Monday, August 29, 2016

New College Rankings

Yet another college rankings is out, though this one is different. Washington Monthly's methodology includes student outcomes you don't normally find in other rankings. Yes there's research rate and graduation rate, but also Peace Corps and community service, and salaries students make upon graduation.

Here's the link to the national university list. UGA, where I teach, is #55. Not terrible. Here's the list of SEC schools in order, with national ranking in parentheses:
  1. Texas A&M (6)
  2. Florida (18)
  3. Vanderbilt (22)
  4. UGA (55)
  5. Missouri (80)
  6. LSU (98)
  7. South Carolina (109)
  8. Auburn (139)
  9. Tennessee (172)
  10. Arkansas (208)
  11. Kentucky (216)
  12. Alabama (260)
That Bama ranking. Ouch. Just ouch. University of Alabama-Birmingham is #124, the highest in the state (best I can tell, just eyeballing the data). 

Revisiting Titular Colonicity

It was back in 2008 when I first wrote about titular colonicity, which isn't near as interesting as it sounds and has to do with the growing use of colons in academic titles, especially as a field matures or becomes more complex. I even did a quick-and-dirty analysis of my field's major journal, Journalism and Mass Communication Quarterly. I've often considered a more formal analysis to submit to a journal, but I suspect it's such a navel gazing exercise that most would pass on publication. Here's the piece that created what's called the Dillion Hypothesis of Titular Colonicity. Always worth a look.

OK, fine Hollander, but why mention this today?

Because I stumbled on this abstract, which talks about the length of journal article titles but also, apparently, touches on titular colonicity. Unfortunately I don't have access to the entire piece (see also this recent analysis). I'd never heard of APA's recommended 12 words or less in an academic journal title. That's interesting (if you're a PhDweeb like me, at least).

So work continues to be done and I really really really should tackle a mass comm journal analysis of colons. Problem is, who the hell would publish it? Maybe as a conference paper. Yeah. Maybe. If nothing else it'd give me the opportunity to say "titular colonicity" in front of others. I'm not above such sophomoric enjoyment.

Friday, August 26, 2016

The Ethics of ... Niagen?

I was discussing with students in class the other day the various journalism codes of ethics that exist, what they say, etc., when I pointed to this page via SPJ that provides a list of "Other Codes of Ethics (see left). Go ahead, visit the page. Click on the RTNDA code.

When you do, you get not exactly the most ethical of codes. It's an ad for niagen, some bogus supplement. What's fascinating is if you look at the URL, it's on the RTNDA server. The url is: If you visit the legitimate RTNDA site and click ethics, it takes you to the real code. How someone managed to plant this bogus supplement ad on the RTNDA site, I'll never know. And why SPJ refers to it, I'll never know either.

Thursday, August 25, 2016

UGA Football Parking Tix

Updated Info Below

It's coming up on UGA football season again (Go Dawgs). As I play with data from all parking tickets written on campus last academic year, it makes sense to look specifically at football parking violations. Here's the short answer -- there were 133 tickets (out of 23,265) that have "football parking" listed as the violation.

I assume some other tickets written on football Saturdays may be listed under some other charge, like "sidewalk/grass," but maybe not. Anyway, here's a breakdown of where the football parking tickets were written:
  • South campus deck: 64 tix
  • North campus deck: 37 tix
  • Carlton Street deck: 24 tix
  • PAC (on east campus, near Music): 8 tix
Obviously each and every one of these were written on a Saturday (duh). Indeed, only 134 parking tickets were written all year on a Saturday, and 133 of them were for football day violations. Yup, we have our priorities in order here at UGA.


I had some more time and played around with the football ticket data. Here are a few factoids:
  • All but one football ticket was written in the morning, nearly all between 6 and 8. Makes sense.
  • There were 19 tickets on 9/5, the first game day, and 28 on the second game day (9/19). I lost interest in counting after that. Sue me.
  • The type of car that parks illegally on game days doesn't look all that different from other tickets, with Toyota leading the way, followed by a tie for second between Honda and Nissan.
  • Same with color, the most popular football day illegal parker drives a black vehicle. Two illegal parkers drove orange cars, the same Kia hatchback on two different days (you thought it'd be a Tennessee fan, didn't you?).
  • Often it appears the same car parked illegally on different game days. I can't be sure about this because I don't have unique identifiers for the vehicles, just the tickets, but for example a black, 4-door Audi got tickets on four different game days in the same lot. Asses. I see this a lot in the data, too often to be coincidence. The same Mercedez-Benz (I suspect) has tickets in the same lot on two different game days. 

