Tuesday, October 6, 2015

Poll Support x Twitter Followers

In the GOP race we all know Trump is ahead (though perhaps slipping), so I decided to examine all the candidates by their national poll numbers and number of Twitter followers. First off, there is a high correlation between the two (r = .80), in large part thanks to Trump's lead in both polls and Twitter followers. Still, remove Trump and the correlation is still .71, so we got something here.

See the scatterplot below. Yeah, that's Trump all alone in the top right corner. If we drew a regression line here, it'd be running up from 0,0 to Trumpland, a nice linear fit. As I tell my students over and over, however, correlation is not causality. You cannot argue lots of Trump followers equals poll popularity, nor the other way around.

Raw data is below.

Poll Twitter
Bush 8.3 326,000
Carson 17.3 705,000
Christie 2.6 55,600
Cruz 6.1 519,000
Fiorina 11 581,000
Graham 0.3 27,800
Huckabee 2.8 411,000
Jindal 0.5 20,900
Kasich 3.1 111,000
Pataki 0.3 54,600
Paul 2.4 693,000
Rubio 9.5 888,000
Santorum 0.4 245,000
Trump 22.8 4,036,000

Millage Rates

I teach journalism, but I'm also a bit of a nerd when it comes to statistics, math, and especially public opinion polls. Today let's look at an "explainer" produced by our Newsource students down stairs. You can watch it here for context. Go ahead. It's fast. Then come back. Oh, and I love the clever bit about what's a millage rate? It's not Milledge Avenue. Nice job on that.

OK, done? Good.

At the 1:27 point we get to my point. Here's a screenshot:

So what's the problem? Not the math. The issue is the "If your assessed property value is $100,000 ..."

That's not quite how it works. Here's a nice, simple description, but I'll save you the visit off site and summarize it for you.

1. You have the assessed value of your home. In our example above, $100,000.
2. You then take 40 percent of that value.
3. And then you remove various exemptions, like homestead exemption. Let's say $10,000
4. This gives you the taxable value.

So in the example above, a $100,000 assessed value would become a tax on $30,000. So instead of a $1,913.10 tax bill, you'd see more like a $765.24.

How'd I catch this? I own a home in Clarke County -- with higher taxes than Jackson -- assessed at more than $100k, and I know my tax bill is below $1,913.

So there's commercial value, assessed value, and taxable value.

Wednesday, September 30, 2015

Rate This Rating

As The Red & Black reported today, UGA (where I teach) finished in the Rate My Professor Top 10 universities. Full list here. We're #8! We're #8! So very exciting -- except that we were #3 last year, so we've dropped.

So, let's rate that rating. In other words, what's their methodology? Here's their methodology page, and from it we can tell a couple of things. There are a lot of ratings, millions of them, and nothing about this is random or systematic. The final rating is a combination of individual faculty ratings and campus ratings, which kinda makes sense, but beyond that there's no rigor in the data collection. It's big data, but necessarily good big data.

All that said, go UGA. We're #8.

Interestingly, in the Top Professor's list, two are from Georgia but neither is from UGA. I suppose we make up for it with lots and lots of well rated, but not top rated, professors.

In full transparency, my rating on a 5-point scale is 4.3.

My thoughts about stuff like this? Harmless fun. Don't take the numbers too seriously, given the lack of rigor, but if I were a student I'd certainly look at them for guidance. Keep in mind that often the students who post are those pissed off about their grades. And the N is small. Mine has 10 ratings based on posts from 2005 to 2015. Not exactly a good data source and, in fact, fairly meaningless.

Running the Name


This is no surprise. I fully expected either (1) UGA's athletics machine would lean on the young lady or (2) he'd apologize and all would be well or (3) it was all, of course, a misunderstanding, and anyone can confuse a death threat, or (4) she felt the pressure and just surrendered to the inevitable.

This makes even less excusable the use of the student's name in the story. See my post below.

Original Post

A 20-year-old UGA student accused a football player, Isaiah McKenzie, of making terroristic threats. According to reports, he said "he was going to call some friends and they were going to come out and he was going to kill her." Yes, that's a felony. It's also a felony to eat at Chili's, where this allegedly happened. Don't we feed our players?

Stories by ESPN's David Ching and AJC's (DawgNation)'s Chip Towers identified the woman by name. They were wrong to do so.

Thankfully, The Red & Black and The Athens Banner-Herald did not name her. They got it right.

