People reading the exact same text thought that the recipe would take longer to complete when it was written in one TYPOGRAPHIC TREATMENT versus another. (My emphasis.)
Kevin, I think you have expressed that correctly.
The problem is that the font used, not the treatment, is getting the credit for making the difference, and before you know it, Arial is being bandied about as the go-to, go-faster typeface.
Perhaps it would be better to not mention the names of the typefaces used in these tests, as it is such a distraction.
After all, if the fonts in this study had been described as Font a, "a legible, grotesk sans serif" and Font b, "a connected handwriting script", that would have been quite adequate for scientific purposes, would it not? (These descriptions are the top MyFonts tags for the two faces).
By mentioning the names of the fonts used, researchers run the risk of becoming unwitting pimps for commercial products, and their serious theoretical work being mistaken for mere product testing.
Without mentioning the names it would not be possible for another scientist to reproduce (support or refute) the results. The commercial point is very interesting however. It would be possible to say "Font A" and "Font B" thereby not plugging a product, which I think good. You may be on to something there Nick. I could design a typeface, then conduct hundreds of sloppy experiments solely in an attempt to broadcast it's name as a marketing ploy. In order to reproduce a study where the fonts aren't named, you would have to contact the experimenters directly which would place a moral obligation on the scientist to share the information so that the community could validate their work.
It has been my experience that not all research scientsts, especially ones in fields with commercial application, are willing to give away their materials, data &c in fear of getting scooped. I'm still new at this however, and have only come across this problem a handful of times. For the most part, research scientists I have contacted are more than wiling to share their information as it only increases the likelihood of them getting referenced.
Regarding Nick's commercial point, I think there may be a parallel in medical research and drug names. Is there a doctor in the house?
Why do they have to use the same fonts?
They don't use the same subjects.
Surely one "legible grotesk sans serif" is as good as another, if it's being compared with a "connected handwriting script".
Anyway, I'm doubtful as to whether researchers are really interested in reproducing one another's experiments.
Where are the follow-ups to this Song and Schwartz affair?
I'm aware of one study, Yager et al., 1998, that compared a "Dutch" and "Swiss" typeface. Times vs. Arial/Helvetica, I would imagine.
"I'm doubtful as to whether researchers are really interested in reproducing one another's experiments."
Reproducing results is a major aspect of all scientific research. How else would we validate results? Pointing to confounds is simple, but if the reported results are significant, and there are confounds that go unnoticed or unreported, then we are accepting false data as true. Imagine the implications of this type of approach in the case of medicine.
I have not read Yager, but for Times versus Helvetica see:
Rudi, W. et al (1993). Performance differences between Times and Helvetica in a reading task. Electronic Publishing 6(3), 41–248.
Not much can be drawn from this study however as they failed to counterbalance the order in which subjects received the conditions.
Without mentioning the names it would not be possible for another scientist to reproduce (support or refute) the results. "
Yes it would. They could call them Font A, Font B etc. as Nick suggests and the fonts can be embedded in the report documentation. Or, how about defining specific criteria for specific features and elements of a typeface.
It is the job of font makers or font distributors to substantiate claims of legibility, just as it is with drug companies. Not scientists studying general principals of legibility and readability. Any mention in a scientific report of specific fonts, (which are products), in a scientific study of legibility amounts to an endorsement (or condemnation, depending on the outcome) of that product, which I doubt is the intended take-away of those conducting the research. And is most certainly unfair to all those designers and retailers of fonts that could be equally as good as or better than those that happen to get used in legibility tests, such as Arial, which is there, in no small part because it is a bundled font that happens to be the default sans serif typeface on PCs, and we all have a pretty good idea of how that hapened.
There isn't some magic bullet than can't be defined which one font has and another doesn't which makes one font more legible than another. It's is as simple as such things as relative x-height, weight, contrast, width to height ratio, tracking, how open or closed Apertures are, relative bowl size, presence or absence of confusable characters and so on. These are the things science and engineering exist to measure and evaluate. When comparing fonts they are completely at sea without either ignoring specifics and simply referring to a short list of fonts, which I suppose are regarded as generic examples of various types of fonts and referring to them by name, or delving right in there and measuring everything to come up with some generic criteria.
For what it is worth, the AODA (Accessibility for Ontarians with Disabilities Act) specifies that fonts for use on signage in public buildings must:
.2 Have a width to height ratio between 3:5 and 1:1 (using an upper case X for
.3 Have a stroke width to height ratio between 1:5 and 1:10;
God knows where that spec came from, but there it is. It doesn’t seem unreasonable. It does seem arbitrary, but outside those parameters, fonts used for wayfinding signage do tend to seem too narrow or too wide and too light or too bold. This ratio by the way makes Tiresias SignFont non-compliant on signage in Ontario.
Reproducing results is a major aspect of all scientific research.
My awareness of reading research is pretty skimpy, but it looks like each researcher wants to do their own thing their own way, rather than repeat someone else's project.
