WEIRD Considerations When Studying Adolescent Sleep Need

Within the past year, the field has developed three major sets of sleep recommendations for children and adolescents, with two of the three emerging in the summer of 2016.^1–3 While coming to slightly varied conclusions, all have been based upon a compilation of expert opinion derived from the available research literature. The process is more deliberate and systematic than prior guideline development, but is intrinsically limited by the available evidence. That evidence has been critiqued as being heavily focused on correlational research that prematurely assumes linear effects and is subject to reporter biases, and on experimental work that has focused more on proof-of-concept than determination of optimal sleep need.^4–6

In this issue of SLEEP, Ojio and colleagues⁷ continued to use a correlational approach, but allowed for the potential that sleep duration is non-linearly related to adolescents' symptoms of anxiety and depression. They looked for a least-risk zone: the duration of sleep at which those psychological symptoms least occur, potentially with rises in symptoms with either greater or less sleep. Although such non-linear approaches are not new, they have only recently been applied to pediatric sleep, and seem more directly useful in setting recommendations for optimal sleep duration than approaches that a priori assume a linear, “more is better” approach.⁵

There is much to like in this paper. The sample was large and likely to have been representative of the areas of Japan from which it was drawn. A validated mood rating form was used, and there was an attempt to control for irregularity of sleep patterns, both of which are commonly overlooked in epidemio-logical work. Further, the large sample allowed for parsed findings by sex and, to some degree, by age. All of these factors offset the limitations accurately noted in the manuscript (e.g., exclusive reliance on self-report, limited sleep measures) to make several contributions to the literature. The authors found that the least risk for anxiety/depression symptoms occurred at around 7.5–8.75 hours of sleep on school nights, with values on the higher end for younger and male students than older and female students. The average sleep duration in the sample was modestly lower than the least-risk point within each age/sex stratum, corresponding to slightly greater risk of anxiety or depression symptoms. Finally, at least in the younger adolescents, the risks associated with getting less than optimum sleep accelerated more steeply than the risks associated with getting more than the optimum sleep. These are important findings that inform the public health debate.

While celebrating that success, this editorial challenges the field to keep pushing the evidence base forward. Like all correlational work, one cannot infer causation from Ojio and colleagues' data. Moreover, those data raise questions about two factors that are important but often overlooked in psychological research (including sleep research). The first has to do with the larger context in which research is conducted: the vast majority of similar studies have been conducted by researchers from Western, Educated, Industrialized, Rich, and Democratic (WEIRD) societies, on WEIRD samples, and using constructs and measurement paradigms that are so deeply rooted in WEIRD epistemologies as to give an illusion of universality.^8,9 Sleep behaviors and their potential correlates are heavily culturally determined, however, so findings can differ depending on the culture in which a study is conducted. For example, some have argued that debates about bed-sharing have overemphasized findings from WEIRD societies.¹⁰ In such societies, bed-sharing may be culturally discouraged, such that its occurrence is disproportionately in reaction to a child's difficulties sleeping on their own, which then drives a statistical association between bed-sharing and child pathology.¹⁰ While the question of adolescent sleep need involves different issues, it is noteworthy that there are strikingly different cultural expectations pertaining to adolescent sleep even across two WEIRD societies.¹¹ As such, it will be important to test whether Ojio's findings generalize cross-culturally, rather than un-questioningly assume universality.

The second often overlooked factor is related to the WEIRD concept, but focuses more on behavioral diversity within a culture. Cultural expectations define what is normal and abnormal (i.e., “weird,” in lower-case letters) for individuals. Violating cultural norms—independent of whether those norms are healthy—imparts its own risks, either causally (e.g., via social feedback) or because individuals who violate cultural norms self-select to do so or engage in other behaviors that increase risk. Here is an example that is outside of sleep, but makes the point. In the mid-20th century, cigarette smoking was widespread in America,¹² smoking was common in the popular media, and statistical associations between smoking and psychopathology were fairly weak.¹³ Smoking was not safe, but it was culturally normal. However, as the social acceptability of smoking plummeted—as it became weirder— its correlation with psychopathology grew.¹³ The biology of smoking changed little, but being weird imparted statistical risk. Applied to sleep research, it is noteworthy that the published studies that have found curvilinear relationships between sleep duration and psychosocial outcomes in adolescents have tended to show least-risk points that approach the average sleep in the studied population.^7,14,15 The gap between actual and optimal sleep may be greater than these studies suggest; if indeed being individually weird within a culture imparts its own risk, then the statistical inflection point in these studies could get distorted toward the population mean. Indeed, insofar as many WEIRD societies prioritize individualism and many non-WEIRD societies value collectivism,⁹ the risk associated with being individually weird might be even greater in a non-WEIRD society.

To be clear, these issues are not blistering critiques of Ojio and colleagues' contribution.⁷ Scientific knowledge grows with the overall corpus of evidence, not a single study. Their study adds nicely to growing correlational evidence—consistent with contemporary sleep recommendations—that there is likely to be an “optimal zone” for adolescent sleep. It also provides important evidence of where that zone may occur, albeit with a clear need for additional work to help differentiate what is unique to the culture, attributable to hidden confounds, and unique to the research design itself.

Researchers do not need to throw up their hands in exasperation when faced with individual and cultural variation. The field can respect its limits. If the evidence base comes from WEIRD societies, it is absolutely reasonable to develop evidence-based guidelines for the millions of adolescents living within them. However, it is unreasonable to uncritically apply those to all of humanity. True cross-cultural work, with a healthy dialogue between empirical and ethnographic methodology, is needed.¹⁶ On this count, it is encouraging to note that Ojio's study took place in Japan, so at least the W in the WEIRD acronym does not apply.

It is also reasonable to use correlational designs that test both linear and non-linear effects. In doing so, I strongly recommend considering the potential impact of individual weirdness by examining a wide a range of potential confounding variables. Future research should also concurrently examine multiple outcome measures and domains (e.g., mood, attention, academic performance, dietary changes).¹⁷ Not only might optimal sleep differ for each of these, it would be difficult to explain the presence of such differences by invoking the “being weird is risky” argument.

Finally, being culturally attuned does not rule out doing good experimental work. To be maximally useful, such experimental work should (a) measure multiple outcome domains, as it seems plausible that the domains will differ in sensitivity to sleep changes, (b) test a wider variety of sleep “doses” to better-inform population-based guidelines,^5,17 (c) design sleep conditions that vary not only on duration, but also sleep phase, to inform both how much and when sleep is optimal,¹⁷ (d) consider whether the methods mean the same thing to the participants as they do to the experimenter,^9,16 and (e) maximize the diversity and population representativeness of the sample^6,7; to date, most experimental sleep research has been with highly atypical samples, most often using White, Elite, Intelligent, Rich, and unusually Determined (WEIRD) participants. Such samples might be convenient, but the findings might turn out to be…unusual.

CITATION

Beebe DW. WEIRD considerations when studying adolescent sleep need. SLEEP 2016;39(8):1491–1492.

DISCLOSURE STATEMENT

Dr. Beebe has indicated no financial conflicts of interest.