Survey Story

Updated Below

There's a good Red & Black story out today on a UGA survey I've critiqued before (see herehere, and here). I don't want to repeat my earlier critiques other than make a few comments about the poll and the story itself. First, there's this graph (bold face by me):
The results were released in July, with an overall response rate of 23 percent, or just more than 10,000 people, according to the executive summary. This was below the target goal of 30 percent, which Cook said was needed for the survey results to be generalized for the entire university population.
I've looked hard to find any justification for that 30 percent in a non-probability survey (nice way of saying a self-selected survey) being generalizable to the population. I can't find it in the survey literature, I can't find it on major survey sites like AAPOR, I can't find it anywhere. I admit I could be missing it, and this 30 percent comes not from the university spokeswoman but the consultant. It's in the report, but with no citation, no support, no justification. Given the 23 percent response rate loaded with far too many admin/staff types, I honestly don't think much of this survey, but that's a different issue.

And then there's this quote just above the 30 percent graph. Here I take off my public opinion hat and put on my journalism professor hat.
“The survey was really an attempt on the part of the University of Georgia to hear the voices and to learn more about the experiences of faculty, staff and students here at the University of Georgia,” said Michelle Cook, associate provost for Institutional Diversity, who chaired the 11-member committee that oversaw the survey.
I could be wrong, but this stinks of an emailed answer to a question provided in advance. If I'm wrong, I apologize. If I'm right, then dammit you should say in the attribution that it was in response to an emailed question. Be up front with the reader, especially given UGA flacks and their crutch use of requiring emailed questions and emailed responses -- which no news org should ever accept.

[I was wrong. Nate Harris, news editor at The R&B, told me via Twitter that his reporter got the quote in a face-to-face interview.]

Again, maybe she actually said it over the phone in exactly that way, but it reads like a written response to me. Tell me, the reader, if it's so.

And now something to praise. That's a helluva good graphic in the story comparing UGA's actual gender and race breakdowns with the survey's breakdowns. And other than my quibbles above, it's a damn good story with lots of comments from campus folks, especially students. Good work on that.

Wednesday, August 24, 2016

The Prez Debates

We're already asking who's gonna win the presidential debate and the first one is still a month away.

I saw this story and then immediately hunted up the crosstabs to see better breakdowns of the data. First off is the lede:
Americans expect Hillary Clinton to do a better job than Donald Trump in the presidential debates this year, according to a new HuffPost/YouGov poll.

By a 10-point margin – 45 percent to 35 percent – Americans think Clinton will outperform Trump, while another 20 percent aren’t sure.
So the news is people expect Clinton to out-debate Trump. No real surprise there. And no surprise if you read later than Dems are more likely to see Clinton will win and GOPers more likely to see Trump as the winner.

Now let's look deeper. Here are the crosstabs, which gives us breakdowns on the "winner" question by gender, age, race, party ID, income, and Census region (South, etc.).

Again, nothing is gonna shock you, but it's interesting nonetheless. Every age category sees Clinton as doing better, though younger respondents under the age of 30 are most likely to predict it. Whites also predicted Trump doing better. Interestingly, the middle income bracket of $50,000 to $100,000 is the only one to predict Trump will win, no doubt due to him having more supporters in that category. Because, face it, people tend to believe their own candidate will do well in a debate and, afterward, will report their loser actually won. People see what they want to see, what we in the biz called the theory of motivated reasoning.

What Clinton has to fight, of course, is high expectations going into the debate. Her campaign folks will in a few weeks start praising Trump's debate skills in hopes of raising expectations of his performance and dampening expectations of her performance. Keep an eye out for it.


