Friday, September 12, 2014

The Big Game

Saturday is a big game for Georgia. The Bulldogs travel to South Carolina to play Steve "Evil Genius" Spurrier's Gamecocks. UGA won its first game, a blowout of a good Clemson team, while South Carolina lost its opening game, a blowout loss to Texas A&M. UGA had a bye last week, while South Carolina played a middling game to beat East Carolina.

Okay, enough with all the proper nouns. What's it all mean?

UGA is a touchdown favorite tomorrow. The fine folks at 538 did a nice piece today on the game. I'm only gonna add something about rankings.

Since 2000, UGA has started the season ranked in the AP Poll's Top 10 six times. Only in three of those seasons, though, did it finish in the Top 10. The most painful, probably, was 2008 when Georgia opened as #1 and finished 13th. Ouch. See the table below for pre-season prediction and final ranking since 2000.

AP Poll on Georgia

Season Pre-Season
2014 12 NA, 6th at present
2013 5 *
2012 6 5
2011 19 19
2010 23 *
2009 13 *
2008 1 13
2007 13 2
2006 15 23
2005 13 10
2004 3 7
2003 11 7
2002 8 3
2001 * 22
2000 10 20
* Not in Top 25

Georgia averages a pre-season ranking of about 12th. I say "about" because there's an unranked year, so I called it 28th. It ends with an average of about 16th (there are some rankings in the 30s as finishes). To summarize, UGA averages an AP Poll ranking of 12, and averages an AP Poll finish of 16.

Hopefully this season it'll go in the opposite direction, starting at 14, finishing at ... well ... it's too soon to say.

Thursday, September 11, 2014

Why UGA Won't Be in AAU

I wrote yesterday about how UGA (where I profess) would dearly love to be invited to the prom -- in this case, the Association of American Universities, the organization of the top research schools. I predicted an invitation by 2017 that elite group of 62.

I'd like to take that back.

First, there's this story about a couple of schools either nudged out or who left the organization, complaining:
...the metrics the group uses to assess its members define the contributions of high-quality research universities too narrowly, focusing excessively on biomedical research and the volume of federally supported research.
UGA does a good bit of this kind of stuff, but not anywhere near enough given there's no sizeable research medical school located within the university.

Plus, among the list of indicators is how the schools rank in terms of NSF, USDA, and other grant funding. I looked it up. UGA is #83 on one NSF list, #84 on another. I couldn't quickly located other rankings, but I'd expect us to be a bit higher on the USDA stuff, but not so much as to overwhelm the underwhelming ranks in the 80s.

So that 2017 prediction? Never mind.

AAU for UGA?

Most people have never heard of the Association of American Universities. Say "AAU" and odds are, if anything at all comes to mind, it's amateur sports. Actually, the AAU is an elite club of the very top research universities in the U.S. and Canada. Sixty-two of them.

UGA (where I teach) would dearly love to belong. You can't come out and say it, of course. How academically gauche. You can't apply. Can't beg. No, it works like this:
Membership in AAU is by invitation and is based on the high quality of programs of academic research and scholarship and undergraduate, graduate, and professional education in a number of fields, as well as general recognition that a university is outstanding by reason of the excellence of its research and education programs.  
There's a policy on membership. It's long, it's full of "indicators," (academics wrote it, after all) and really it boils down to grants and research productivity. UGA's lack of major medical and engineering programs does hurt. So who does belong? You can see the list yourself. What I did is download it and sort it by year to see what schools have been added most recently, and how often a school gets asked to the prom. Let's take a peak:

2012 Boston University (most recent)
2010 Georgia Tech

Okay, let's stop right there. Who imagines Georgia Tech would never be happy to see Georgia join its club? Luckily it requires a three-fourths vote to be invited, so Tech alone couldn't stop us. And yet, and yet.

2001 Stony Brook University
2001 Texas A&M

Okay, I know nothing about Stony Brook other than I love the name. A&M, it I know, and am not surprised. Good school.

1996 Univ Cal-Davis
1996 Univ Cal-Irvine

So 1996 is Cal Year. Funny, as Cal-Davis is one of UGA's "peer institutions." So are several others in the club (University of Florida, UNC-Chapel Hill, etc.). Oddly (another post, another day) few of UGA's self-identified "peer" universities list UGA as their peer institution. There's a good story in there somewhere.

1995 Emory University
1995 Univ Cal-Santa Barbara

Emory is another school likely to be less than enamored with the idea of the nearby football factory being invited into its nerdhouse. As you can no doubt see, membership comes in batches, sometimes several years apart (the biggest gap seems to be 9 years, unless I'm miscounting, from 2001 to 2010 when Tech joined). Frankly, given its world-class program, I'm surprised it took that long for Tech to get in.

Will UGA ever get an invite? I'm betting yes. The latest USN&WR rankings name it the 20th best public university in the country. While I haven't studied the rankings closely, UGA is higher on that list than many AAU members. To be clear, this is an apples and oranges comparison. AAU membership is more focused on research and faculty productivity, along with other "indicators," a few of which have to do with actually educating students.

