Contra Hanania on Who Reads and Who Watches TV...

Or, why making up a story based on some statistics without knowledge of history will send you down a bad path.

I don’t have a particular issue with Richard Hanania. I get his news letter and read it. Only about half the time am I then motivated to write something in the vein of “You are wrong, here’s why.” That’s as opposed to other authors where I am often motivated to write an essay almost entirely expressed by holding shift and smashing the numbers row on my keyboard. Then I don’t do it because there isn’t even anything coherent enough to discuss or argue with, or even really stay mad at, because the author produced the typed equivalent of a parrot squawking madly in order to get a treat. Those eventually I just unsubscribe to, whereas Hanania is interesting enough to keep reading.

This post, however… it might just have enough bad reasoning and ignorance of the subject to poison everything else Hanania has written. Just a little poison, but enough that it makes being charitable towards assertions and vaguaries difficult. Let me explain.

Hanania’s post is long, and part of a longer series where he tries to sort out what the differences between the USA left and right, Democrats and Republicans, liberals and conservatives.¹ He points out that some popular descriptions, more vs less authoritarianism, Haidt’s moral foundations theory, psychological accounts that he doesn’t really describe enough to know to whom he refers, all those have important gaps that demand a new theory to properly explain.

Hanania proposes this theory to fill those gaps: “liberals live in a world dominated by the written word, while conservatism is something of a pre-literate culture.” Shortly after, in the caveats section, Hanania narrows that a bit to (emphasis mine):

This is not a theory about the average Democratic or Republican voter. TV watchers are the human norm, so if you’re talking about the bottom 80% of the population or so in political knowledge, the two sides are going to be pretty similar. Even among Democrats, many more watch CNN than read the NYT. This is a theory about the political class, that is, the community of journalists, activists, informed voters, and politicians on each side. And just as how most Democratic voters being TV watchers does not contradict the theory presented, neither does the fact that conservatives have an intellectual class. The arguments are mostly about the dominant forces among the politicized members of the public, not a way to think about everyone on one side or the other.

I already have a problem with the “bottom 80%” sentence there, but more on that later. I would sum up Hanania’s argument as:

The key difference between liberals and conservatives who are highly engaged, and thus tend to direct their respective side as elites, is that liberals cleave much more to written sources of news and information while conservatives cleave to the spoken word, e.g. TV and radio.

The first section of Hanania’s essay gives some statistical data and analysis on this subject and builds the basis for the rest, which covers a wide range of outcomes of his basic notion that the elites of each side differ on their interest in written vs spoken words. In fact, Hanania includes a really handy chart detailing the implications of his theory, which I want to applaud. It makes the whole long essay easier to navigate.

The first section is what I want to address, primarily because it is so flawed that it makes the rest of an essay an amusing exercise in how easy it is to make any and every observed fact fit a narrative, no matter how ridiculous that narrative may be.

---

Hanania’s theory that (elite) liberals like to read and (elite) conservatives like to watch TV or listen to the radio is supported by survey data presented in a number of graphs, such as the following:

The researchers from Pew asked each respondent about their use and feelings about a set of 30 news sources, as well as an open ended question about what their main source of political or election news was, according to the methodology section. So we have a survey looking at how people with Democrat or Republican tendencies rate various mainstream (by popularity) political news sources. The above graph shows the % that responded yes to a question of whether they used a given one of those 30 sources for information over the last week.