Monday, August 22, 2016

A Ticket Time of Day

What time of day are you more likely to get a UGA parking ticket? I finally got around to knocking out the specific times. It wasn't as easy as you might think. My Excel skills are slipping. So here we go, the top times of the day when tickets are written:
  1. 8:30 a.m.
  2. 9:40 a.m.
  3. 10:36 a.m. (see a trend here?)
  4. 2:10 p.m. (wait, what?)
  5. 9:50 a.m.
  6. 9:46 a.m.
  7. 2:07 p.m.
  8. 9:45 a.m.
  9. 10:14 a.m.
  10. 9:41 a.m.
As you can see above, 8 of 10 are morning times. If we flip this, the least likely times to get tickets are, obviously, evenings or very early in the mornings. There are way too many 1-ticket-written times to list here. However ... there was one ticket written at 1:07 a.m.  That's an early ticket, written (of course) in the Ramsey lot.

Fun Fact: A silver Tesla got three parking tickets, all bunched together (9/15, 9/22, and 9/28). But in three different lots (Physics, Brooks, and Lower Brooks). I can only assume that third ticket got the overindulged Tesla owner's attention.

Fun Fact II: A gray Maserati got two tickets, one at Ramsey Center and another nearby at East Village. Not nearly enough, to me, for anyone driving a Maserati.

Fun Fact III: The most ticketed vehicle at UGA is a Toyota, followed by Honda, Ford, and Nissan. This, of course, merely reflects how many staff, faculty, and students drive those popular brands. Black cars got the most tickets, pink (yes, pink) the fewest (three the entire academic year, and yes in all three cases it appears to be the same pink scooter).

Friday, August 19, 2016

New Georgia Prez Poll

There's a new Georgia presidential horserace poll out (crosstabs here). The lede is, of course, it's a tie between Hillary Clinton and Donald Trump (43-43).

But I want to point out something interesting I found deep on page 12 of those crosstabs. It breaks down the questions by landline versus cell phone method of reaching respondents. Respondents who were reached by landline favored Trump 52-36 over Clinton. Among those reached by cell phone, Clinton wins 59-21.

Beating the UGA Parking Ticket Odds

So what's the best or worst time to park in the most ticketed UGA parking lots? I'm glad you asked, because I've got the data. Below I list the most ticketed lots and the day or time you're most and least likely to get a ticket. By time I use a gross measure of a.m. or p.m., not specific time of day like 9:30 a.m.

Legion Pool. Of the 1,448 tickets written at the top lot over last academic year, most were written on Wednesday. Tuesdays and Fridays are the least likely to see tickets. IMPORTANT -- more tickets are written in afternoons than mornings in this lot, especially Wednesday afternoons.

Ramsey Center. This lot always ranks high, with 1,408 tickets last academic year. Thursdays are the worst, but basically it's bad every day. Fridays are best. Mornings are worst for tickets

East River Road. Our third place lot is ticketed pretty much every day, but Mondays are slightly the worst for its total of 743 tickets. By far mornings is the popular time for this lot, with only 1 out of 8 tickets coming in the afternoons.

Ransey Lower. Yes, Ramsey again. Of the 676 tickets written in this lot, Wednesday is the worst, but only barely. Afternoons worst, making it different than Ramsey Center. Weird.

North River Road. Spread across all days, these 556 tickets, but Monday is the worst. Fairly even on mornings versus afternoons.

Overall, Monday is the day most tickets are written, 22.2 percent of them. Every day we see fewer and fewer tickets written. See the graphic below for raw numbers. By the way, it's no surprise that Monday mornings are the worst overall.

Thursday, August 18, 2016

Parking on UGA Campus

It's data time.

I have all the parking tickets written on the UGA campus last academic year. For fun, let's compare 2014-2015 with 2015-2016 (last year) to see if any differences emerge. First off, more tickets were written in the recent year (23,265) than in the previous year (23,029).  That's a modest 1 percent increase.

Below are the top 10 ticket locations for the two academic years. Legion Pool topped Ramsey, which is a surprise. I'd love to know what's driving that (every pun intended). Health Sciences drops, perhaps fewer students going there for pizza at The Niche. There are some interesting changes as well, such as North River Road now a factor, and Driftmier Engineering. Yup, there's a story in here somewhere.

Ramsey Center
Legion Pool
Health Sciences
Ramsey Center
Ramsey Lower
E. River Road
Legion Pool
Ramsey Lower
E. River Road
North River
Life Sciences Upper
Driftmier Eng
Academic Achieve
Carlton St.
Carlton St.
Chi Psi House
Life Sciences Upper
Hardman Hall
Health Sciences

Of course I also have other data, such as time of day the tickets were written, and which days were more likely (Monday, Tuesday, etc.) to see tickets on cars (spoiler, it's Monday). I'll do another post or posts to get into that.

