User preferences for reading text on the web
A few words on methodology
We devised an on-screen test which allowed participants to change the type face, font size, line height and column width of a piece of on-screen text, adjusting all four until they found the text to be 'most readable'.
There was no need to account for screen resolution, or even a particular user's anti-aliasing settings. The environment they took the test in was the one they were used to so we felt those system variables would be correctly accounted for across the spectrum of respondents.
This test didn't actually measure their ability to read the passage of text; it relied on the user to tell us which settings they found most readable. You could call this 'perceived readability' as opposed to 'actual readability' as my friend and colleague Luke Scheybeler points out.
We ran the survey a total of three times, with slightly differing methodologies.
After the initial survey we realised we needed to randomise the default font to remove what we saw as a probable bias towards Arial.
Then we wondered about the defaults for font size, column width and leading. We set the defaults alternating between extremely high or low values in order to force the user to make a selection for all values.
The first change reduced the preference for Arial by a significant degree, confirming our suspicions about the flaw in our initial survey. The reduction in Arial was taken up by the other sans serif fonts, with Verdana appearing to benefit the most.
The data used for this report are based on all three surveys, with the results of the later ones applied to the earlier ones in order to correct the early flaws.
So, which are the most readable column widths, types faces, font sizes and line heights according to our TextPrefs survey?
Type face is probably the single most important factor when it comes to readability, although all of the variables this report addresses come into play.
Arial pips Verdana to the post, but only just. The closeness is such that it probably lies within the margin of error, so it should probably be declared a tie! Anecdotally I have often heard people express a liking for Verdana, so I was interested to see that reflected in a more formal survey. Both hover around the 20% mark, garnering 40% between the two Microsoft stalwarts.
Helvetica is next up, beating Lucida by a similarly tight margin, at around 15%, a few points above Trebuchet.
The most popular of the serif fonts on offer was Georgia, with a showing almost as strong as that for Trebuchet at around 11%. This font has increased in popularity in recent years, being used extensively by the 'standards compliance' crowd, including Message!
Times, the classic serif font, does predictably badly, coming in at around two thirds the popularity of the more modern Georgia.
Courier, which we put in almost as a 'control', gleans a negligible percentage point or two.
If you client or your design brief requires a serif font, Georgia is likely to be most readable by the majority of a given audience. For your sans serif requirements, you can flip a coin between Arial and Verdana, or bring your design judgement to bear!
The other fonts, despite not being the most popular in our survey, do of course have their places. Courier, for example, is often used in technical websites to display examples of programming code. And you might decide to use Helvetica in preference to Arial for larger headings if you prefer the slightly more sophisticated cut of the characters. (Please don't email me to explain why you think Arial is nicer!)
(Remember though that Helvetica is not installed by default on Windows computers, so for practical purposes the vast majority of your visitors are likely to see Arial in any case.)
In tomorrow's installment: Font sizes!