Nate Silver's Statistical Analysis - Sends IRS Critiques To "The Fainting Couch"

[kpete]

In the table below, I’ve estimated the number of taxpayers in each income group who were audited in 2012, as derived from statistics in the I.R.S.’s 2012 Data Book. It is also possible to estimate how many Mitt Romney and Barack Obama voters would have been audited last year. The calculation assumes that an individual’s chance of being audited was related to their income, but not to their political views.

I estimate the number of voters in each income bracket from the 2012 Current Population Survey. I then estimate the share of the vote in each income bracket that went to Mr. Romney and Mr. Obama based on last year’s national exit poll. (Note that the income brackets used in the exit poll and the Current Population Survey do not exactly match the income brackets listed in the I.R.S.’s audit data, so I use the closest available approximations.)

This results in an estimate that about 380,000 of Mr. Romney’s voters were audited last year, as were about 480,000 of Mr. Obama’s voters.

It’s almost like the IRS audits low-income Americans far more than the wealthy because there are far more low-income Americans to audit, and they are much more likely to receive a refund from the government, and that these low-income Americans are primarily Obama voters. It’s also almost like Peggy Noonan found a couple people who had been audited and concluded, without any evidence whatsoever, that they had been deliberately targeted because of their political views, and that she then dismissed every other possible explanation for the audits.


And then Nate Silver came along and said “Wow, this woman is insane, TO THE NATE CAVE” and then squished her. Then he probably had a tasty beverage.

It tasted like victory and Noonan tears.

http://fivethirtyeight.blogs.nytimes.com/2013/05/17/new-audit-allegations-show-flawed-statistical-thinking/?smid=tw-fivethirtyeight&seid=auto
http://www.balloon-juice.com/2013/05/19/the-plural-of-noonan-is-not-data/#comments

[Baitball Blogger]

1. Me likes.

[Auntie Bush]

2. Thanks Nate for that graphic...splains a lot.

[ZRT2209]

3. Noonan is an idiot.

[bullwinkle428]

4. "STATISTICS HAVE A LIBRUL BIAS!!1!1!"

We should start a pool to see which wing-nut is the first to make this claim! My money is on Gohmert.

[ProSense]

5. CNN Confirms Gallup's Poll: The Obama Bounce Continues

CNN Confirms Gallup's Poll: The Obama Bounce Continues
http://www.dailykos.com/story/2013/05/19/1210181/-CNN-Confirms-Gallup-s-Poll-The-Obama-Bounce-Continues

Charles Blow, NYTimes: Americans Yawning at GOP Phony Scandals
http://www.democraticunderground.com/10022863252

[okaawhatever]

6. One thing that wasn't taken into account in both instances is the presence of an outlier or anomaly

that would trigger an audit. Statistical anomalies trigger audits. Someone with super high medical bills one year, someone with out of the norm expenses for business. Even with businesses the expenses are broken down by category. One example Noonan gave was a man who had donated a million dollars to a campaign. It's conceivable that that person used savings or investments to pay for that so their tax returns would show donations exceeding income. That would probably trigger an audit. What's important is if the IRS has a computer program that selects audits based on statistical analysis and if so whether the computer selected these returns or individuals did. Shame on people like Noonan for helping drive this distrust and paranoia.