Bryan Caplan  

Where Is the Political Flynn Effect?

Government Conspiracy to Silen... A Pareto-Optimal Move...
According to IQ tests, we're getting smarter.  But when I was reading Warren Harding's "Return to Normalcy" speech, it seemed way over the heads of a modern audience.  The anomaly inspired me to plug Harding's words into an online grade level applet.

The result: The average estimated grade level required to understand Harding's speech was 16.06 years.

By way of comparison, Obama's 2008 acceptance speech had an average estimated grade level of 9.64 years - and McCain's was 7.72!

If you adjust for the fact that average education levels are much higher in 2008 than they were 1920, these numbers are just weird.  Harding was talking at least a standard deviation over the head of his median voter.  Obama and McCain are talking at least a standard deviation below the head of theirs.

If I didn't know better, I'd assume that there's been a drastic expansion of the franchise since 1920.  It looks like we've gone from a world where only Harvard grads could vote, to one where all you need is passing grades in kindergarten.

Comments and Sharing

COMMENTS (36 to date)
Bob writes:

To be politically incorrect...

1920 is coincidentally the start of women's voting (Nineteenth Amendment passed that year)

Simfish InquilineKea writes:

I also find this extremely interesting. Even the uneducated presidents (like Lincoln) made speeches that seem way over the heads of a modern audience. I actually find the speeches of the founding fathers to be intellectual - they're pretty fun to read if you want to read something that will force your brain to focus.

Washington's Farewell Address ( ) is 18. Lincoln's Gettysburg Address seems low though.

mgroves writes:

Maybe that's the solution then to the irrational voting problem? Just reduce the number of voters, and especially reduce the number of irrational ones. Of course, a more educated person can vote just as irrationally, but I would submit that they are likely to vote less irrationally.

Zdeno writes:

But they look so adorable all lined up at the polling stations...

From the data I've seen, female IQs are either identical to men's or the difference is so small as to be practically insignificant. However, women do seem to be much worse voters than men, or so says this libertarian. Big-government democrats may disagree.

Brandon Berg writes:

I would bet that television has something to do with it. Speeches weren't widely broadcast in those days (even on radio, AFAIK). Presumably they were printed in newspapers, and the newspaper-reading population is more educated than the television-watching population.

Carl Shulman writes:

I think much of this can be attributed to TV. Attendees at speeches and newspaper readers who would actually read a written speech were more selected audiences. Also, modern speeches need to be amenable to carving out short clips.

Frejus writes:

You're barking up the wrong tree. Try putting a Henry James novel through that analysis software. Because it will probably consider him erudite but, in the end, after having read most of his novels, I'd say he's just plain stuffy. And boring.

That Harding speech is as well.

Robinson writes:

I think you're underestimating the effect of television and the internet. The people that would have either seen Harding's speech in person or would have bothered to read it in newspapers are not the same audience (relatively speaking) that watches acceptance speeches today.

I'm not positive on the effect of radio, of course.

Jason Malloy writes:

While I am always the first to argue the evidence goes against the Flynn Effect representing a rise in intelligence, Dr. Caplan's example probably just isn't representative.

In fact one of the evidences used by James Flynn to argue there has been a genuine rise in intelligence is the paper by James Rosenau & Michael Fagen showing that political speeches, writing, and reporting have become more complex and logical over the last 100 years in a number of different countries:

"... we tested the predicted generational changes by analyzing the skills of three types of individuals—elected officials in the U.S. Congress, witnesses at congressional hearings, and contributors to the daily press in three countries—in two widely separated epochs as they evaluated events across three issue areas—foreign affairs, international trade, and human rights. In doing so we randomly selected nearly one thousand paragraph-sized statements and coded them according to the methodology prescribed by the Integrative Complexity Coding Manual. All in all, our findings supported the hypothesis: the skill level of the sampled individuals was found to have increased over several generations... "

Gil writes:

I suspect that it's correct that the main difference is that older speeches were to be read, and modern ones are to be watched.

