Thread by @ClaireBerlinski

Okay, some thoughts. The effect, though "statistically significant," is also small: considerably smaller than I'd expect given my prejudices, which are firmly in the "television rots your mind" camp.

https://twitter.com/PieroCastellano/status/1147498979296448512

Correlation isn't causation. As they note, grudgingly. "Finally, early Mediaset viewers may have simply been more likely to know who Berlusconi was when he first ran for office,"

"or they may have been more sympathetic toward him out of gratitude for the unprecedented entertainment opportunities offered by his TV channels."

"exposure to entertainment TV during childhood
increases the probability of receiving very low scores on psychometric tests by 8 to
25 percent over the baseline," they right.

So, what kind of parent puts their kid in front of crap TV all day?

Can we actually disaggregate the effect of being exposed for hours to crap TV in childhood from the effect of having parents who stick you in front of that crap TV for hours? Not easily.

Propensity to perform well on psychometric tests is either a matter of nature, nurture, or luck. Whatever the cause, "parents who let you watch crap TV for hours" instead of reading Plutarch seem just as likely to be the causal factor here as the TV.

This kind of research is beastly difficult to do. I applaud them for trying, but in writing about the results, we should be clear about what we have and haven't learned. They've learned there's a small correlation between being exposed to tons of crap TV in early or old age--

--and voting for populist parties. In Italy. Which doesn't mean we should expect these results elsewhere; Italy has its own very unique culture. But is the crap TV to blame, or circumstances such that parents thought it a good idea to use it their TV as a babysitter?

That may suggest low IQ on the part of the parents, perhaps inherited. It may suggest other social stresses on families that affect academic performance; it may suggest that watching crap on TV leaves you with less time to spend on things that train you think more rigorously.

"One reason populist leaders may be particularly appealing to less sophisticated voters is because they use a language that is more direct and easier for ordinary citizens to understand." Well, sure. That's also why these channels may have been popular.

Mediaset may have been successful in these regions precisely because these are regions full of voters who were already less sophisticated. Something else might be causing that--the arrow of causation hasn't been established.

You won't get far trying to make me watch Mediaset for hours. I find it dumb and unsophisticated, which is best explained in terms of my class background.

It's hard to run an experiment in which you take people from *exactly the same* socioeconomic background,

separate their kids, put 1/3 before the Tube all day,send 1/3 of the rest off to practice their Latin declensions; and as a control send the rest to dig and fill holes all day but call it "education," right?

"We find little evidence that early Mediaset viewers were more likely to evaluate Berlusconi more positively: of the six coefficients in columns 3 to 8, only one is significantly different from zero.

"The effect on the overall rating of Berlusconi, on a scale between 1 and 10, is also not significantly different from zero (column 9)." This, they say, "seem[s] to exclude that early Mediaset viewers had either better knowledge of Berlusconi or a better opinion about him."

"Indeed, “name recognition” hardly played any role in the elections after 1994, when Berlusconi became Italy’s best-known politician." Really? How do they reckon this? As they say, "More than 90 percent of respondents already knew of Berlusconi when he first entered politics."

They may not have found evidence that they viewed him positively, but they sure found evidence they knew his name. That might explain everything. We know "name recognition" is a key variable in voter choice.

They say the effect is "quite sizable." I don't think 1.5 is "quite sizable." It's trivial.

"Our findings," they conclude, "offer the first systematic evidence that exposure to entertainment
television influences voting behavior." No, it doesn't. It shows a correlation between

television watching and voting behavior; it does not show that one causes the other. It's just as plausible that stupid people like stupid television; that Berlusconi was a good businessman and thus figured out where to find the stupid people

for his stupid product, and that subsequently these people voted for stupid parties--as you'd expect--because they're stupid.

They add that it " suggests that this effect is mediated by deeper cognitive and cultural transformations." IOW, they think stupid TV rots your brain.

I do too, but I don't think this gives my case some kind of scientific support. We just don't know enough about how intelligence is handed down or nurtured. Some identical twin studies here might have been interesting.

"Mediaset coverage in 1985 may have been correlated with other local characteristics that could have affected electoral outcomes in ways other than through TV," they say, anticipating this criticism; so they try to quasi-randomize it

by identifying "from the residual variation in signal strength due to idiosyncratic geographic factors within narrow areas, which is uncorrelated with both past electoral outcomes and a wide range of municipal characteristics." IOW, they look for pockets exposed "by chance."

But do we know they were exposed "by chance?" Do we have evidence that they weren't chosen by market research suggesting that they'd be especially likely to enjoy crap TV? Mediaset was a pretty shrewd business, maybe it wasn't an accident that they knew their customers well.

In any event: the main finding is that there's a robust correlation but a small one, and not one apt to have shaped the course of history. "A 1 standard deviation increase in Signal is associated with a
2.85 percentage point increase in the vote share of Forza Italia." TBC

Me? I don't get out of bed for less than two standard deviations. I'm not going to war with Crap TV over a 2.85 percentage point increase in vote share. Let's see how this compares with other factors correlated with dumbness and dumb voting before we cause a moral panic, I say.

We've got to figure out what's causing reverse Flynn Effect throughout the developed world; this might be part of the puzzle, but I suspect like all things it's complex, and probably not the priority target for fixing it.

they *write*