Fun fact, only one ticket out of 23,265 last academic year was written on a Sunday. Turns out it was a ticket on November 1 in the East Deck for "unauthorized area," which I assume it means someone invented a space.

Wednesday, August 17, 2016

A Statistical Tie?

The conservative site Breitbart breathlessly reports a new Zogby poll that says Donald Trump and Hillary Clinton are in a statistical tie, with Clinton at 38 and Trump at 36 and, as the Zogby site says, "we are back to a close race!" Let me stress the exclamation point is theirs. You're only allowed three in a career and I wouldn't burn one of mine on this.

OK, but is it a tie? No. Of course not.

The Zogby poll is probably an outlier, a firm that gets only a C- from 538's famous pollster ratings. The internet-only Zogby poll has a Republican house effect, meaning it tends to lean that way in its results. Pay attention to the quality polls, especially the ones that use live phone calls to landlines and cell phones.

Tuesday, August 16, 2016

Fox Gets it Wrong

The Fox News site has a Bias Alert section (insert your own joke here) and today it paid attention to MSNBC's Morning Joe and a comments from Cokie Roberts. Here's the lede:
Anyone supporting Donald Trump is “morally tainted,” ABC and NPR commentator Cokie Roberts declared Tuesday, while making no mention of possible ethical dilemmas facing backers of Hillary Clinton.
I happened to have been watching this morning when Roberts spoke, and she wasn't talking about "anyone" but the conversation was about Republican leaders and elected officials. So the bias alert is, in itself, biased.

I'm no Roberts fan, not at all. Don't think much of her insights or analyses, and mostly she regurgitates what we already know, but Fox gets is wrong here in trying to score a couple of cheap partisan points.

Work on your bias alert.

Prez Debates

The debate commission has released which polls will be used to determine if a third-party candidate, in this case Libertarian Gary Johnson, will make it to the presidential debate stage. The polls are:
  • ABC-Washington Post 
  • CBS-New York Times 
  • CNN-Opinion Research Corporation 
  • Fox News 
  • NBC-Wall Street Journal
According to a CNN report, the numbers for Johnson if you add up the results so far is 10 percent. I've not gone back and verified this, but let's assume CNN can count. The commission long ago said 15 percent would be the threshold to be on the stage, but according to Politico it sent signals recently that if Johnson gets close, he may earn a spot.

Ron Faucheux reports in his daily email blast on polls: "Note in the poll that Gary Johnson, at 11%, is getting a little closer to the 15% debate threshold." But this is a report on the NBC/SurveyMonkey poll, which is different methodologically from the NBC-'Wall Street Journal poll (if I'm wrong on this, let me know). So I don't know it means a lot unless, being new it signals some trend for Johnson.

My take

If Johnson can get solidly within the margin of error of the magic 15 percent mark the commission gives him a seat on the stage. A recent ABC/WaPo national poll, one of those included above, had him at 8 percent and a 4 percent margin of error, meaning he'd miss it on that single poll.  If I have time I'll compute his running average on the five polls above with a margin of error, but figure he needs to hit around 12 percent as an average to make the MOE argument.

Wednesday, August 10, 2016

Is the Election Rigged?

Of course the upcoming presidential election isn't rigged -- but that hasn't stopped Donald Trump from suggesting it (coincidentally as he was falling further and further behind in the polls).

How many folks buy his "rigged" argument? Yes, we've got a poll for that. If you look here you can see the breakdowns of a Bloomberg poll. The question was:
When it comes to the presidential election, is it your sense the election will or will not be rigged?
Turns out, 34 percent say it's rigged and, thankfully, 60 percent say it's not rigged. Six percent aren't sure.

I don't have access to the crosstabs, but according to this article, among Trump supporters 56 percent say it's rigged. Because, I suppose, Trump says so. Our elections are screwed up sometimes, but it's not about invalid voters or cheating, it's more about making it too hard for people to vote, for lines being too long, and all that.

Why does this matter? This deserves a very long, very thoughtful piece, but I'm not gonna bother. Let me just say that a sense of a "rigged" election hurts democracy, harms people's faith in the electoral process, and is generally bad all around. I can't do it any better than this NPR piece does. Read it. Save me the writing time.