Estimating Use Patterns for TV vs Radio vs Text Sources by Individual Source Use

What does that graph suggest regarding tv vs radio vs text as a source of information? It is a little hard to tell… while 60% of Republicans tuned into Fox News, 53% watched CNN, and all the rest of the channels get 30-40% ratings of Democrats and 20-30% of Republicans, with the exception of MSNBC which Republicans apparently really don’t like. Ignoring the question of whether the respondents are elite, high political knowledge people or not, how do we know which side watches more tv? The best we can tell from that graph is that at minimum 54% of Democrats and 60% of Republicans got news from the TV, at at most 100% if the TV viewers don’t perfectly overlap. Strangely, the Pew report that Hanania references does not itself seem to mention any distinction between parties regarding the number of respondents that watch zero TV for news. To me this suggests that the numbers of TV viewers on each side are likely very close to each other as otherwise you would expect that to be mentioned. My guess would be that over 95% of respondents said they got some election news from a major TV source during an election year, regardless of party.²

So, we can’t tell much about TV as a source of information other than “Most people surveyed on both sides got news from the TV, at least 54%-60% and possibly all the people.” What about radio? That one is pretty clear cut, at least in the rough outline. NPR (the radio network) is listened to by 30% of Democrats and 11% of Republicans, while Hannity (the radio show) and Limbaugh (also a radio show) both served ~1% of Dem’s to 19% and 17% of ‘Pubs.³ Much like TV we can only know the minimum number of respondents who got news from the radio and guess at the maximum number, but at least for Democrats both numbers are 30%, plus or minus a bit, whereas for Republicans at least 19% but perhaps as many as 47% got information from the radio. Considering, however, that the two biggest sources for radio news for Republicans were individual shows, it seems more likely that those sources overlap and the number of Republican respondents who listed radio as a source of news is pretty close to the 30% of democrats. Even if it is 30% vs 47% of respondents, that’s not a huge difference and less than I would have guessed given that individual Republican shows serve nearly half the number of respondents as an entire network.

Then we get to text, both traditional print and websites. Here things fall apart, as without being told how many respondents of each party didn’t list any text source, we really have no idea how many are using text of one source or another. It does seem that at the low end Democrats are using text more than Republicans, as 30% of Democrats listed the NYT as a source, so at least 30% use a text source. Of course, if Republicans have little overlap in sources, that is, respondents read one and not others instead of reading a bunch, you can get to 30% as well. My bigger concern would be that the range of Republican text sources is much wider than Democrats, but who knows. Overall, however, it does seem that the text sources in the list of 30 options are more heavily used by Democrats than Republicans.

So, does Hanania’s claim that liberals like the written word while conservatives prefer the spoken hold up? Well, only if we assume that liberals and conservatives map 1:1 with Democrat and Republican leanings, but more importantly that the Pew survey says anything directly about the breakdown of TV vs Radio vs Text, which it doesn’t according to the report Hanania references.⁴ So far, all we can say for sure is that something like 54-100% of Democrats and 60%-100% of Republicans got some news from TV, ~30% of Democrats and 19-47% of Republicans got some news from radio, and 31% -100% of Democrats and 10-95% of Republicans got some news from text sources. I would hazard that, of those 30 sources, more Democrats said they read them than Republicans, but given the numbers we have there is nothing much to be said about relative TV and radio use.

We also can’t seem to say much about whether those who watched TV eschewed newspapers or radio; it is entirely possible, and I think likely, that many respondents got news from multiple source types, some from just one, and many from none or even just one or two sources regardless of type. Recall that the survey report does not give any information on use to source type, just specific sources.⁵

Estimating Use Patterns for TV vs Radio vs Text Sources by Individual Source Trust

Well, maybe if we can’t use the types of sources used to estimate differences in preferences for TV vs Radio vs Text, we can look at the trust numbers Pew reports to get an idea of who prefers what.

At first glance, it looks like we are onto something. Republicans trust Fox News and two radio shows the most, and really don’t trust anything else. Democrats trust just about everything that isn’t Fox News or those two radio shows. Clearly, Democrats love text more than Republicans, and Republicans love TV and radio more than text!

But wait… Democrats trust every TV source other than Fox, by very wide margins. NPR is also more trusted than not by Democrats, who distrust the two explicitly right wing shows (surprise!) Just like TV, Democrats trust just about every text source that isn’t explicitly right wing.

