[jasode]

The (controversial) economist Steven D. Levitt pointed out that there was no correlation between parents reading to children and reading test scores.[1]

I don't know which studies are accurate. I just remember that story because the findings were counter-intuitive.

[1] "Early Childhood Longitudinal Study, which tracks the progress of more than 20,000 American schoolchildren from kindergarten through the fifth grade." -- https://usatoday30.usatoday.com/news/opinion/editorials/2005...

[b_emery]

The editorial you site is from 2005, while the other paper is from 2014. The latter looks like a large and well designed study. Here is a quote from the conclusion:

"What are the implications of our findings? Our main finding is that it is important that young children are being read to. This is an early-life intervention that seems to be beneficial for the rest of their lives. We show that there is an important role for parents in the educational performance of their children. The evidence strongly suggests that parental reading to children gives them a head-start in life."

I haven't studied either enough to say for sure, but it appears the the 2014 paper is trying to measure the marginal benefit of reading, while the Levitt study was looking at larger correlations. The Levitt study suggests that smart parents end up with smart kids, reading or not. What is more interesting to me at least, is what can I do to help my kids succeed? The 2014 paper comes closer to answering this question, and suggests reading to them helps.

[jasode]

>The Levitt study suggests that smart parents end up with smart kids, reading or not.

To clarify, it wasn't a "Levitt study" per se. It was government study. The government had questionnaires asking parents "how often do you read to your children?" and it had the children's reading test scores. Levitt found no correlation in those 2 datapoints.

[dragonwriter]

The only thing I find to be less valuable than popular media news stories about scientific research results is popular media editorials interpreting research results.

Worse yet is such an editorial reporting conclusions from public data, without details of the analytical methodology.

[jasode]

>The only thing I find to be less valuable [...] is popular media editorials interpreting research results

In this case, the medium is not the message here. That page wasn't an "editorial" in the sense that a dilettante newspaper editor put his own spin on research he didn't know.

Instead, Levitt himself was writing an article that was short enough for the editorial space alloted. His longer text is chapter 5 in his Freakonomics book. (In other words, the so-called "editorial" was advertising hype for his book.)

I just linked the USA Today page because I thought it was more convenient than trying to deep-link a copyrighted book.[1][2]

It was a government study so presumably, Levitt got the raw data and reports from here: https://nces.ed.gov/pubsearch/getpubcats.asp?sid=024

[1] best link I could do but it's not the whole book so it's incomplete: https://books.google.com/books?id=pGJ-BwAAQBAJ&pg=PT25&dq=fr...

[2] Google Books doesn't let you deep-link to Chapter 5 of Freakonomics: https://books.google.com/books?id=wNPnl5zYA-cC&printsec=fron...

[tpeo]

I find it odd that it's Levitt and Dubner saying that, because their most famous papers are the ones that use instrumental variable regressions to squeeze causal relationships out of the data. Now they're saying that a correlation from a longitudinal study can be taken at face value?

Maybe I'm reading it wrong, but it is very strange.