An excerpt from a study (and conversation) by and with writer and filmmaker Errol Morris, published in the New York Times:
Until about 150 years ago most people wrote out documents by hand. Since the advent of typewriters (from John J. Pratt’s pterotype in the 1860s to word processors in the 1980s), few people write by hand anymore, and we now have a vast array of typefaces available to us. It is an easy matter to change an entire document from Bembo to Garamond to Caslon to Palatino. We forget that written manuscripts, letters and journals were once unique objects often containing clues about the writer and the context of when and how they were written. [...]
DAVID DUNNING: Baskerville seems to be the king of fonts. What I did is I pushed and pulled at the data and threw nasty criteria at it. But it is clear in the data that Baskerville is different from the other fonts in terms of the response it is soliciting. Now, it may seem small but it is impressive.
ERROL MORRIS: I am completely surprised by this. If you asked me in advance, I would have guessed Georgia or Computer Modern, something that has the imprimatur of, I don’t know, truth — truthiness.
DAVID DUNNING: The word that comes to my mind is gravitas. There are some fonts that are informal — Comic Sans, obviously — and other fonts that are a little bit more tuxedo. It seems to me that Georgia is slightly tuxedo. Computer Modern is a little bit more tuxedo and Baskerville has just a tad more starchiness. I would have expected that if you are going to have a winner in Baskerville, you are also going to have a winner in Computer Modern. But we did not. And there can be a number of explanations for that. Maybe there is a slight difference in how they are rendered in PCs or laptops that causes the starch in Computer Modern to be a little softer than the starch in Baskerville.
ERROL MORRIS: Starchiness?
DAVID DUNNING: Fonts have different personalities. It seems to me that one thing you can say about Baskerville is that it feels more formal or looks more formal. So that may give it a push in terms of its level of authority. This is, of course, speculation. I don’t really know. What one would do with, when you get surprising results is you now have to think about, O.K., what do we do to take that back-ended speculation and support it with data?
ERROL MORRIS: How surprised are you by this?
DAVID DUNNING: I’m surprised that the damn thing worked at all — because you are conducting an experiment in an uncontrolled environment. Who knows what’s going on at the other end of a computer screen? Their kids could be screaming in the background for all we know. It could be two a.m. It could be two p.m. They’ve had their coffee. They haven’t had their coffee… The font is on their desktops. There is just a ton of stuff out there that could obscure any results whatsoever. That’s why I made sure to have those six levels of confidence —
ERROL MORRIS: Because —
DAVID DUNNING: Because, basically, there are two different types of questions you can ask in a survey. You can ask yes/no. Do you agree with X? And that is a rather crude question, because if a person says yes, you don’t know if they are saying, “Yes, God damn it,” or if they are saying, “Ye-es.” [in a meek voice]. They both qualify as yes. However, if you ask about gradations of the “yes” (or gradations of agreeing), then if there is a more subtle phenomenon going on you have a better chance of catching it. You catch people going from “Ye-es” to “Yeah.”
ERROL MORRIS: And what did you learn from this data?
DAVID DUNNING: That people either agree or they disagree. They are not hovering around the middle at all. They choose a decisive yes or no. But I thought that some fonts would be rejected rather than that one font was going to be the winner.
ERROL MORRIS: For example, Comic Sans would be a loser.
DAVID DUNNING: Exactly.
ERROL MORRIS: The loser font.
DAVID DUNNING: The inappropriate font. What is this font doing here? [laughter] But no. That doesn’t seem to have been the case. And that’s why you do the studies. Sometimes you get exactly what you expect. O.K., great. You publish them. The fun happens when you do a study that comes out in a way that no one would ever have expected. Now you’ve got to sit back and say, how do I explain that? Can I explain that?