, and social psychology that have culminated on the following conclusions:

  1. Spanking is not associated with any positive outcome.
  2. Spanking does not correct problem behavior. In other words, spanking is not effective.
  3. Spanking is associated with only negative outcomes in children.

The most recent contribution to the spanking literature is an excellent and comprehensive meta-analysis published in 2016 by Elizabeth Gershoff from the University of Texas at Austin, and Andrew Grogan-Kaylor from the University of Michigan. Despite the consensus views regarding the negative outcomes for children as a result of spanking, their meta-analysis addresses some remaining concerns in the literature.

One primary concern addressed in the 2016 meta-analysis is ensuring that spanking (as defined above) is not confounded with harsher forms of physical punishment perpetrated against children (e.g., hitting children with objects; abusive behaviors including choking and beating, in other words, what would be considered physical abuse). This distinction is important: whereas there is a clear consensus that child physical abuse can be detrimental for child development, it is less clear – and still debated – whether controlled spanking is: (1) an effective form of positive punishment that reduces problem behavior, and (2) directly (and causally)associated with negative psychosocial outcomes throughout development and into adulthood.

The results of the Gershoff and Grogan-Kaylor (2016) meta-analysis suggest that spanking is, indeed, associated with wide-ranging negative psychological and behavioral outcomes across development. Psychosocial outcomes resulting from spanking vs. non-spanking, importantly, did not differ as a function of study design (e.g., retrospective, longitudinal), measurement (e.g., observational, child-report, parent-report), or age of children at the time of spanking (ranging from toddlerhood to adolescence). In summary, the results of Gershoff and Grogan-Kaylor align with the general consensus that spanking is consistently associated with poor developmental outcomes, appears to be largely ineffective at reducing problem behavior, and has not been shown to have any positive effects on child development. (Articles covering this paper can be found here and here.)

Although research does clearly show that spanking is associated with negative psychosocial outcomes, not all confounds have been adequately addressed. This limitation of the literature is where my coauthors Eric Connolly, Madi Sogge, Todd Shackelford, Brian Boutwell, and I decided to focus our empirical efforts. The confound we were interested in was genetic confounds. Genetic confounds in this context would reflect the idea that being spanked and the psychosocial outcomes of interests such as externalizing behavior and self-regulation, for example, covary at the genetic level; meaning that the genetic variation underlying children’s propensity to be spanked are the same, to some extent, as the genetic variation underlying their problem behaviors presumed to be an outcome of spanking.

Some research addressing these genetic confounds have been conducted with results generally supporting the notion that genetic covariation can account for a large proportion of the association between spanking and psychosocial outcomes. A study by Sara Jaffee et al. in 2004 found that genetic covariation between spanking and antisocial behavior accounted for 86% of the observed association, whereas the remainder of the effect was primarily explained by nonshared environmental covariation. Another twin study conducted by Button and colleagues in 2008 found that genetic covariation between parental punitive punishment and children’s externalizing behaviors explained between 61% and 98% of the association. (Importantly this is NOT the case for the association between physical abuse and psychosocial outcomes).

Genetically informed research is hard to come by in the spanking literature. Whereas the non-genetic literature includes a wide range of psychosocial outcomes – 17 in the Gershoff and Grogan-Kaylor (2016) meta-analysis – the available genetically-informative research on spanking includes only a limited range of outcomes, such as internalizing and externalizing problems, and antisocial or conduct behavioral problems. The purpose of our research was to use all the information available in the spanking literature to provide an up-to-date analysis of the role that genetic confounds may play in a wide range of psychosocial outcomes. Specifically, we provided probable ranges of estimates of the degree to which genetic and nonshared environmental covariation could explain the reported associations from the Gershoff and Gorgan-Kaylor meta-analysis.

We conducted two studies to address our aims. The first study reported results from the Children of the National Longitudinal Survey of Youth (CNLSY). Similar to the results of earlier studies, our analyses showed that for alcohol use, delinquency, and depression, 49%, 66%, and 80% of the association, respectively between spanking and the outcomes was accounted for by genetic covariation between the traits, with the remainder explained by non-shared environmental covariation.

Our second study took a simulation modeling approach to estimate the extent to which the association between spanking and 17 psychosocial outcomes was likely to be explained by genetic covariation and non-shared environmental covariation. For all outcomes, with the exception of physical abuse, the estimates showed that genetic covariation could potentially account for the majority of the association between spanking and the psychosocial outcome of interest. Non-shared environmental covariance accounted for a relatively small, but non-trivial amount of the association between spanking and psychosocial outcomes.

So, what do these results mean for our understanding of how spanking effects children psychologically? First, our results are consistent with the notion that it is plausible to suggest that spanking, in some cases, may exert a negative impact, and certainly not a positive impact, on psychosocial development across the life course. However, there is an important caveat to this claim: The effect sizes are likely to be much smaller than what the current literature suggests. Overall, the results of genetically informative research suggesting moderate-to-large degrees of genetic covariation between spanking and psychosocial outcomes imply that once genetic variation is considered, the environmentally-mediated (casual) effects of spanking on psychosocial outcomes will be much smaller in magnitude than what non-genetically-informed studies suggest.

That the true effects of spanking on child development are likely smaller is magnitude than what the reported literature suggests is not at all to say that the negative effects of spanking on child development are non-existent. The extant research and the data in our new study strongly suggest that non-shared environmental covariation explains a non-trivial proportion of the phenotypic effect – a finding that is, in fact, consistent with causal interpretations of the negative effects of spanking.

Although the spanking debate may be ‘over’ with regard to whether spanking negatively effects children’s development, our understanding of the effects from a genetically informative perspective is far from resolved. Our work demonstrates how little genetically informative research has actually been done on the topic, and the potential impact that genetically informative research can have on the estimates we rely on to make informed policy decisions and moral arguments. So, yes, all research points to spanking having negative effects on psychosocial outcomes, but the magnitude of the effect is likely to be much smaller thank what we currently think to be the case.

You can find the OSF project here, and the pre-print of our paper here, which is currently under review for publication.

Photo by Caleb Woods on Unsplash