But, maybe part of the answer is that people used to pick a president who they thought of as an expert, someone above the voter's level, who seemed hyper-competent and qualified. And, today, people prefer somebody they can connect with emotionally.

Since voting for the inferior candidate doesn't matter (low probability of decisiveness), they choose somebody that they feel good about, rather than someone whose policies they've analyzed and judge more effective.

Mike writes:

On a related note, a site just went up ( that presents a voter fitness test that measures knowledge of U.S. government, politics, and the 2008 presidential candidates' positions on issues. On this site votes are weighted by test score so people with a reasonable understanding of government, politics, and candidate positions have a stronger voice in the election.

Nyle Kardatzke writes:

How could theater audiences in the 1600's understand Shakespeare's complex language and historical allusions when they didn't have the benefit of compulsory, state-financed schooling? Would they have made better voters than those we have today?

Billare writes:

Jason Malloy, please start a goddamned blog of your own. I'm tired of seeing your insightful, interesting knowledge/commentary languishing in the comments section of the various ones I read.

CK writes:

In 1920, the Progressive-Era reforms hadn't taken effect or had just barely take effect.

But doesn't Progressivism lead to rule of the experts? Only in governance, not out on the campaign trail or on the stump.

So the Presidential candidate of 1920 was making most of his pitch to party bosses. They chose the nominee, after all. And they knew what they had to do--which often didn't involve "message" at all--to get the people to the polls. Party bosses are not rationally ignorant. This is their job, people.

Then came radio and Fireside Chats, television and the Nixon-Kennedy debates, and it's off to the races. We appeal to the Masses who are, after all, asses.

In short, democratization (it's not the ladies; it's all us proles) and mass communication (I'm talking to you, Mr. Average Voter) and then more democratization (Hello, primary elections; goodbye, smoke-filled backrooms).

Policy is a function of the bureaucracy while politics is a mass game.

What was it that Hanson says? Vote on values (done); bet on policies (alas, not yet) . . .

Brian writes:

I imagine that there is also another element of the progressive legacy at work here as well: progressive taxation coupled with redistributive policies.

Once the lower, less educated classes began to be showered with money and goodies, they began to pay more attention to who was doing the showering, which in turn led speeches and addresses that spoke to their level of education.

Brandon Turner writes:

I think Gil's onto something. If you sat the median voter down, gave her that speech, and asked her what equipoise meant, she'd probably be able to figure it out just through pattern recognition. The problem isn't the vocabulary (and since when did vocabulary indicate the sophistication of thinking?)--it's the way we regard people with large vocabularies.

We just twice-elected a man who is all-too-eager to juxtapose himself with the sort of person who uses words like "equipoise." I think this is a matter of political culture, rather than demographics.

Brian writes:

Wouldn't political culture follow demographics? I'd be interested to see if the people who elected Bush have the same demographic composition as those who elected Harding.

bjk writes:

I blame advertising. We write and think like those composers of "prose poems to the potato chip," copywriters. Strunk and White should be subtitled "How to Write Punchy Copy."

Brandon Turner writes:

Brian, I think it does, yes, but only to a degree. The reason folksy anti-intellectualism still thrives in an increasingly educated populace might be because more and more people who are *actually* educated--people who would have been reading Harding's speeches--identify themselves as anti-intellectuals. Call it bourgeois false consciousness.

floccina writes:

I think that it only shows that the candidates are better at getting votes today. If Harding and crew knew that it was better to keep to it simple they would have.

Steve Roth writes:

>all you need is passing grades in kindergarten.

Well that does pretty much characterize the world view of our current president, and of the current Republican vice-presidential nominee.

"Everything I need to know, I learned in..."

dcpi writes:

Is our education really this bad people!

The Seventeenth Amendment was passed in 1912 and took effect in 1914. For those of you who skipped history to study economics, that amendment allowed for the direct election of Senators for the first time in U.S. history. Harding, it turns out, was the very first Senator from Ohio elected by the citizens of Ohio rather than the state legislature.

Perhaps, he was delivering the speech to the standards of the day which presumably would have been to write for the state politicians and not the unwashed masses.

