Wednesday, July 30, 2014

Good Titles

Some people have a flair for titles of their work published in academic journal articles. You can't go wrong with one like this, the stuff after the colon (The Ignorant Public Reassessed). Even better, this study attempts to challenge the conventional wisdom. From the abstract:
Challenging some established findings, the study shows that people in rural communities know more about local politics than urban counterparts, that women know more about local politics than men and that young people are equally knowledgeable about the EU as older people. The results thus indicate that people are to varying extents knowledgeable about varying aspects of politics.
It's a Finnish study, so it's difficult to know whether it's culture-bound. Still, quite interesting.


Friday, July 25, 2014

Data (the word, not the Star Trek guy)

It's the era of Big Data. So, in honor of that on a Friday, here's the trend in the use of the word "data" via the magic of Google's Ngram viewer, which searches books through 2000. We can only assume since 2000 it's dramatically increased.



And Let the Polling Insanity Begin ...

The Political Insider page at the AJC reports two new polls out about Georgia's U.S. Senate race. Dueling polls, really.
  • A Rasmussen poll has the Republican, David Perdue, over the Democrat, Michelle Nunn, 46-40.
  • Yes, but a Landmark poll has it 47-43 -- for Nunn.
Of course, Landmark had Kingston beating Perdue for the GOP nomination, 48-41. Perdue won by a couple of percentage points. But the real issue here is, of course, which one is right?

Answer: neither

First, both results are within the margins of error, so "leading" here means damn little. Second, both appear to be robo-polls, which are often flawed, or a bit biased, or just plain suck. Take your pick, but just look at how poorly these polls just performed only a week ago. There's a good argument to be made, journalistically, for not even reporting on these polls. Not that that's gonna happen, sad to say. And that's odd. Imagine you're a reporter (I know, but stay with me) and there's a source who pretty much blows bullshit at you day after day. You check the stuff out, it's never right.

Do you keep going back for quotes? Of course not.And yet, and yet. Yes, we're the enablers.

Thursday, July 24, 2014

On My Research ...

UGA pushed out a press release today on my most recent published research. Read it. Over and over again. Make it part of your life.



Talk Radio ... and Concussions?

I have a long history of research into talk radio. Hell, it got me tenure in the 1990s, I published so much while the topic was hot stuff and I was arguably the leader in the field. So I keep up with what's happening in research. I stumbled across this study today on, of all things, recovery from concussions. Here's the key part for me:

While many activities were associated with longer average recovery times, only reading (p=0.024) and listening to audio books, talk radio, or podcasts (0=0.0003) were statistically significant. 

Wow. Talk radio, still evil.

Of course it's more about audio, something about listening, that leads to longer recovery times. I wish we knew what kind of talk radio -- Car Talk? NPR stuff? Rush Limbaugh? Okay, the latter might lead not only to a slower recovery but suicide, but you get my point. And clearly the audio books, talk radio, and podcasts are a combined category. In other words, we can't really blame talk radio. Not completely.

Wednesday, July 23, 2014

In Georgia ... Pick a Poll at your Own Risk

We just had our party nomination runoff elections and we have two really interesting general elections in Georgia to look forward to -- Governor, and U.S. Senate.

Let's look at the polling.

Right now, according to this nifty Wikipedia page (scroll down a bit to the table labeled General Election) the polls between Republican and incumbent Gov. Nathan Deal and Democratic challenger Jason Carter are all over the place. We've got carter up 49-41. No, wait. It's a tie. No, wait again, Deal is up 47-40.

Here's the lesson for the day. Look hard at how the polls were conducted. If it's a robo-poll, ignore it. Take, for example, the recent GOP runoff for U.S. Senate. InsiderAdvantage had Kingston ahead by 5 percentage points. Perdue won by a couple. Yes, it was a robo-poll, meaning landline only, or some odd way of contacting people online in combination with landline phones, which skew older, more conservative. Or, as this story points out, should simply be ignored by journalists. Landmark also has Kingston winning by 7 percentage points. Indeed, every poll but one favored Kingston.



Monday, July 21, 2014

Political Knowledge and False Projection

Among my favorite (how nerdy is this?) research areas is the false consensus effect. A study just out looks at an aspect of this combined with news media exposure and political knowledge.

Simply put, people tend to think others think the way they do. The study finds that, as you'd expect (but not always find) news exposure and political knowledge result in greater accuracy about the true distribution of opinions.