Racial Attitudes and Sports

This is less than a first blush, but I'm messing with a large data set that includes questions about how interested you are in certain sports and also happens to have questions elsewhere that asks about attitudes about race and racism.

The news hook is obvious, the Clippers owner being banned for life from the NBA.

This is a difficult data set to work with and I'm trying to grade papers at the end of the semester, but lemme just share a glimpse of what I'm talking about in simple correlations. That means there's a lot of other explanatory factors that need to be controlled for, but just go with it for now. I'm busy on other stuff.So, I'm looking at how interested one is in various pro sports and how that stacks up with how rare one sees racial problems. High scores on the "racial" question means you see them as rare.

• NBA : -.09*
• MLB :  .04
• NFL :  -.05
• College Basketball : .02
• NASCAR : -.08*

What's this mean in English? If there is an asterisk (*) that means the relationship is statistically significant, so a relationship is found only for interest in NASCAR and NBA. In both cases the coefficient is negative, meaning the more interest you have in those two sports, the more you disagree with the idea that racial problems are rare and isolated. To put it another way, the more you follow those sports, the more likely you think racial problems exist today. Interesting.

Here's another. In this question respondents were asked whether racism is a major problem. Being interested in any of the sports above was unrelated to this question. In other words, following these sports, any of them, had no relationship to racial attitudes, at least among the 600 or so folks surveyed.

Again, these are correlations, roughed out in a hurry.

I need to root around in the codebook for a really good measure of racism itself and see how it stacks up with these variables, if at all.

When the Null Hypothesis IS Your Research Finding

In journalism, the notion of native advertising is seen both as controversial and as a path toward financial stability. The question, then, is whether native advertising threatens the credibility of a news site. In other words:

They worry that, at its core, native advertising is about tricking your reader into reading an ad and thinking its editorial content. 

The results in this study (via Nieman Journalism Lab) suggest no.

Read the article for yourself, or the original research here (begins on page 78 of the PDF). Lemme point out I know nothing about this journal or its peer review process.  Skimming the research itself, I like its use of a college student and human (i.e., adult) samples as a comparison. I hate the use of "mediums" (please, media), but otherwise the procedure seems straightforward. Subjects were randomly assigned to view one of two versions of a site -- one with native advertising, one with "traditional" advertising.

Okay, but here's the weird part:

H1: Ad type will have minimal effect on credibility.

All of you who suffered through and passed a graduate class in research methods will recognize this as an example of hypothesizing the null hypothesis. In other words, this is usually not done. The null hypothesis is of no effect and you hypothesize a difference, and what they've done here is run an experiment with a fairly mild, subtle manipulation and have taken the lack of an effect on credibility as a, excuse the pun, credible finding.

No. No. No.

I'm buried in grading so I can't vent more, but this result should not be taken as evidence that native advertising has no effect on a news site's credibility. Perhaps there is no effect, but this study is not conduced in a fashion to truly answer that question.

Not Gonna "Make"

I hoped to teach a graduate class this summer in secondary analysis. In other words, how to avoid IRBs and human subjects committees and analyze all the data that's out there, freely available, for scholars to exploit. I'm talking ANES, GSS, ICPSR, ABCD, EFG, and all the other acronyms.

The class is stuck at four. Ya need five to "make." I've told the grad office to let students know so they can make other plans.

I could push it, maybe. Certainly other faculty in an unnamed department downstairs have insisted their grad class "make" with fewer students, but that's not me. I certainly have enough research to do in the summer without teaching a class -- and it's partly my fault, or my vocal cord's fault, that I had to offer it in the "through summer" session rather than one of those short, brief, all-at-once semesters. I can't talk four hours a day.

Anyway, this means a lot of other stuff I need to do -- manuscripts, yard work, PlayStation games, will get my attention instead.

The Importance of "Say"

So this tweet just came across my Twitter feed:

The reasons why Americans own guns have changed over time. http://t.co/6WUefY6VsV pic.twitter.com/OLDlNsY9Yv
— Pew Research Center (@pewresearch) April 28, 2014

Let's think about this for a moment.

