Do people really spend so little time reading news online?

From the December/January 2008 edition of the American Journalism Review:

Newspaper Web sites are attracting lots of visitors, but aren’t keeping them around for long. The typical visitor to, which attracts more than 10 percent of the entire newspaper industry’s traffic online, spent an average of just 34 minutes and 53 seconds browsing its richly detailed offerings in October. That’s 34 minutes and 53 seconds per month, or about 68 seconds per day online. Slim as that is, it’s actually about three times longer than the average of the next nine largest newspaper sites.

The recent Reuters report What’s Happening to our News? gives similar figures for a discrepancy between time spent reading the news online and in print, this time in Europe:

…visitors to the leading UK newspaper websites (as measured by overall traffic) typically only spend a few minutes each day perusing the content. The Daily Mail leads the pack, with an average daily visit of only 8.7 minutes; followed by the Guardian (5.4 minutes), News of the World (3.7 minutes), The Sun (3.7 minutes) and The Times (3.3 minutes).In contrast, McKinsey estimates that, on average, consumers spend roughly eight times longer reading a physical newspaper, compared to the equivalent time they spend at a news-paper website.

In other words, someone who reads the Daily Mail online spends 8.7 minutes doing so, but a reader with a physical copy of the newspaper spends 69.6 minutes at it.

In line with this, (this may have been where McKinsey got their data, I’m not 100% as that data isn’t available online) the UK National Readership survey estimates the average UK print newspaper reader spends 30 minutes reading it per day, with just over 20% spending around an hour.

That’s a pretty big discrepancy between time spent reading online and reading print.  And the AJR sees it as having an effect on the amount of money online news providers can charge for ads:

Harvard’s Patterson offers a theory about why it’s hard to squeeze more money out of online advertisers: Web ads may not be as effective as the traditional kind. “I’m not sure [advertisers] are convinced yet about how terrific a sales tool [a Web display ad] is,” he says. “The evidence isn’t strong yet that it can drive people into a store the way a full-page newspaper ad can. They’re less confident about what they’re getting online.”

Moreover, unlike their here-and-gone counterparts on the Internet, print subscribers still stay around long enough to see an ad. Some 80 percent of print readers say they spent 16 or more minutes per day with their newspaper, according to Scarborough Research.

(emphasis added)

The fleeting nature of online news reading isn’t the only reason to worry about funding newspapers online through advertising, of course – the web is a more competitive environment than print ever was, although I’m still not convinced that it’s as competitive as all that (for the simple reason that some advertisers will continue with a preference for  advertising with big name news sites targeted at certain demographics).  Leaving that issue aside, a bit of speculation.  Calculating the time a particular user spends on a site per day is precise and objective – it’s a figure derived from that site’s statistics.  It’s observational data, in other words.  Time spent reading the print edition of a newspaper is not so objective.  Short of surreptitiously following a reader around for a day and night and timing them, there’s no way to get observational data on how much time they spend reading a newspaper.  The only alternative is a self-report.

The National Readership Survey in the UK asks respondents to rate on a seven point scale how long they spent reading a newspaper the previous day.  The points on the scale are:


A couple of points.  First of all, self-reports of anything are frequently unreliable (e.g. Cronbach, 1970; Fiske, 1980).  Memory is unreliable, and something called social desirability bias can frequently creep in.  Respondents, wishing to present themselves in a favorable light, may (consciously or unconsciously) tailor their answers to make their behavior appear less deviant and more socially desirable (Bradburn, Sudman, et al, 1980).   “The crucial problem with self-report, if it is to be interpreted as a picture of typical behavior, is honesty… .Even when [the respondent] tries to be truthful we cannot expect him to be detached and impartial. His report about himself is certain to be distorted to some degree” (Cronbach, 1970, p. 40).

“Assuming that a researcher constructs a meaningful scale, respondents use the scale as a source of information in computing an estimate.  Specifically, they assume that values in the middle range of the scale reflect the “average” or “usual” behavioural frequency whereas the extremes of the scale correspond to the extremes of the distribution. Hence the use the range of the scale as a frame of reference in estimating their own behavioural frequency.” (Sternberg et al, 2007, p. 70)

Look again at the NRS graph.  It’s almost perfectly normally distributed, with the middle score of the scale shown to respondents (‘About 30 minutes’) being the most frequent response.  There are almost certainly two factors at play here: social desirability, and the effect of the scale in shaping a respondent’s estimate of their reading time.  These two factors likely combine to produce an inflated estimate of that time.

Whether that inflation is enough to explain the disparity between time spent reading news online and in print would definitely be worth some study (maybe someone already has?).  If reading time is a factor in pushing down the cost of advertising online, with advertisers believing their message is not getting across because readers aren’t hanging around long enough to register the ad, it would seem fairly important to figure out if anyone actually ever did spend as much time reading a newspaper as they said they did.

Refs: Cronbach, L. J. (1970). Essentials of psychological testing (3rd. ed.). New York: Harper & Row.
Fiske, D. W. (1980). When are verbal reports veridical? New Directions for Methodology of Social Behavioral Sciences, 4, 59–67.

Sternberg, R.J., Roediger, H.L., Halpern, D.F. (2007). Critical Thinking in Psychology. Cambridge, Cambridge University Press.