Tuesday, August 9, 2016

Ranking University Food

Here's a perfect demonstration of how different ranking systems get you completely different results. In one system, by Niche, UGA is #8 in food on campus. On another, CollegeRank, it's not even in the top 50.


On CollegeRank I couldn't find much on methodology, how they decided who is #1 and who is #50. Yeah, there's some writing at the top with fancy words like versatility and sustainability, I just couldn't easily find out how they measured this stuff. Niche is more upfront with its methodology, though it hardly rises to the level of that term. Turns out, 85 percent is from visitors to the site, another 15 percent based on the cost of meals via Department of Education stats. As we all know, the people who visit and rank something on a site either love or hate it, rarely are they in between.

That said, UGA has good food for undergrads on a meal plan (winner of 81 national awards, it says so right on its web site so it must be true). No mystery meat like when I went to college. There are milkshake bars, coffee bars, and a jillion other things designed to keep happy our overindulged yuppie larvae. It's all about competition, about being a better place than the other guys, and dorms and food and recreational facilities are all part of the competition. We can pamper better than you can pamper, that's our motto, should we ever start having a motto.

I should add that BestColleges has UGA as #17 in food. I'll be honest, this one seems a little more rigorous than the others and it even mentions the national award given to university food services. So maybe it has a clue. And Business Insider had UGA #8 in best dining hall. Makes you wonder about that failure above to make the top 50.

Monday, August 8, 2016

UGA Frats & Sororities

Back in 2014 I pointed to this page that ranked fraternities and sororities at UGA. I happened to look again today as we're cranking up for rush again (classes begin Thursday). The rankings from that earlier post have hardly changed, mainly because I suspect the data from 2014 is still included in the recent rankings.

Rank 2014
Rank 2016
Sigma Chi
Phi Gamma Delta
Phi Gamma Delta
Sigma Chi
Pi Kappa Phi
Pi Kappa Alpha
Sigma Nu
Phi Delta Theta
Kappa Sigma
Kappa Alpha

Pi Beta Phi
Pi Beta Phi
Delta Zeta
Delta Zeta
Delta Gamma
Delta Gamma
Alpha Gamma Delta
Alpha Gamma Delta
Gamma Phi Beta
Gamma Phi Beta

Read my 2014 post for a discussion of the strengths (few) and weaknesses (many) of this ranking system. I see little reason to repeat it here other than to say the weaknesses far far far outweigh any strengths you may find in the ranking scheme.

Sexual Identity at UGA

There was this huge survey conducted at UGA last year and the report is now available. I've been slowly working my way through it. Here's a table I found fascinating.

Sexual Identity
Not Listed
Missing Data

I doubt there's little surprise here on the sheer dominance of hetereosexual over the various other identities at a place like UGA. Plus this sample is made up of students, faculty, staff, and administrators, so all the respondents other than students may skew the numbers.

If we look at undergrads, there aren't a lot of differences. The breakdowns provided aren't complete, but it shows 87.6 percent are hetereosexual, 11.0 percent LGBQ (including pansexual), 0.4 percent asexual (no sexual feelings or desires), and 0.6 percent "other." By comparison, 90 percent of staff/administrator is hetereosexual.

Transgender has received a lot of news attention, of course. In this survey, 39 undergrads identified themselves as transgender, and 11 grad students did so. Six faculty and 10 staff/administrators said they are transgender.

American Accuracy

What kinds of political knowledge questions are Americans best at? I was messing with a 2012 national data set that includes several such questions so I decided to run a quick-and-not-at-all-clean analysis to see which questions respondents performed the best on, which they sucked at.

% Correct
Terms Prez Can Serve
Who is VP
Size of Defense Budget
What is Medicare
Who Controls House
Mitt Romney’s Religion
Who Controls Senate
Who is Speaker of House
Barack Obama’s Religion
Length of Senate Term
Spending Comparison
Who is Chief Justice
Who is Prime Minster UK

Lots of folks know the president can serve up to two terms, or the name of the vice president. Far fewer know the names of the chief justice or the prime minster of the U.K. No surprise there. And Obama's religion scores rather low because of the various theories that he's secretly Muslim.

New Georgia Poll

There's a new poll out on Georgia.

Ignore it.

Well, kinda sorta ignore it.