That said, as other posters noted, radio and television and now the internet all have further redefined how political speeches are written, delivered and heard.

For extra credit: When were property requirements for voters dropped and what was the impact on the subsequent elections? When was the first secret ballot in a presidential election? Which presidential candidate ran the first "modern" campaign?

For extra extra credit for Bryan: Who was the first president who was elected based on "pledged" electors in the electoral college based on how the state's population voted?

Bob Hawkins writes:

I've listened to a lot of old-time radio from the 1930s and 1940s. It is not high-brow stuff. But the jokes sometimes rely on the audience knowing the plots of the more popular operas, and Shakespeare's plays, and knowing the leaders of the major labor unions. There are plenty of references to the pop culture of the time, as well.

Entertainment used to be in unimaginably short supply. People used to go to political speeches to be entertained, for God's sake. When radio came in, people listened to operas on the radio because there weren't 20 other channels.

In those days, a successful politician put as much into a speech as possible. Today, people don't listen to political speeches for entertainment. It's work. So successful politicians make their speeches as easy to take as possible. And you can still listen to opera on the radio, but you listen to the 3 channels of sports talk, the 12 channels of pop, or the 5 channels of rock instead.

Nathan Whitehead writes:

My view is that political science has progressed. It is undoubtedly effective to lower the knowledge and education required to understand your speech if the goal is to get more people understanding and agreeing with your ideas. This is progress.

It is not a bad thing. It's like subway maps. They used to be detailed maps of routes, but now they are "dumbed-down" versions designed for the mentally challenged. They benefit not only the mentally challenged but everyone who now have to put less effort into understand the maps.

dearieme writes:

Compare the jazz of Harding's day to the rap of today - a certain coarsening, wouldn't you say?

Prakhar Goel writes:

@bob and Zdeno

See Paper #60 at

CK writes:

Bob Hawkins wins the thread! (Along with, ahem, my own previous comment.)

hacs writes:

It's difficult to ratiocinate on goodness of cultural fit tests as I.Q. tests. Maybe it was better to think about effective years of study (the correspondent number of years of study that a person understands despite of his/her education) of the voters with relation to some fixed school curriculum (a benchmark for comparisons between years of study in 1937 and years of study in 2007, for example). Perhaps the expansion of the right to vote and its effects on the metrics mentioned above could explain the quoted phenomenon.

Steve Sailer writes:

Obama's big race speech from last March comes in a couple of grade levels harder than his convention speeches on the MS Word grade level meter. It was delivered on a weekday morning, so it was presumably meant to be read rather than watched on prime time, and was intended to affect elite opinion rather than mass opinion. Also, he wrote it in a few days, so he didn't have time to dumb it down.

Alan writes:

I was struck by the phrase in the fourth paragraph of Harding's speech where he writes of a citizenship which "seeks what it may do for the government rather then what the government may do for individuals"... Apparently Harding was the unreferenced source for one of the more popular sentences in the Kennedy Inauguration address which reads: "fellow Americans: ask not what your country can do for you -- ask what you can do for your country." The Kennedy speech, by the way, had an average estimated grade level of 10.96

Michael E Sullivan writes:

Our education is not as bad as this test makes it sound.

My guess is that Harding's speech would have sounded much less "intellectual" in its day, and that much of the high-grade level reading comes from using words that are unfamiliar to modern readers which were relatively common in 1900.

There may well be a literacy gap, and a seriousness gap, but this demonstration doesn't really do what you think it does.

One thing to consider is the *vast* amount of knowledge that is taken for granted today, which would be a complete mystery to those in 1900. The complete lack of these metaphors and statements in a speech from 1900 doesn't show up in your grade level test, but the things they've replaced in our general mindspace *do* show up there, and look like high-grade level items, because it's mostly scholars and nerds who read that kind of thing regularly today.

hacs writes:

Readability Scores

The text you entered has been checked, and scored as follows:
Original Text

"The I.Q. tests are goodness of cultural fit tests inasmuch as they are formatted to achieve ideological (in the broad sense) choices despite of the contents tested. For example, a math test can emphasize the speed to accomplish a long series of very simple calculus or it can require deepness in abstract deductive thought. Furthermore, a I.Q. test can only count right and wrong answers or to take into account the creativity of how the problems have been attacked."

