More People Need(ed) to Read Harry Potter

[annoni-no.livejournal.com]

I know, a provocative title in this community, but we have concrete evidence that reading Harry Potter leads to a small, but significant, increase in antipathy toward Donald Trump and his policies.

https://www.asc.upenn.edu/news-events/news/new-study-shows-reading-harry-potter-lowers-americans%E2%80%99-opinions-donald-trump ; (Link through to the actual study in article.)

A while back I posted about a study that found that identifying with Harry Potter led to decreased bias toward stigmatized minorities.  At the time, I wondered how reading the series led people to feel about how to deal with their enemies given the vindictiveness the series shows in a close reading.  As it turns out, the more Harry Potter books someone has read, even controlling for "party identification, gender, education level, age, evangelical self-identification, and social dominance orientation," the more opposed they were to violence and punitive policies (like torturing their enemies as advocated by Trump) and authoritarianism.  This is in addition to confirmation of the decreased bias against outgroups.

You don't have to like Harry Potter, and I completely agree that the books have a lot of problems.  But let's not loose sight of the fact that the world is entering a dangerous, if not outright fascistic period.  There's too much hatred and divisiveness driving our politics; hate crimes have risen by several hundred percent since Trump's election.  If reading Harry Potter does help lead people to greater tolerance and mercy, we shouldn't let the perfect be the enemy of the good.

Comments

[zigadenus.livejournal.com]

[I tried posting this earlier, but it tagged it as spam on account of hyperlinks. Links are still here, because looking at the source is always good in this kind of exercise, but you'll have to remove '$' I added.]

Mmmkay. Study pdf is here, in case anyone else wants it (www$cambridge$org/core/services/aop-cambridge-core/content/view/84B3BED39ACA703DC7B8BE2D5486B185/S1049096516001633a.pdf/harry-potter-and-the-deathly-donald.pdf)

I read over their methods, and looked at how they constructed their analytical model. --Again, no ideological stance on this (I haven't even read the preamble and arguments, I'm just looking at research methodology, approaching this from a science/good data/good study design perspective.). Pertinent bits are:

total of 1,142 respondents completed both waves of the survey relevant to this study. To measure the two independent variables, a survey administered in 2014 asked all respondents about their extent of exposure to the Harry Potter story through either books or movies (see online appendix). Each person’s scores were summed.

Okay, 1) good sample size. Check. 2) online appendix comprises the survey questions, so we can see what the author actually asked. Awesome. So let's take a look:

(static$cambridge$org/resource/id/urn:cambridge.org:id:binary:20161118060719324-0118:S1049096516001633:S1049096516001633sup001.pdf)

3) Now we get into the problematic areas. My expressed suspicions are justified: the author asks only about Harry Potter -- Methodologically there are two arguments to be made here: A,the author is only interested in the specific effects of Harry Potter vs B, total exposure to reading (anything, even Harry Potter) may be driving the trend that the author reports. To untangle what's really going on here, this study needs more data. If the trend is real (i.e., if reading Harry Potter does make you more empathetic or whatever), the study needs to explicitly include people who read a lot, but didn't read much Harry Potter. Otherwise, this is simply a correlation with some other explanatory variable (maybe "total reading", maybe something else entirely -- we don't know, and we can't know, given how this data was collected).

But ok, let's go on to the control variables the author accounted for:

I include control variables in all models in order to take into account potentially spurious causes of both Trump support and exposure to Harry Potter. All models included gender (females were expected to rate Trump poorly), education (expected to negatively predict Trump support), age (expected to positively predict Trump support), and evangelical self-identification (expected to discourage both tolerance of Muslims and gays, and consuming stories about wizards). Two dummy variables accounted for party identification, and ideology was measured on the usual seven-point scale.

4) Ok, methodologically, these are all good controls. Pass. But it still doesn't address whether it's exposure to reading anything at all, or exposure to reading Harry Potter specifically. The 'education' variable gets closest, but is only a proxy -- there is no way of determining (among the data collected by this study)how much leisure reading any of the participants did, regardless of whether they are highly educated or not. And no, you can't make the argument that 'more educated people read more in general'. Maybe they do. But you don't know if that is true of the people in this particular study, because it wasn't addressed in how the data was collected.

So in summary: this study was structured badly, and it is technically impossible, given the type of data collected, to determine whether it is reading Harry Potter, specifically, that drives any of these trends. Maybe reading Pride and Prejudice would do the same thing. Point is, we can't tell from these data.