Republicans, on the other hand, don’t seem to trust any news source. Fox News gets the largest amount of trust from Republicans for any single source at 65%, but 19% say they don’t trust it. Hannity comes in second with 30% trusting and 10% distrusting, and it is downhill from there. Even Rush Limbaugh only has a 13% Trust/Distrust spread, and the other trusted sources never have a spread greater than 7%. Indeed, only 7 sources are more trusted than not by Republicans, compared to 22 for Democrats.

Yet both trusted lists, Democrat and Republican, include some TV, some radio and some text sources. Considering the paucity of sources Republicans trust more than not, getting all three categories in there is pretty surprising.

Whither Theory Hanania’s?

Little suggests that Hanania’s theory can be supported by the report he cites. At best we can say that Democrats read some few selected sources enough to remember to list them as sources of political and election news they used over the past week. If Republicans are reading a lot of articles and essays for news, it is more the equivalent of going on Real Clear Politics and seeing what George Will or Krautheimer has written, or subscribing to a lot of small Substack writers *ahem*…

… not going to specific online newspapers or other story aggregators.

Does the data presented in the report say anything else about what the “Mode of Communication” from the chart is for each party? No. In fact, where Hanania got the notion that the elite liberal prefers books, or the elite republican memes, remains a bit of a mystery. Hitting Ctrl+F and looking for ‘book’ only turns up 3 cases, none of which apply, and meme never pops up. Again, the only part of his theory his cited evidence supports is that Democrats read large text content sources more than Republicans. Really I should say “Large, frequently publishing text content sources”, as I don’t see National Review, Reason, City Journal, or other non-Time magazines on the list, even in their online incarnations, which strikes me as odd. I wonder what those numbers would look like, especially if those magazines had published an issue the week before.

---

So what’s a better interpretation?

The most notable thing about the Pew data in the report is the huge gap in trust of news sources between Democrats and Republicans. Democrats trust CNN the most, with 67% trusting vs 10% distrusting, while Republicans top out with Fox News, 65% trust vs 19% distrust. Democrats overwhelmingly report trusting the major news networks and textual source, along with NPR, with very little distrust. Republicans, however, barely trust any source, even those 7 of 30 they trust more than distrust. Let’s look at those trust charts side by side:

If you are trying to identify the fundamental difference between Democrats and Republicans based only on the Pew report, the difference should be “Democrats are very trusting of the news media overall, while Republicans don’t trust the news media much, regardless of media type.

Hey! Remember the 80’s?

But wait! We don’t have to use only the Pew data! Some of us remember the world before the new millenium. We remember why Rush Limbaugh became such a massive hit when his show was nationally syndicated in 1988: Republicans considered the mainstream media to be heavily biased towards the left. Talk radio became known as the domain of Republican news and commentary while the big networks and newspapers were abandoned. It wasn’t until 1996 when Fox News Channel was launched to serve the ever expanding number of Republicans dissatisfied with CNN and the major news networks. A “New York Times, but for Republicans” was never created.

This little tidbit of history casts a new light on the Pew data suggesting they read less: Republicans distrust all media, particularly legacy media. While radio and tv had their replacements rise in the 80’s and 90’s, the collapse of newspapers obviated the creation of a Republican leaning paper in the same fashion as Fox was created to compete with CNN. Possibly the slack was taken up by the Republican news magazines and various independent writers online. Or books, because we have zero information on that from Pew. Democrats and Republicans consume all types of media in roughly the same amounts, except that Democrats read larger legacy style text media a bit more. It is hard to make any firm conclusions, but this is a far cry from “liberals live in a world dominated by the written word, while conservatism is something of a pre-literate culture” for which there is zero evidence.

BUT WAIT! THERE’S MORE PEW PEW!

In fact, there is the raw data from the survey available from Pew! Let’s make a Pew account, download SPSS so I can open the file, convert the file to Stata so I can use it, and see what happens!⁶

There is one question about the type of media: What is the most common way you get political and election news? Using the raw data I can split that out between Democrats and Republicans. The result is… entirely unsurprising.