Average Grade Level

Average Readability Level: 8.48
Average of grade levels scores that follow.
Approximation of number of years of education required* to read text.
Specific Scores

Flesch-Kincaid Reading Ease (Wikipedia): 65.6
Aim for 60 to 80. The higher the score, the more readable the text.

Flesch-Kincaid Grade Level (Wikipedia): 7
Approximation of number of years of education required* to read text.

Gunning-Fog Score (Wikipedia): 10.1
Approximation of number of years of education required* to read text.

Coleman-Liau Index (Wikipedia): 11.6
Approximation of number of years of education required* to read text.

SMOG Index (Wikipedia): 7.4
Approximation of number of years of education required* to read text.

Automated Readability Index (Wikipedia): 6.3
Approximation of number of years of education required* to read text.

* These scores return a "grade level", based on the USA education system. A grade level is equivalent to the number of years of education a person has had.

Darren writes:

I just ran through the first speech of the Australian prime minister Kevin Rudd - 12.9 and the opposition leader Malcom Turnbull - 10.56 - not sure if this says anything about Australian politicians. Interesting exercize as Kevin Rudd would be the more accessable and populist of the two.....

The Snob writes:

Part of this is I think the "classic rock was so much better" problem. Oldies stations can make a full playlist of nothing but the top 5-10% of a genre.

DrMrKen writes:

1. Voters might have preferred a candidate they couldn't understand back then as a sign of his erudition, a quality that counts as a negative now (as suggested obliquely above).
2. My latest submitted philosophy paper for publication in a peer reviewed philosophy journal has a score of 14.8, below that of Harding's speech. Most people wouldn't say that technical philosophy is very readable. So Harding's speech must have been particularly complex by the current computations. (Consider that the score is supposed to be the number of years of education one needs to understand the speech: 14.8 corresponds to someone at the end of her sophomore year of college, 16.06 is therefore someone who is already in grad school!)
3. Harding spoke in 1920, the readability scores were developed in the 50s 60s and 70s based on how long the words used are and how many words there are per sentence. It is easily possible that standard English usage is what has changed and that the formulas developed in the 50s-70s (with specific non-varying coefficients), were based on usage patterns that now emphasize shorter sentences and words than did common usage in the 20s. So a person with less education in the 20s might just have been using an English with longer sentences (which seems more likely than words). This is not too far fetched, given the more recent emphasis in style manuals and composition classes on decreasing the incidence of subordinate clauses. In a world with lots more subordinate clauses in use, the readability levels would be much higher by the formulas developed in the 50s-70s.
4. These are tests of READABILITY, not comprehensibility when spoken. This isn't just an issue of TV vs radio (as mentioned by the comments above), but a question of whether a speaker can get away with a speech that would be very complex if read on the page. That isn't too hard to fathom given the role that inflection and tone have in spoken communication. Again, consider the problem of the subordinate clause: reading it on the page can be extremely difficult and confusing if it is lengthy because you don't know how long it will be when you start and you can easily lose track of the sentence. This is solved in the spoken word by lowering one's voice or changing one's tone so that the listener knows to keep the other part of the sentence in mind. Now this would only be a reason not to be surprised by Harding's rating, not a reason for the difference between that speech and those today. But factor in again the loss of favor of the subordinate clause: speech writers today just don't use it as much even for spoken speeches since they were taught to minimize its use and are not as likely to consider that it would be more acceptable in a voiced lecture.

guthrie writes:

@Simfish as an aside from the main, and picking up on Nyle's reasoning...

To be fair, Lincon wasn't UNeducated, rather SELFeducated... even though he didn't have a government-sponsored representitive drilling him on vocabulary, he sure as hell knew what his words meant!

Perhaps that's what we're missing? Self-motivated learning?

Comments for this entry have been closed
Return to top