[sweettalkeress.livejournal.com]

That was my first thought as well. Especially since, in my own personal experience, people who find "deep messages" in what they read tend to have read those things for affirmation, rather than for instruction. My first thought is that what's going on here is that there is some variable that influences both love of Harry Potter and antipathy toward Donald Trump and/or racism more generally, that couldn't be measured in the study. I would be very, very surprised to learn that one book, or even one series, was enough to influence people's opinions on racism in any really major way at all.

[zigadenus.livejournal.com]

there is some variable that influences both love of Harry Potter and antipathy toward Donald Trump and/or racism more generally, that couldn't be measured in the study.

Could very well be! It's important to remember, too, that the study didn't ask if participants were fans of Harry Potter -- only whether they'd read the books (and how many), or seen the movies. So we're on slippery ground in even trying to deduce anything about whether the participants liked the books or just read them. There's too little data available to support the majority of the conclusions.

[madderbrad.livejournal.com]

... they read tend to have read those things for affirmation, rather than for instruction.

Or the Harry Potter text provides 'confirmation bias' for any negative feelings about Trump. I've been seeing confirmation bias mentioned a lot lately. :-)

[annoni-no.livejournal.com]

The difference between Harry Potter and all other books is that Harry Potter was a cultural phenomenon unlike any seen in recent memory. While I do wish they had specifically asked about other reading, they didn't give us the nothing you suggest. There was *no statistical difference* in the number of people who read Harry Potter when sorted by political affiliation, even though Republicans are less likely to read books than Democrats.

A lot of people who aren't normally readers read Harry Potter. We've known that for years. Even with a representative sample, it's possible that they somehow reached only Republicans who read more than the average or Democrats who read less, but it's unlikely.

The worst you can say about this study is that it only confirms what other studies have told us: reading, especially fiction, promotes tolerance. I thought it particularly relevant because a number of members of this community had expressed concern about the vindictiveness and cruelty in the Potter books, as well as other social justice concerns Potter handled badly, and this study showed that the correlation ran the opposite way from what people were concerned about.

[zigadenus.livejournal.com]

Eh, I think my point was lost somewhere. I didn't, in any way, negate the position that literacy promotes empathy. Personally, I'm of the opinion that Stephen Pinker is spot-on, there. I only performed a methodological analysis of the construction of their study, and pointed out there was a simple way that they could have explicitly tested the hypothesis that it's Harry that matters, and they didn't do that. My entire post was more in the nature of explaining that we can only reason so far with the data as collected -- and that neither the author's arguments, nor inferences about HPfen characteristics can be adequately treated with the data as it was collected. And the data didn't address Republican vs Democrat at all -- the study only asked the participants' attitude toward Trump in particular.

[zigadenus.livejournal.com]

And the data didn't address Republican vs Democrat at all

I should clarify this in that specific ideologies defining Republican vs Democrat were not addressed. There could be many reasons why a person chooses one or another. Everything from critical study to family tradition of voting for a particular party may be at play. Thus, voting affiliation may tell us very little about personal political beliefs. A better way to determine ideology (if you want to make sweeping statements about Republican vs Democrat), is to include in the survey a set of questions where the participant is asked to rank how much they agree or disagree with lists of ideological statements -- just like VoteCompass. From that, you cluster the participants as ideological Republicans, Democrats, etc. It has the value of being replicable.

To explain what I mean by that, let's examine how it is potentially erroneous and leading to draw parallels between this study and statements like "Republicans are less likely to read books". While there may indeed be studies that demonstrate that, we do not know if those studies are defining 'Republican' the same way as it is defined here. 'Voter affiliation' is useless without context -- someone who affiliated themselves with the Republicans under GW Bush might hesitate to do so under Trump; someone with Democratic affiliation under Clinton may not have voted for Gore. So depending upon when a study was conducted, it may be comparing different parts of the population entirely.

[kahran042]

If you want to white-knight the series, do it on your own blog.

[zigadenus.livejournal.com]

On more sober reflection of this topic, please do just ignore my other responses here. I think we're at risk of discussing this from two very different perspectives -- for my part, I'm talking research methodology alone, and couldn't honestly care less whether the topic is "Harry makes people nicer" or "Kale consumption predicts empathy". It's just not that interesting to me, and I don't think people reading or not-reading Harry Potter is going to be a solution to any of the social and economic problems that have been on the rise in your country.

[jana-ch.livejournal.com]

I want to thank you for taking a look at the methodology. Since I was the first to question the solidity of the study, I felt I ought to check it out rather than just snarking it sight-unseen, but I was too lazy. Fifty “Geography 426” points to your House for taking on the job.

I suspect the study is picking up on reading in general, not just reading the Potter books.