"Normal People" and Data Journalism

So I come across the following this morning

Why Normal People Don't Trust Data Journalism

and I think, oh cool, some kind of research, some kind of analysis on whether the hottest journalism trend is actually succeeding or failing. We need this.

Instead, we get this. Read it for yourself. Yeah, I know, Business Insider. Get over yourself. Read it anyway. There's a lot to learn here. As the author, Milo Yiannopoulos writes:

Data journalism, “fact check” journalism and the new generation of “explainer” websites, which map out context behind big news stories, sound unobjectionable—but they’re not resonating with the public.

At this point I'm thinking -- okay, cool -- how did you define and measure "resonating with the public?" I'd love to see whether Vox and 538 and Upshot are working or not. Read the story. You'll find no definitions, no measures, no analysis. Not much of anything at all. How the hell do we know if it's "resonating" or not? Just you, Milo? Your opinion? With not a fucking bit of data to back it up? I suppose because, ya know, data is boring. Facts -- boring. If we're boring you, it must not be working.

That's why, Yiannopoulos, data journalism exists. Because people are tired of "journalists" and "pundits" just making shit up.

We won't really know whether such sites are "resonating" until we have a better sense of their traffic, their ad sales, their ability to stay online and active in a competitive media marketplace. And even then they can "resonate" if they are influential, if they play a role in the chattering class and social/political debate. See? Measuring "resonate" is tough, because what resonates for me may not resonate for someone down the road.

We've seen some of these criticisms elsewhere, mostly from a pundit class threatened by the successes of 538 and others to point out how wrong they often are (as Nate Silver famously did in the 2012 elections).

But this is also a conservative response to what is perceived to be liberal sites (though a conservative Heritage version just launched). As he writes:

So-called “actually” journalism doesn’t speak to these people because it doesn’t use the language of emotion or common sense, as most of us do in most of our lives. And too often readers get the sense that there is a gigantic bait and switch going on: in the language of impartiality, they are being fed politics.

Now there's a good point in there somewhere, that data journalism seems bloodless. And I agree -- but that matters only if data journalism pretended to be a single, comprehensive site designed to explain all of what's happening in the world today, like the New York Times does, or other major news orgs. That's not data journalism's mission. First, Milo misunderstands the term "data journalism" and is really talking about these explainer sites, and second there is a testable hypothesis in that graph above, about whether people do perceive a "bait and switch" baffling with bullshit kind of thing happening in the stories. I dunno. That's why we gather data. To find out. Not to just write what we think, making shit up as we go.
 

Now What?

I just uploaded a conspiracy theory manuscript I've been fighting with for months.Now what?I have a couple of studies on deck and I'm not sure which to tackle first. They are:

• U.S. Hispanic consumption of news in English or Spanish. There's some literature here on the effects this can have on voting and political participation. I have great data on this.
• Sports and racism. I've just pulled these data together. I can look at fandom of various sports and whether or not it's associated with racism. This is way out of my comfort zone in the literature.

So we'll see which wins.

Partisan Tilt in TV Prez Coverage?

A study, just out this week, looks at the last three U.S. presidential campaigns and asks that age-old question: is there a partisan tilt? In this case, it asks it about the traditional broadcast alphabet soup networks of ABC, CBS, and NBC.

The answer might surprise you.

Some networks had a Republican tilt, some a Democratic tilt. That'll send some folks off to froth at the mouth, insisting that all media (save Fox, and even then some wingnuts aren't quite so sure) have a liberal-Democratic lean to their coverage. In 2008, there was an aggregate Republican tilt among the networks. Wow. Of course it all comes down to how you code stuff, but I'm writing this in a hurry and may take time later today to get into the guts of the study.

The Berghdal Deal -- A Poll

I stumbled across this Fox poll first via Townhall (yeah, I know). Here's the breathless Townhall story on it with the following hed:

Poll: Vast Majority Fears
Berghdal Deal Incentivizes
Hostage-Taking, Obama
Approval Slumps

It was the "vast majority" that got my attention because sometimes partisan sites see a vast majority in a bare majority scenario (see any time Rush Limbaugh talks about a poll, usually badly). Not so much here. Indeed, more than 8 out of 10 say they have some kind of concern about this. But here's the actual question:

How concerned are you that negotiating with terrorist groups will encourage these groups to take more American soldiers hostage?

Okay, the real issue is -- is this a fair, or is it a leading, question? Is it a talking point? Or a real concern? Keep in mind this is a sound poll methodologically, one that taps landline and cell phones. A pro job. So now we have to ask whether the question itself has any kind of a surprising question, or is merely partisan hackery disguised as a poll.

The only way to truly test this is to ask a more open-ended question and see if this concern pops up without being prompted. My hunch? It would, in part because it's a valid concern, in part because it's been a major Republican and conservative talking point. But mostly the former. Let's look closer.

What's fascinating is if you go to the actual Fox poll report and scroll way down to the very very very bottom. There's a crosstab of this question by age, gender, party, etc. You'd kinda expect to see a serious Republican effect if the "talking point hypothesis" is correct. Seventy percent of Republicans are "very concerned" while only 45 percent of Democrats are "very concerned." That supports a talking points argument, for sure, but you can also argue that nearly half of all Democrats are very concerned about this. That's not good if you work for Obama. We see roughly the same thing between self-described conservatives and liberals.

In other breakdowns, such as age and gender, not a lot to see. Younger are less concerned than older respondents, women and men are about the same. In all, it's party identification and ideology that suggest support for a "talking points" interpretation of at least some of the results, the rest no doubt due to very real concerns about what this deal may mean for the future.

In all, the Fox poll makes life miserable for the Obama administration. And probably makes Hillary Clinton happy to be out of office, with some distance, if and when she runs for president.

News Portals as Great Levelers

We know from recent research that television news, for all its faults, acts as a great leveler of what people know. In other words, TV news helps those who know the least. This study out of Japan suggests that news portals can act in much the same way.

As the authors say in their conclusions:

The results suggest that even people with a strong preference for entertainment can acquire political knowledge by being exposed to news headlines in the topics sections of portal sites. Or, more optimistically, it may be that “[t]he average citizens want to review headlines to make sure they are at least aware of the basic events and issues everyone else is likely to know about” (Neuman,1986, p. 156).  

Call it The Death of Inadvertent Exposure.

Poll Takedown

Here's an interesting takedown of a gun poll that appeared on MSNBC. Scroll to the end, especially the part about the Hispanic women segment. The critic, who points out some issues with the poll, misses an even better criticism. In a survey of 1,000 folks in Texas, exactly how many are likely to be Hispanic women? The margin of error of such a small number explodes and, to be honest, the results become useless. You have to oversample if you want to break results down by ethnicity or race, otherwise you simply have far too few people to make meaningful comparisons.

My own political preferences aside, there's a lot to question in this poll.

Do Debates Matter

Political debates increase political knowledge, especially among the least informed, but this study examines whether the format of the debate can make a difference. To put it simply, as I'm headed out the door, foreign policy debates improve knowledge the most and on such topics the town hall approach actually does a little good, though not much.