It's summertime–when kids head outdoors for beach time, pool parties, and picnics. Time spent outdoors is not only fun; it's also associated with lower rates of near-sightedness (myopia). A journalist from NPR covered this topic in a recent story, titled, Want to protect your kids' eyes from myopia? Get them to play outside.
Myopia has been increasing around the world:
In the U.S., 42% of people are now myopic – up from 25% back in the 1970s. In some East Asian countries, as many as 90% of people are myopic by the time they're young adults.
Although there are treatments for myopia, researchers are also interested in prevention. One Australian scientist, Ian Morgan, decided to test light exposure. According to NPR:
A couple of decades ago, he noticed that the rates of myopia in East Asia were much higher than they were in Sydney. […] He knew from animal studies that light stimulates the eye to release the neurotransmitter dopamine, which can slow the eyeball from stretching. "Australians are famous for their outdoor-oriented lifestyle," he thought. "Maybe there's a link between getting outside a lot and preventing the development of myopia."
To test that theory, he and his colleagues designed a two-year study involving more than 4,000 6 and 12-year-olds in Sydney. Turns out, the researchers were right.
"The children who reported spending more time outdoors were less likely to be myopic and, we showed later on, less likely to become myopic," Morgan says of the finding, which was published in 2008.
a) In the Morgan team's study, what are the two key variables? Were they measured or manipulated?
In the empirical article in which the Morgan team's study was published, the conclusion reads,
"Higher levels of total time spent outdoors …were associated with less myopia … after adjusting for near work, parental myopia, and ethnicity"
b) The language above is a sign that the researchers used multiple regression to control for potential confounds. What would be the criterion variable in this analysis? What were the four predictor variables?
c) This regression result means we can rule out internal validity threats such as amount of near work, parent's levels of myopia, and ethnicity. What is another variable that you would like to see controlled for in this study? You'll need to select a third variable (C) and explain how it is associated with both myopia (A) and time spent outdoors (B). (C–>A and C–>B)
The NPR journalist reported on two, quasi-experimental designs as well:
Morgan's research caught the attention of Dr. Pei-Chang Wu, an ophthalmologist in Taiwan….
At the time, Wu's young son was starting first grade, and he worried about Taiwan's sky-high rates of myopia. Around 90 percent of teens there have it by the end of high school. Wu says the academic culture in Taiwan's primary schools didn't allow for much outdoor recess. "Many teachers want students to practice their homework during recess," he says.
But Wu convinced his son's elementary school to increase outdoor time. He also recruited a control school. A year later, his son's school had half as many new myopia cases as the other school. "We saw the results – they were very successful," Wu says.
d) What are the two variables in Dr. Wu's study?
e) What kind of quasi-experimental design is this? Pick one of the four from Chapter 13 (non-equivalent control group posttest-only, non-equivalent control group pretest-posttest, interrupted time series, non-equivalent control group interrupted time series).
Here is a second quasi-experiment:
He did more research, at more schools, and eventually convinced Taiwan's Ministry of Education to encourage all primary schools to send students out doors for at least 2 hours a day, every day. The program launched in September 2010. And after decades of trending upward, the rate of myopia among Taiwan's elementary school students began falling – from an all-time high of 50% in 2011 down to 45.1% by 2015.
f) What kind of quasi-experimental design is this? To help you answer, take a peek at the figures in the Wu team's empirical article, which is open access here. Pick one of the four from Chapter 13 (non-equivalent control group posttest-only, non-equivalent control group pretest-posttest, interrupted time series, non-equivalent control group interrupted time series).
g) What do you think of the size of the effect here?