"The pediatrics group’s guidelines point out that research to date suggests a “correlation between television viewing and developmental problems, but they cannot show causality.”"
What is that statement based on?
If A and B are correlated, it might mean that A -> B, or that B -> A, or that there's a C that leads to both A and B. Without further data, there is no way to distinguish between these possibilities. And often the actual causation is different from what you'd guess based on intuition alone.
More and more papers these days rely on loose correlation to point to conclusions (not saying the linked one is because I didn't actually read the paper, just the link).
I'm sure it's a similar scenario all around, but at least in Brazil, universities are very, very production-driven, so any paper out is better than some good papers out, which leads to a rather poor quality overall.
I see it as a very real possibility that there is a "C" leading to both increased TV consumption and developmental challenges, namely, poor parenting.
There are statistical tools that can be used to control for confounding variables, unfortunately they are not always used correctly. Often a study in psychology relies on methodology to control for confounding variables, using only a chi-squared test (or similar) as their statistical measure. In many of these cases, the data is screaming for the use of a marginally more complex statistical method. A simple linear regression attempting to control for omitted variable bias would be a great improvement in many cases.
When you're making a claim about causation, there is certainly more required than simply statistical evidence of correlation. Even if some of these studies did more work on controlling for omitted variable bias, you would never know it from reading this article and for that reason should take it with a grain of salt.
In fact, where there is a clear and sane hypothesis in play (e.g. "time spent watching TV is time not spent learning to talk") it almost always works out that further science shows the causation that you expect. That's true across fields, and it really shouldn't surprise anyone.
It's just that the experiments required to show that are harder. Simple demographic studies are a lot easier and cheaper. So you do those first, then work up the hard stuff when you know where to look.
Taking your point literally, it would never be useful to do demographics like this because you can't "prove" the causation. But of course that's ridiculous; these studies are immensely useful and improve all our lives.
There can be more than one clear and sane hypothesis, even if it doesn't occur to you immediately. For example, Wilduck below suggests poor parenting as an underlying cause for both these effects.
Here's another example. It is well-known that there is a correlation between the global temperature and the levels of CO2 in the atmosphere (from looking at historical data). It is widely believed that higher CO2 levels lead to higher temperatures. On this basis governments have capped CO2 emissions, with real economic implications. This seems reasonable, doesn't it?
But some scientists who are working on this suggest an alternative explanation; here's the gist of it. There is a lot of CO2 trapped in ice, due to some historical reason that escapes me at the moment (it's not my field). As temperatures rise, ice melts and CO2 is released, leading to higher CO2 levels in the atmosphere. If this is the correct explanation, following your advice and putting caps on CO2 emissions certainly did not "improve all our lives".
"it almost always works out that further science shows the causation that you expect."
Evidence?
"Taking your point literally, it would never be useful to do demographics like this because you can't "prove" the causation. But of course that's ridiculous; these studies are immensely useful and improve all our lives."
Not at all. Sometimes correlation is enough. For example, if an insurance company finds a correlation between having red hair and being involved in more car accidents, they can use this information to their advantage.
Is it? A child that does not speak much takes more effort to engage with in play that does not get boring and tedious to the parent, and may get frustrated with the parent more easily from finding it hard to get their meaning across. In this case more TV would be a "solution" for both parties, and might be arrived at simply because it causes fewer tantrums and less stress for the parents.
I don't believe that's the most likely cause, but I also do believe one should be exceedingly careful about jumping to conclusions about cause based on "common sense" and correlation, because a whole lot of explanations that people would never even think of suddenly seem like common sense after the fact.
> In fact, where there is a clear and sane hypothesis in play (e.g. "time spent watching TV is time not spent learning to talk") it almost always works out that further science shows the causation that you expect.
Does it?
Of course, the best way to demonstrate causality is empirical test with a treatment and control group, but this is rarely possible in social sciences, and in studies like this would be downright unethical.
Thus, we're stuck trying to determine causality, and correlation is a strong tool to help us get there.