The fine folks at Pew, who do amazing work, have a survey that suggests people are increasingly likely to own a gun for protection. Face it, a 22-percentage-point increase, that's significant. That's huge. That's ... friggin scary, as in mean and scary world scary. Hunting going down, that's reasonable and not particularly surprising. But has the world become so very scary? Not really. Crime has, instead, gone down in this time period.

Let's talk, methodologically, about a tiny word. That word is: say.

My lede for this would be that the reason people say they own a gun has changed. I suspect there's a lot more at work here, more than a single survey item can capture. You'd need some follow-up questions and a bit of sophisticated analysis to tease out what's happening. Ultimately, though, what we have is a crashing of perception and reality. Fear. And a host of other partisan and ideological mishmash of information that probably explains this.

Of course, perception is reality. So perhaps using "say" isn't necessary after all.

And thus ends this methodological moment.

Walking and ...

As we come into campus many mornings I'll see this young woman at the intersection of Oak and Baldwin streets, near Carr's Hill, waiting to cross. She's always looking down, staring at the object in her hands. Even when she crosses.

And no, it's not a phone. It's always a book.

I find this fascinating. She may be the only person at UGA walking while reading a book rather than a smartphone. I've wanted to get out and ask her why. Is she a word nerd? Work at the nearby grad school office? Studying for a big GRE or MCAT or LSAT? Any close calls while walking in most mornings, eyes locked on her book? And does she even own a mobile phone?

There's a story there.

That Pesky Margin of Error

There's a Politico horserace poll story today that, among other things, finds Rand Paul leading Hillary Clinton in Colorado.

Sen. Rand Paul appears to be the man to beat in Colorado in 2016, a new poll says.

Colorado voters would favor the Kentucky Republican over former Secretary of State Hillary Clinton by 48 percent to 43 percent in a potential 2016 presidential race, according to a new Quinnipiac University poll.

The poll has a 2.7 percent margin of error.

Let's do the math:

• 48 + 2.7 means the actual result could be from 45.3 to 50.7
• 43 + 2.7 means the actual result could be from 40.3 to 45.7

So here's the thing. Paul could be as low as 45.3 and Clinton as high at 45.7. That's called a statistical tie. Yes, only barely a tie.

On the other hand, if you look at the polling shop's previous Paul vs. Clinton polling it's been fairly consistent, with Paul at either 47 or 48 percent in Colorado, Clinton at either 43 or 44 percent. So while the pesky margin of error suggests something of a tie, if you look at the polls as a whole, it's fair to say Paul leads Clinton in that state.

Fudging the Numbers on Race-Based Admits?

As many of us know, there was a U.S. Supreme Court decision this week about affirmative action in universities, essentially upholding a Michigan law that bans attempts to diversity campuses by race.I'm not getting into the for it or against it argument. Instead, this quote in a The New York Times article caught my eye:

“I think this issue is largely settled,” said Ward Connerly, president of the American Civil Rights Institute. “Most Americans have made up their minds that the government should not treat people differently based on race, and they’re kind of impatient that we continue to wrestle with the question.”

Whenever a partisan hack sums up public opinion, I wonder whether the person is right or just blowing smoke out their ass. So, what do people think? Luckily, Pew released a survey recently on just this topic. See the graphic below.

The poll above suggests smoke is being blown out of said ass. But not so fast, my friends. Note the wording. Is affirmative action a "good thing?" Well, yeah. Sounds good to me. But what about race specifically? This survey for ABC asked respondents:

Overall do you support or oppose allowing universities to consider applicants' race as a factor in deciding which students to admit?

The result? Seventy-six percent of U.S. adults oppose race as a factor (62 percent do so "strongly") and only 22 percent support it (and only 8 percent "strongly").

Our lesson here? When it comes to surveys, it all depends on how you ask the question, but in this case the guy quoted in the article seems to be right in summarizing the public's thoughts on the issue.