I think their method does count as a natural experiment and as such does have empirical evidence. From the article:
“Preschoolers from a 2010 study (612 children) were compared to preschoolers from a 2014 study (492 children) because of the digital explosion into everyday life that occurred between the two time periods. Television viewing patterns have remained similar over time, but technological device usage has increased.”
> If it were true that usage of touchscreens cause lower intelligence test scores, we would have seen this show up as a population-wide reduction in scores as touchscreens became widely available.
Exposure in formative years is vastly different than adult exposure. The affected cohort in the study is now only in middle school and population wide effects will have a lagged appearance.
Strictly speaking, the discussed research of toddlers using touchscreens is not an experiment also. In experimental setup researcher must control independant variable (touchscreen use). For this research it would mean to get a group of toddlers, split them randomly into two groups, give touchscreens to one group and some "placebo" to other. Researchers didn't do that, they split toddlers into two groups on the ground of the existing touchscreen use. So what they did is called a "correlational research", not an "experiment".
Correlation doesn't allow to speak about casuality. Maybe children ability to a visual search drive them to use touchscreens. Maybe touchscreens teach them to search. Maybe some other factor (like parenting style, or family income) teach them to love touchscreens and to effectively search for visual stimuli.
To be frank, I do not like this reseach much. It is correlational reseach, and there were no attempts to exclude interfering variables: if toddler used touchscreen more, than he/she would be better at tasks on a screen, wouldn't he/she?
* Spatial intelligence is positively correlated with socio-economic status.
* Screen (TV + Digital) time is negatively correlated with socio-economic status.
Concluding from this that spatial intelligence is negatively correlated with screen time is dubious. Similar reasoning is used to promote all kinds of class-correlated fashions, e.g. breast feeding.
I agree that this study doesn't measure anything useful, but in my circles, I have heard of and seen children younger than 2 years old learn to read by playing touchscreen games and children younger than 3 learn the names of states, capitals, and other trivia from playing touchscreen games. Achieving that is possible without touchscreens, but it would require an incredible amount of work from the caretaker.