Here's a brief AJC story on the poll, and if you're more interested go here for a better breakdown from the polling firm itself. The AJC story fails to mention this is a robo-poll, an automated phone survey that probably -- though it's unclear -- called only landline telephones (by federal law you can't robo-call cell phones).

Robo-polls are notoriously flawed, so be skeptical. How skeptical? 538's new Georgia forecast lists the poll but grades the polling firm only with a "C" for historical accuracy and quality of methodology, which is damn low. 538 weights polls in its overall average, with extra points for being more recent. Even with this new poll's recency, it's weighted only 0.49 while a poll from last week by the AJC (that used humans calling landline and cell phones) gets weighted 0.65 in 538's overall forecast. That should tell you something.

Back to that new Georgia poll. It gives Hillary Clinton a 44-37 advantage over Donald Trump. The margin of error is 4 percent so, technically, this difference falls just within being considered a statistical tie. That said, the handful of recent polls all point to a slight Clinton lead, all within the margins of error. Georgia may indeed be in play, which is good if you're a political junkie and bad if you hate to see the airwaves filled with political ads.

Note the breakdown provided does not include age. Often you end up with far too few young respondents in a robo-call to landlines. It's unclear that the data were weighted in any way.

UGA Starts This Week. And That's Stupid.

UGA starts this week.

Yeah, that's stupid.

UGA classes begin August 11, the earliest of any school in the SEC. By far. Check it out below:

8/11: UGA (yes, way too early)
8/16: Auburn and Mississippi State
8/17: Alabama and Tennessee
8/18: South Carolina
8/22: Arkansas, Ole Miss, and Florida
8/24: Kentucky, LSU, Missouri, and Vanderbilt

Wow, so four schools start two weeks later than us. Our 8/11 start is the earliest I can find in the history of UGA, but I'm only going back to 2005. In one previous year we began on 8/12, but nothing as early as 8/11. And of course back in saner days of the quarter system we didn't begin until mid-September -- as it should be.

To make things worse, we start on 8/10 next year. Effing insane. And again on a Thursday. And again, stupid.

I wonder whether UGA is opening classes earlier than any other major university in the country. That would take forever to figure out and I'm not about to try, but my casual eyeballing of the data says this is the case. But for fun let's look at a few of UGA's "aspirational" institutions, schools we'd like to be like. When do they start? Well, UNC and Virginia are on the list and they begin classes on 8/23. Penn State and University of Texas are aspirationals too, and they open 8/22.

Saving myself the time, let's just argue that UGA is opening earlier than any other major university and ask ourselves, WTF?

Friday, August 5, 2016

A Libertarian in the Presidential Debates?

Can Gary Johnson, the Libertarian candidate, get a seat in the presidential debates? I think probably not, but let's dig into the details.

The debate commission criteria present three challenges. First, you must be qualified to serve as president. That's an easy one. Second, you must be on enough state ballots to mathematically have a shot at the 270 electoral votes needed to be elected president. That one's covered.

The third one, though, is tougher. Here's the language with my boldface:
The CPD's third criterion requires that the candidate have a level of support of at least 15% (fifteen percent) of the national electorate as determined by five national public opinion polling organizations selected by CPD, using the average of those organizations' most recent publicly-reported results at the time of the determination. CPD will rely on the advice of a recognized expert or experts in public opinion polling in determining the polls it will rely upon. The polls to be relied upon will be selected based on the quality of the methodology employed, the reputation of the polling organizations and the frequency of the polling conducted. CPD will identify the selected polling organizations well in advance of the time the criteria are applied.
The polls will likely be "gold standard" surveys, meaning they use live callers (not robo-polls) of both landlines and cell phones (probably not online panels, though there's a good one out there). The major news organizations spend the money for such quality polls. Pew also produces these quality polls, as do a few other shops. Looking at the most likely polls, Johnson ranges today between 8 and 12 percent -- and he's dipped a bit in the most recent polls, but we're talking a point here or there, a statistical blip. Can change in a heartbeat, or a Trump miscue.

The commission does not make a decision on who gets a chair on the stage until after Labor Day, though of course in plenty of time before the first debate, which is Sept. 26.

Remember, third party candidates poll better than they actually perform in the election. Gary Johnson also ran in 2012. One poll had him at 6 percent support, another at 7 percent. He captured 0.99 percent of the popular vote. In other words, a hair under 1 percent, while polling at 7 percent.