Nick, the point is not that all experiments are repeated, but that every experiment must be repeatable such that it can be independently verified that the results are reproducible. If you conduct an experiment that no one else is able to repeat, then your results don't count for anything because they can't be verified. The circumstances under which an experiment is actually repeated vary, but this typically happens if the result is counter-intuitive or contrary to the results of other, related experiments, or if the procedure appears in some way possibly flawed to another researcher, or if a later experiment produces a result that makes the earlier result suspect.
This isn't particular to reading research. It is how experimental science is conducted.
They were kind enough to show us:http://typophile.com/files/song%20schwarz_3412.PNG
Here was the crit:"That's pathetic."
"You should at least set these fonts to the same x-height. "
"As designed, the whole rigmarole is utterly ludicrous"
Then You told us Schwartz said: "...if you'd read the study you would have noticed that it is NOT about type fonts..."
Well, I read the study. So I said: "the widespread and most read answer from this study is that Arial is the best font for reading, and [yet S says] the study is NOT about type fonts, (?)"
Get content, target content, design solution so content hits target.
I thought they taught this in business school. Studies have, after all convincingly proven that students (the subjects of this study), have an extremely low tolerance for any recipe not ending in debauchery of some kind.
You can say, this is not about who the subjects are, and you can say that this is not about typography, and you can say that this is not about fonts, or content, but then what's it about?
They got one right here so far, up to now, to this point, in part one... for an arbitrary match of content and target, something interesting happens when everything is wrong about the type and typography of sample B, and less is wrong about sample A.
Here is a newspaper article about the larger Schwarz program of research. Perhaps it will clarify what they were demonstrating. I'm confident that it never occurred (or to me) that their study could be used to sell more copies of Arial.http://www.boston.com/bostonglobe/ideas/articles/2010/01/31/easy__true/?...
Oh good, thanks.
You now have the ultimate power in all the universe over disfluency. ;)
What will you do?
I think I'm beginning to understand this cognitive fluency stuff.
Research has been done on reading and sound.
How would investigation of another sense, taste, work out?
Subjects would be given two bowls.
One contains Fruit Loops™ and milk, the other an equal amount of jalapeno tortilla chips and milk.
How much of each is eaten in a given period of time is measured.
Based on the general theory of cognitive fluency, it can be predicted that, "if it tastes bad, don't eat it".
Although the study is not, of course, about cereals, news services would pick up the story, "Fruit Loops™ is best for breakfast, scientists find".
People eat food for nutrition, but this research would demonstrate that, because of cognitive processes, taste is also important.
So there may be commercial applications.
After Nick's research was published, a perplexed scientist would attempt to reproduce the experiment but would replace the fruit loops with wassabe. Results would show that no body ever wanted breakfast anymore except for in Japan and Mexico.
Almost Nick, but your example isn’t about cognitive fluency. With reading, the main effect isn't about the speed, ease, or completeness of the reading itself, but about the effect on cognition. People estimated that the completion of non-reading tasks would be faster after reading about them with better typography. In your example, we don’t care if people eat fruit loops faster or more completely. Instead we’re interested in what effects the more delicious breakfast has on your cognition. Will the more delicious breakfast make people predict that they will complete a later task more successfully?
I completely agree with you that newspaper accounts of scientific inquiries are too often misleading or completely wrong. I thought the Boston Globe article I pointed to was well written.
Will the more delicious breakfast make people predict that they will complete a later task more successfully?
When tasting, task-prediction isn't relevant to cognitive perception.
It's not the cognitive purpose of tasting food to instruct people on how to complete tasks, as was the case with looking at the recipe cards.
The purpose of tasting food is to determine, or predict if you like, whether it's good to eat.
Surely the principle of "if it tastes bad, don't eat it" concerns the principle of cognitive fluency, as applied to mental processing of taste perceptions.
In my cereal test, measurement of how much, or how fast the participants eat, indicates the degree of cognitive fluency they are experiencing tasting the food.
Kevin: People estimated that the completion of non-reading tasks would be faster after reading about them with better typography.
Which is why I'm inclined to suspect that the subjects didn't understand anything about the nature of the non-reading task. Anyone who has made sushi before already knows how long these tasks are going to take, and that experience will inform their estimation. So what the results of the study suggest to me is that typography influences cognition in the absence of other cognitive 'handles'.
So what the results of the study suggest to me is that typography influences cognition in the absence of other cognitive 'handles'.
I’m sure that you are correct that your experiences will impact your task time estimations. Cognitive fluency predicts that if you find it easy to make sushi-rolls, you will underestimate how long it actually takes, and you will overestimate how long it takes if you think they are difficult to make. People are not particularly accurate at estimating the duration of tasks even if they just completed the task. I’ve asked study participants how long they just had been reading. With a better typography condition, I found people underestimating how long they were just reading by 25% on average (discussed at ATypI-Helsinki).
I would still bet that sushi-roll making participants will also predict that sushi-roll making is faster when reading the recipe with the good typographic condition than the poor typographic condition.
Kevin: I would still bet that sushi-roll making participants will also predict that sushi-roll making is faster when reading the recipe with the good typographic condition than the poor typographic condition.
Agreed. I would be interested, though, to know if such estimations were closer to accurate than those of inexperienced sushi makers.