Interpreting data is risky business. What might seem clear one moment becomes a ghost in the evidence the next. As my grandpa, a professor of statistics at Yale, is fond of saying, “reading data is high wire act between skepticism and conjecture. No models are ever completely accurate, the most we can hope for is that some are useful.”
Perhaps one of the most seductive syllogisms of data is that of parallel direction. If A. is moving in the same direction as B. and they are even in relative acceleration and velocity to each other, I’d like to assume a relationship between the two. There is a certain logic to this. However, upon more detailed scrutinizing, this assumption can often falter.
The mounds of data that both "Election 2006 Online" as well as “The State of the News Media” presented seemed, at least in part, to be governed by the above assumption as well as some others.
It would, of course, be amiss of me not to provide an example. The “Election 2006 Online” states the following:
"The growing importance of the internet in political life is tied at least in part to the spread of broadband connections in American homes. From November 2002 to November 2006, the share of adult Americans with high-speed connections at home grew from 17% to 45%. These “always on” internet connections draw people to online news of any kind, political news included" (ii.).
Just because “always on” internet connections have increased does not mean that those connection “draw people to online news of any kind, political news included.”
In fact, if you were to consider the framing of that logic for a moment, you would find that the sentence implies that the greater accessibility of the internet causes people to seek online news.
Does accessibility draw people to online news sites? My gut says no. How could greater accessibility to the large window that is the internet result in a specific increase to a small sector?
I’m not sure.
But, let’s consider its viability for the moment. The study goes on to examine the influence of broadband and mentions it 42 more times (43 total). And as the evidence mounts, the less plausible the study’s initial claim seems.
"As we documented earlier this year, those who have broadband at home are different news consumers from non-internet users and dial-up users. In this survey, broadband users are just as likely to turn to the internet for news as they are their local newspaper on a typical day. Some 38% of home broadband users get news online on a typical day, the same percentage as home broadband users who read a newspaper on a typical day. For home dial-up users, however, online news is not as much an everyday activity" (4.).
To me, this section undermines the claim of a draw. Since it presents the same suspiciously same percentage for both the usage of newspapers and internet among broadband users. This data tells me that broadband users are just as likely to turn on safari as they are to pick up the times. It also tells me that broadband has the same drawing power as regular old newspapers.
I might be missing something, though, the original claim might be relative to dial up users. If I were to assume that then the claim is absolutely plausible. 27% of broadband users get most of their election news online, while only 13% of dial up users get their news online.
But, the original claim does not, to my powers, imply or state a relationship between broadband or dial up. It claims that, in no absolute terms, broadband attracts more people to news on the internet.
But, now I might have an even bigger problem. The chart on page 8 clearly states that 27% of broadband users receive most of their campaign coverage online. However, in an earlier paragraph, already quoted in this article, the study claimed that “38% of home broadband users get news online on a typical day.”
Now, we could parse the language. That the chart presents information pertaining to solely the campaign, and the paragraph presents data relating solely to news. But, at this point in time, I’m beginning to question the accuracy of the study’s numbers. I want to see how these percents were calculated in the first place. I’d like to know about this organization, etc.
I hope this excerise in close reading gave everyone a healthy cup of skepticism before class on Monday.
2 comments:
How could greater accessibility to the large window that is the internet result in a specific increase to a small sector?
This doesn't seem like a particularly bold claim to me. Broadband, which is faster and always online, encourages all sorts of internet behavior. So heavy rain raises the water level everywhere; internet gaming, internet chat rooms, and yes, internet news.
Yeah, I was thinking about that when I wrote the piece. I was bothered by the study's choice of words though. The term "draw" I felt was misleading at best.
Truth be told, too. Your read of their claim, I don't think was part of the argument they were making. The way the study represented its argument seemed to imply the water rose in only one area, after raining everywhere.
Post a Comment