My money is Johnson doesn't make the stage and instead will find a ready seat on any number of late-night television shows to make his case.

Weighting the Survey Data

There's a new Georgia poll out that shows Hillary Clinton 4 percentage points ahead of Donald Trump, just within the margin of error but signaling that the state is, indeed, in play.

Election surveys are a mix of art and science. The art can be question wording and the special sauce that goes into deciding who is a likely voter. The science is how to weight the data to reflect the population you're trying to describe. Simply put, survey responses are usually off to some degree -- too many women, too few young people, not enough African-Americans, etc. In Georgia surveyors can look at the demographics of previous elections and "weight" their results to reflect the population and those who vote.

Here are the crosstabs of a new AJC poll. Most folks don't bother with this stuff, but you'll find interesting tidbits. See the two rows called Unweighted Base and Total Respondents? Unweighted Base reflects the raw numbers, in this case 847. The Total Respondents reflects (I'm guessing) the weighted numbers of 767. It's possible the total reflects "likely" versus "unlikely" voters but I can find no evidence of that, so I'm gonna assume I'm reading this correctly until told otherwise.

Check this out:
  • Unweighted number of young respondents (ages 18-39): 165, or 19.5 percent of total unweighted respondents.
  • Weighted number of young respondents: 248, or 32.3 percent of total.
How did they invent 83 respondents? They didn't. Again, if I'm reading these xtabs correctly, they've weighted the data to reflect how many young people failed to respond to the poll but actually participate in elections. (We can argue about 18-39 being "young" on another day). That means each "young" person's opinion gets a little heavier in the final calculation. When that happens, someone loses, probably older respondents. In this survey, those 65 and older made up 29.9 percent of the unweighted base, but only 19.6 percent of the "Total Respondents." We see the same for blacks, though not as dramatic, with 26.2 percent of raw respondents in the survey but 30.8 percent in the "total respondents" category, meaning they got weighted slightly higher and whites slightly lower.

What's this all mean? In Georgia, not much, as we have good socio-demographics collected at the voting booth, giving us a solid comparison of previous voters. Some states do this, some don't. Often we have to guess as to "likely voters," which is the special sauce I mentioned above. Different polling shops have slightly different methods, and sometimes their methods are so off that they end up badly projecting an election.

By the way, the weighting didn't seriously affect party identification. For example, 29.0 percent were Democrats in the raw data, 28.8 in the weighted data.

Also I can't see in the AJC story how the survey was conducted. Landlines? Cells? Smoke signal? Live callers, or robo calls? Given the firm they use, real pros, I'm assuming a quality survey.

Wednesday, August 3, 2016

Who is John Roberts?

There's a set of survey questions the ANES asks every four years. Respondents are provided the name and asked what office that person holds, in a multiple choice format. The office holders are vice president, prime minister of the U.K., speaker of the House, and chief justice of the U.S. Supreme Court. The names, obviously, change depending on who holds that job, and these items are often used to construct a political knowledge index. Sometimes they stand alone, sometimes they're combined with other questions, such as which party controls the House or Senate.

For fun, let's take the chief justice question. That's John Roberts, in the 2012 survey. Respondents were provided his name and asked to answer what "job or political office" he now holds. If they didn't know, they were probed with "Well, what's your best guess?"

Some of my favorite answers are below. Note: spelling is from them as they could type in their answers in the national survey. I've added a few comments in parentheses.
  • Don't know where did you come up with these people (by the way, this respondent didn't know any of the office holders)
  •  Fox News Correspondent (there is a guy by this name on Fox)
  •  Hair stylist (love this answer)
  • I haven't a clue, I've been more preocupied with looking for a job than the election.
  •  I have no idea.//Is he the Supreme Court Chief Justice?//The only judge I know is Judge Turnbull.
  •  Member of Obamas team (some Republicans agree with this after the Obamacare ruling)
  •  Oh I don't know.//Chairman of the Republican Comittee, if he's a black guy, if I remember, the name and face.
  • i think he is part of the school dept. 
  • I dk I dk whose whom sometines I don't care makes me feel seilly n n=knowing but dont care
  •  Who the hell is that chic....? (I have no explanation for this)
This question is somewhat controversial, or as controversial as questions get among survey academics. Seems that for years the answers were coded as incorrect even when actually correct. If I answer "supreme court justice" that's technically correct even though he's Chief Justice. If I say "judge on the supreme court" that's correct too, but was getting marked as incorrect. But someone this survey answered "Republican judge." How do I code that? Correct? Kinda sorta correct? I've not yet taken a stab at this coding.

