Steering the ElephantHaving lived in Louisiana for 7 years and appreciating its history, social vibrancy, culture, food, and music, I also came to appreciate its quirkier sides. I had a neighbor who thought that if a quart of beer could get cold after one day of refrigeration, it would be really cold after a week. I knew other people who would point up over their heads when answering the question, “Where is north?” There was a women living down the street who had scores of plastic rose buds blooming from her Camellia bushes all year round. And there are people there who do some very interesting things with chicken claws and other bird parts to keep the world in balance. So if someone were to ask me whether I thought it likely that a certain portion of the population of Louisiana, say less than 1.2%, thought the sun rose in the west rather than the east, I’d answer, ”possible… but not probable.”

Public Policy Polling released favorability data on August 21, 2013 for the State of Louisiana. The results from the poll included favorability ratings to questions about Presidential and State Governor performance levels, same sex marriage, gun sales background checks, Mother Theresa, Abraham Lincoln, Jesus Christ, prominent regional football coaches, and who was responsible for the lack of Federal response to Hurricane Katrina – George Bush or Barack Obama.

Guess which one was talked about in the News media?

Here are a few of the headlines:

More Louisiana Republicans Blame President Obama for Hurricane Katrina Response Than Bush

A Third Of Louisiana Republicans Blame Obama For Hurricane Katrina Response Under Bush

Forgetting the relevant Katrina detail, eight years later

Dana Milbank: Embracing misinformation on Obama

Every one of these, and many other articles, reported that 29% of Louisiana Republicans believed that Barack Obama was responsible for the lack of Federal response to Katrina. They went on to analyze why so many Republicans would believe this, or, gave reasons why Republicans would answer this way. Nowhere was there an analysis of the poll. Nowhere, in any of these articles, was there a reaction of ‘possible… but not probable.’

When I heard the report that 29% of Louisiana Republicans blamed Barack Obama for the lack of Federal response to Katrina (2005), my first reaction was, “Obama was not President then”, which lead to my second reaction, “Show me the data!” What were the polling questions, who was polled, how were they polled, how many were polled, and what was the margin of error?

Through an automated phone survey, (you know, the ones that come up on Caller ID as “Polling” or “Unavailable” – the one phone call we all have been waiting for and want to answer), 721 people were asked to respond to 26 questions. Within the 721 respondents, 274 identified themselves as Republicans and were asked 5 more questions.

From the 274 self-identified Republicans with the special extra 5 questions, 80 respondents (29%) answered question 2, stating that Barack Obama was responsible for the poor Federal response to Hurricane Katrina. The error rate for this question is +/- 5.9%.

Let’s put this into some context:

80 people out of a state population of 4.6 million people — 0.00174%
80 people out of 2.9 million people registered to vote — 0.00276%
80 people out of 789,000 registered Republicans — 0.01014%
80 people out of 186,000 Republicans likely to vote — 0.04301%
80 people who answered a robo-call and did not hang up!
80 people — using the survey sample and backward projecting to the population of likely voting Republicans — would represent 54,059 (+/- 5.9%) people
54,059 people as a proportion of the state population – 1.18%

Is it likely that 29% of Louisiana Republicans think that Barack Obama is responsible for the lack of Federal response to Hurricane Katrina?

Possible… but not probable.

Is it likely that the media, with all of its analytical resources, will publish and comment on scintillating statistics without questioning the statistic?

They did it, in this case. Likely, possible, or probable do not figure into it.

What about us? How likely are we to use a statistic without questioning its validity?