If I were to bet, after studying the list and how often schools are added, I'd predict UGA gets an invite by 2017.

All it needs is 47 out of 62 votes.

Tuesday, September 9, 2014

Attitudes toward Blacks (and sports fandom)

Among the big sports stories this week is, of course, the NBA and racism, this time involving the team just down the road from me (Atlanta Hawks). So I decided to check out the data, see what it could tell us about fans of different sports and their attitudes about African Americans. 
This is a first blush, an initial glimpse of the data, after I merged two connected datasets that include questions on:
  1. how interested respondents are in various sports, and
  2. attitudes toward blacks, etc.
What I've done, after merging, is select out only whites for analysis below. What I'm reporting below are simple correlations on two variables, (1) interest in that sport, and (2) warmth toward blacks on a 7-point scale, so a high score means you like them more, a low score means you like them less.

A positive correlation coefficient means the more interested you are in a sport, the more warm you're also about blacks. A negative coefficient means just the opposite, that less interest in a sport is associated with warmth or, vice versa, greater interest in that sport is associated with less warmth. I've reported only statistically significant correlations.

Positive Relationships

Boxing (.06)
NBA (.05)
NASCAR (.04)

Negative Relationships

Olympics (-.07)
WNBA (-.07)

And that's it. A bunch of stuff, no relationship among whites (NFL, PGA, mixed martial arts, college football, etc.).

What's it tell us? Attitudes about blacks, at least as measured in this survey item, are largely unconnected to sports fandom. The coefficients above are, while statistically significant, relatively weak. If I were writing an academic paper, I might hypothesize that, for whites, sports in which there are large number of black competitors should lead to a positive relationship between being a fan of that sport and attitudes toward blacks. Certainly the NBA coefficient above supports that, at least at the simple bivariate level. Then again, so does the NASCAR number. And what's up with the negative coefficient and WNBA fandom? I have no idea.

This first attitude measure is a bit clumsy, and as the week progresses I'll dip deeper into the questionnaire and pull out something more useful. I should also point out that the survey includes a huge number of sports questions. I just used the most obvious ones. It also includes a very powerful measure of racism, one conducted at the subconscious level to respondents, but it'll take me some work to get it into the analyses here.

About the Data: I used the 2008-2009 ANES panel study and merged the base file with the supplemental data file (not all respondents participated in both, but a couple of thousand did). I have not yet weighted the data in any meaningful way. 

Don't try this at home.

New Stuff (added 12:12 p.m.)

Just ran the sports fandom items against a question on how much a respondent admires blacks. Not a lot different, but similar to above a positive correlation coefficient means the more you like a sport, the more you also admire blacks.

Positives: NBA, NASCAR, mixed martial arts.
Negatives: Olympics, tennis, WNBA, men's college basketball.

So, for example, the more you like the Olympics or tennis, the less you admire blacks -- or the more you admire blacks, the less you like tennis, etc. This is not causality, of course. The college basketball relationship is interesting and confusing.

Monday, September 8, 2014

Time Trumps Data

Too little time, too much data. If I had the time, given the latest NBA racism kerfuffle, I'd merge two data sets and have something interesting to say about it all.

I wrote a bit about it last April, when the first NBA controversy arose. But the list of liking a sport is actually much longer. See my post here. What I really need to do, though, is merge two data sets to make it work. One has the sports attitudes, the other has a long list of racism measures. The same people completed both sets of interviews, a national survey that probably runs into a couple of thousand folks.

Sigh, time trumps data, because the merge -- while I've done it before -- can eat up a few hours when you add in time recoding variables, cleaning, and getting it up and running for analysis. It'd be fun, for example, to separate out white fans of various sports and see if and when racism plays a role. I suspect so.

Friday, September 5, 2014

Letters to the Editor

First off, I'm going to protect the guilty.

I'm reading a scholarly book by a couple of folks not in mass comm but who, as part of their study, include a massive content analysis of newspaper letters to the editor. It's an excellent book (so far), and there's nothing wrong with how they analyzed the letters. Again, I'm protecting the guilty, so I won't even name their academic discipline, but the underlying assumption in their work is people choose to publish letters to the editor, and there's no mention at all of the fact the editors decide what letters get published. Indeed, if you read the book, you get the sense the letters reflect people's opinions on a topic, as if by magic their letters automatically appear in the newspaper.
Given the nature of the study, this is kinda a big deal. As someone steeped in journalism, I am of course well aware of the somewhat screwy process involved in what letters get published, what letters get tossed, and what letters get pasted to the newsroom bulletin board for a good laugh.

Clearly, these two fine scholars -- and again, it's a damn good book -- are innocent of such newsroom experience. Someone should have caught this when the book was being proposed or vetted. Then again, it was probably someone in their discipline, with little news experience, who vetted it.

Last One (I Promise ... maybe)

I've written a few times recently about the military gear passed on from the feds to local law enforcement. So has many other people. One final bit as I put these data away for good. Georgia cops like rifles. Best I can tell, Georgia law enforcement agencies have received 10,265 rifles of various types as part of the program. That's a lot of guns.