Tuesday, August 2, 2016

Is Georgia In Play?

Two polls out suggest Georgia is "in play" in terms of the upcoming U.S. presidential election. One poll is by Landmark/Rosetta, which gets only a "B" in 538's pollster grading scheme, but SurveyUSA gets an "A" in the same scale. The "why" of these grades has to do with reliance on robo-calls, on landlines (bad, and HuffPo just announced it is no longer using such polls in its aggregate measure of opinion), accuracy in previous polls, and being forthcoming in poll methodology. The Landmark poll has it dead even between Hillary Clinton and Donald Trump (45-45), but the SurveyUSA poll has it Trump 46, Clinton 42, just within the 4 percentage point margin of error. I'd lean toward the latter poll and a nod toward Trump given Georgia's previous election tendencies and fundamentals.

So, is Georgia "in play?" Not in the same way Florida or Ohio or Pennsylvania are "in play," despite these polls. They will require Trump to spend a few more bucks here than Romney did in 2012 (Romney aired all of four ads, best I can tell, and Obama zero). Then again, as tightfisted as Trump has been, he'll probably visit Atlanta more than Romney did in 2012. In other words, Georgia is just close enough that Trump will need to play a little defense to keep Georgia red.

Other interesting stuff from the polls.

Landmark has it 2/3 of independents do not believe Trump has the temperament to be president. Those independent voters matter, and that's a damaging number (crosstabs here).  We see similar stuff in the SurveyUSA details (available here). Clinton has only double of Trump's Hispanic vote. I thought it'd be more, though keep in mind the numbers surveyed are relatively small, so these numbers are iffy.

Monday, August 1, 2016

Other (specify)

Often in surveys we ask such questions as my favorite -- what religion is Barack Obama? In the 2012 ANES survey of 5,914 U.S. adults, responses offered to candidates were generic: Protestant, Catholic, Jewish, Muslim, Mormon, and Not Religious. Respondents could volunteer Christian or Baptist or some other denomination, but some took a different journey. Today I was digging into that question a bit deeper, into the raw text responses of people who took the Other (specify) option, some 306 of them.

In other words, they can come up with their own answers.

Here we find such gems as Obama is "a hybrid of Protestant with Muslim teachings from childhood" and another said he's "Muslim trying to be Christian." Another listed Obama's religion as "Marxist" and another as "Mason" still another as "Freemason." You get the idea. Two respondents see something even more dark, with one saying Obama's religion is "Satan" (dunno how that works) and another more specific, as "Satanist."

Another said it "depends on who he's [Obama] talking to."

Heh heh.

And finally, "Who knows for sure what he is?" Which, when you get right down to it, is an honest response. All we really know is what religion someone says he or she is.

Overall, the most popular response among the categories listed in the first paragraph was, of course, Protestant, followed by Muslim (1 out of every 5 responses). Most of the Other (specify) options were various versions of Christian in some form or another.

Students and UGA

Buried in a lengthy survey about the climate at UGA are some questions student media, particularly The Red & Black, could mine for stories. Take page 228 of the report, for instance. The subhead is:

Students Who Have Seriously Considered Leaving UGA

There's rich stuff. Allow me to hit a few high points (or low points, if you prefer):
  • A quarter of all students (undergrad and grad) have considered leaving UGA, most of them in their first year as a student here.
  • Thirty-eight percent of African American undergrads considered leaving.
  • Forty-seven percent of tho undergrads with multiple disabilities considered leaving.
A few caveats. This was not a random survey, but rather a self-selected survey. See my methodological criticism of it here.

What are the top reasons students considered leaving? Here ya go:
  1. I didn't belong
  2. Campus climate was not welcoming
  3. Personal reasons (medical, mental, etc.)
  4. Financial reasons
  5. I didn't have a support group
There's a treasure trove of data here, starting points for stories about student life at UGA. The numbers alone are not a story. You need to find folks who illustrate the issues brought forth in the data. And you must include caveats about the survey itself. Indeed, the survey probably deserves a longish R&B piece, or a multi-part story with graphics and sidebars.