Get the Kids to Bed on Time

As reported in the Wall Street Journal, a new study of sleep habits in children found that children with regular bedtimes performed better on cognitive tests than children with inconsistent bedtimes. Boys and girls with regular bedtime at age 3 performed better on cognitive tests given at age 7 than children with non-regular bedtimes. Cognitive tests included tests of reading, math, and spatial skills. The mother reported whether the child went to bed on time or not. The mother also reported the time the child went to bed on a weeknight (while school was in session). Over 11,000 children participated in the study.

The study found that children’s bedtimes became more regular as they got older. 19.5% of 3-year-old children didn't have a regular bedtime. This dropped to 9.1% of 5-year-old children, and 8.2% of 7-year-old children not having a regular bedtime. Children who were more socially disadvantaged tended to have less regular and later bedtimes. The study controlled for confounding factors that could influence test performance, including mother’s age, birth order, family income, parental education, parental employment, discipline strategies, hours spent watching TV, whether the child has a TV in the bedroom, and whether the mother drank or smoked during pregnancy.

With one exception, early or late bedtime didn't significantly affect children’s test scores, after other factors were taken into account. Late bedtime at age 7 reduced girls’ reading scores. The paper didn't present results on the association between test scores and specific early or late bedtime ranges at ages 3 or 5.

After controlling for the various factors affecting cognitive test performance, not having a regular bedtime at age 7 was associated with lower test scores for girls, but not for boys. For both girls and boys, non-regular bedtimes at age 3 were associated with lower test scores at age 7. The size of the association is small (the authors quantify the size of the association in the Discussion section, but as I mention in the next paragraph I question their methodology). Non-regular bedtimes at age 5 were associated with lower reading scores in girls, and math scores in boys. Girls, but not boys, who never had regular bedtimes at ages 3, 5, and 7 had significantly lower test scores at age 7. Somewhat surprisingly, boys who didn't have regular bedtimes at any 2 of the 3 ages had lower test scores, but boys who didn't have regular bedtimes at all 3 ages did not have significantly lower test scores.

It must be remembered that this is only one study, and replication is required before scientific acceptance of the link between stable bedtime and cognitive test scores. The study used the mother’s self report about whether her child went to bed at a regular time. This is both prone to recall errors, and also is inherently subjective. What exactly is a “regular time”? One mother may say that a regular time is +/- 15 minutes from a target bedtime. Another mother may say that a regular time is +/- 30 minutes from a target bedtime. I had problems with the table presentation of the results. Regression beta coefficients were presented along with confidence intervals, but statistically significant results were presented along with nonsignificant results, and there were no indicators separating significant from nonsignificant. There was no overall R^2 reported. I also had issues with the statistical inferences. A regression beta value for a binary variable like non-regular bedtime can’t be translated into the usual interpretation of beta: 1 SD of change in independent variable corresponds to (beta) SD change in dependent variable, but that’s apparently what the authors did. (How did the reviewers and editors miss this?) In the Discussion section of the paper, the authors report on the size of the associations in terms of standard deviations of test scores. I’m assuming that they are using the beta scores to determine this (the numbers they quote match the beta scores). If they are, then the association sizes in terms of standard deviations of test scores that the authors report in the Discussion Section are incorrect.

Assuming the link between stable early childhood bedtimes and cognitive test scores holds, it’s interesting to relate this to my hypothesis about ideal bedtime that I report in my human magnetoreception research paper. I've found that I have an ideal bedtime that varies according to where I am located, and whether or not daylight saving time is in effect. In the summer in Salt Lake City, UT, my ideal bedtime is about 11:35 p.m. I feel best both when my bedtime is stable, and when it is close to ideal. I’m assuming that sometime in my childhood my ideal bedtime was “set”. I grew up in Northern New Jersey. My ideal bedtime there is about 10:35 p.m. when daylight saving time is in effect. Did I consistently go to bed around 10:35 p.m. when I was a young child?

Another important thing to consider is the effect of daylight saving time. I have found that there is a 2 hour ideal bedtime difference when the time changes. (The explanation for this is rather complicated and I won’t go into this here). Ideal bedtime in Northern New Jersey is 8:35 p.m. during standard time, and 10:35 p.m. during daylight saving time. Perhaps I went to bed around 8:30 p.m. as a young child (more realistic than 10:30 p.m.). This 2 hour difference affects the meaning of “stable bedtime”. A child who goes to bed at 10 p.m. in the summer and 8 p.m. in the winter has a stable bedtime with reference to the natural environment. A child who goes to bed at 8 p.m. year round has an unstable bedtime—he goes to bed 2 hours earlier with reference to the natural environment when daylight saving time goes into effect. Studies such as this one don’t take this daylight saving time factor into account.

It’s possible that stable early childhood bedtimes can affect future mental health, as it can affect IQ. Perhaps unstable early childhood bedtimes can lead to sleep or mood disorders later in life. What about unstable early childhood environments (e.g. frequently moving)? What about the magnetic characteristics of the child’s bedroom (e.g. compass orientation of bed, and artificial magnetic fields in the sleeping environment)? These are the kinds of questions that researchers should be asking as they explore different and previously unknown environmental effects on child development.