48.7% of Democrats list a text like thing as their main source of political news (Print, Social Media or News Website), compared to 39.84% of Republicans. If we pull out Social Media from that, we get 34.89% for Dem’s and 29.29% for ‘Pubs. Is anyone surprised that Democrats like Social Media more than Republicans? Is a 6% difference really a fundamental difference between Democrats and Republicans, especially given that the rate for Radio is nearly identical, and the TV gap is under 7% (41.77% vs 48.33%)?

The numbers do shift a bit if we consider only some version of Hanania’s elite. He does specify that the bottom 80% of political knowledge are going to be TV watchers and largely identical, so presumably his theory applies only to the top 20% of knowledge in each group.

We don’t have any survey questions about “Are you in the top 20%?” but we do have two avenue to get there. The first is a question regarding how closely do you follow political news, and the second is a series of questions regarding factual data.

If we consider “elites” to be those who think they follow political and election news closely, we can use the survey question ”How closely do you follow political and election news?” and only consider the 24% who answered “Very Closely”. How does that break down?

Well, that’s a bit better for Hanania. 41.12% of Democrats most commonly get their news from print or news websites, compared to 30.94% of Republicans. Radio is pretty much identical between the two, and TV goes 42.76% D to 50.98% R. I am not sure I would hang my hat on the 10% difference in text and ~7% in TV between the D & R as the fundamental difference.

What about knowledge? Well, the survey asks four questions about political facts: Did the deficit go up or down, did unemployment go up or down, what are tariffs and what determines the number of electors each state gets. If we define the elite as those respondents who answered all four questions correctly, we can look at their favorite sources and compare.

In this case, we get very little. 44.24% of Democrats vs 39.97% of Republicans prefer text, while more Democrats prefer radio than Republicans. There is again only a 7% gap between Democrats and Republicans on overall TV use, so the overall effect is shrinking.

Let’s go SUPER elite, and only consider those respondents who said they were very engaged AND got all four questions correct!

Text: 45.83% D vs 39% R. TV: 37.04% D vs 45.7% R. Does a 6.83% difference in text and a 8.66% difference in TV preference seem like a fundamental driver of all those differences Hanania mentions? I don’t think so. Particularly when the difference between the elites and the commoners doesn’t seem to matter.

---

Apparently I am reaching the email size limit, so I will wrap up here. The short version is that yes, Democrats seem to like text based sources of news a little more than Republicans, but only a little. So little that I think any theory that makes that difference the key difference in explaining the differences in behavior is really over reaching, especially considering how easily those differences are explained by historical trends in media trust.

What’s more, the differences I found are not what Hanania is presenting as evidence, but rather downstream data in charts that does not actually support his argument. Perhaps he accessed and analyzed the raw data and just neglected to describe it in his post, and instead opted to use the pre-made graphics that fail to support his claim without noticing that the didn’t. I don’t know… I am having a hard time being charitable beyond “This is just very sloppy and inappropriate analysis.” I hope that Hanania can do better going forward, as I do enjoy reading his blog.

And seriously, why are Democrats so much more trusting of nearly everything than Republicans?

1. These are used not exactly interchangeably, but pretty close. For instance, Hanania writes “…within conservatism, the more ideological forces like the pro-life movement and libertarians…” Libertarians are a subset of conservatives? Really?

2. Also confounding this, when asked about the main source, the open ended question, CNN and cnn.com are considered the same source, regardless of whether the cnn.com was a video clip of a show or a text based article. The Pew researchers really don’t seem interested in the divide between types of media.

3. I made the point about network vs individual shows because we are not looking at apples and oranges here. It isn’t entirely clear to me why Pew would lump all the programming content of NPR into one line item and not do the same with other talk radio stations.

4. I did get the raw data, however…

5. One question does ask to how you use sources: Which of the following best describes how you use the news source you turn to most often for political and election news?
57% reported they use multiple, 32% reported “I rely on this source far more than any other” so at least 57% of all respondents are using many sources, although they might all be the same type.

6. This is why I saved it towards the end…