She's filed a complaint. Until McKenzie is arrested, there is no compelling reason to name her. Indeed, look hard at the SPJ Code of Ethics, especially the part about "minimizing harm." College football fans are nuts, and that's putting it nicely. Just look at some of the Twitter and Facebook comments about this young woman. Sickening. By putting her name out there, Ching and Towers, you've put her at risk. Congratulations.

Let's look at some of the arguments for naming her.
  • It's a public record. Yes, it is. So is the name of a rape victim, but we (the royal journalistic we) make an ethical choice to not name the victim. Just because it is a public record, that doesn't mean it has to be published or aired.
  • It's unfair to McKenzie to name him and not her. Sorry, he has the whole University mouthpiece industry to back him. She has herself. This is a false argument. Again, if he gets arrested, that changes matters. It's just a complaint.
  • It's all a plot by Bama, or the cops who hate football, or someone. Yes, conspiracy theorists, unite now. Jeez.
Quite simply, you do not name her. Not yet. Hell, they even named her father, the head of a government agency in Atlanta. Really? It's not like he's the friggin mayor or something. I expect this kind of crap from ESPN, defender of all that is athletic, but I expect more from the AJC.

Sports reporters, just don't do the cop beat. You'll hurt someone.

Oh, by the way, I tweeted this last night. Never got a response or justification from either. Sigh.

Tuesday, September 29, 2015

The Ethics of Deleting Tweets

I don't often write about journalism ethics here, mainly because I have a certain moral flexibility when it comes to getting the story. A rather harmless Twitter mistake fluttered across my desk just a few moments ago and it got me wondering about the ethics of deleting tweets. So here's the tweet from our student newsroom:

A harmless misspelling of Heisman. No biggie. Three minutes later I responded:
And a few minutes later the original tweet was deleted and a corrected version posted. Without a thank you, I should note, so me being me I replied via Twitter: "Better. You're Welcome." That was a little while ago. So far, no thank you.

Anyway, it raises the question (to me, perhaps not to you) whether you should delete the original offending tweet. Keep in mind this is about journalism ethics, not PR, in which obviously you'd delete a mistake or something embarrassing, especially to the boss or client.

On more serious stories, this piece does a good job arguing that, to me, you should correct but never delete the offending post. It's all about transparency. But does this apply to dumb typos? Especially typos that make it seem like Nick Chubb made the mistake when, we all know, it was a harmless mistype by a social media news editor. My own feeling is that it's OK to delete a tweet like above but it's not OK to delete a more serious mistake, as in saying someone is dead when they're not. Again, transparency. At some point we'd like to see a Twitter correction function or even a flag that says this tweet corrects an earlier one.

Or, perhaps, you can just put CORRECTION in the tweet, especially when there's still plenty of space to burn.

And thank your audience when it corrects you.

Yes, you're welcome. Again.

Monday, September 28, 2015

SEC Power Rankings

This week's SEC power rankings are out. I'm just gonna report the top three for several sites or sources. Three are national, one is the UGA newspaper, which I'll let pass without comment because I don't want to spawn any more "mommy he's picking on me" editorials. I find the SB Nation one interesting, with UGA 4th, otherwise no real surprises.

SEC Country
SB Nation
Ole Miss
Ole Miss
Ole Miss
Ole Miss


(UGA 4th)

One based on 13 voters for all NCAA football is somewhat different. The highest SEC school is, of course, Ole Miss (#2 nationally, no real surprise), followed by LSU (#6), UGA (#8), Bama (#11), and Texas A&M (#12). The highest any ESPN individual voter has UGA is #6 (you can check out each individual voter's list). Skimming them, UGA's QB hasn't convinced anyone yet and the Bama game will say a lot about whether UGA is for real and moves into the Top 5.

As a Bulldog fan who remembers the last Bama blackout game, please God let no one suggest such a stupid idea again.

Friday, September 25, 2015

Predicting Political Knowledge

Scholars have long tried to understand what factors predict political knowledge. Education is the standby, the consistent predictor -- with the more educated respondents doing better on political knowledge tests than those with less education. A few other factors come and go, depending on the analysis strategy and how you model the data. Media use, for example, sometimes is a factor and sometimes has no role to play.

This study asks a somewhat different question -- do personality traits better explain this than does education? The answer is, well, read below:
Openness to experience and intelligence are found to be positive predictors of political knowledge and Neuroticism a negative predictor of political knowledge. In both studies education remains the single strongest predictor of political knowledge. Furthermore education can, to a large extent, even out the differences in political knowledge between those with high and low cognitive abilities
So education continues to boss around all the other variables, even interesting ones like personality traits such as openness to experience.