On October 25 I was giving my Google News Feed a quick check when I stumbled upon the a story at the top of the US section. Business Insider reported “A poll found respect for police is nearing a record high nationwide.” The article was headed by a picture of a policeman casting a shadow over a huge American flag. It reported that 76% of Americans had “great respect” for Police, a number which is up by 12% from when it was taken a year ago and at the greatest high since 1967.
This seemed out of touch with the actual experience of actual people who have to deal with actual police. As the Mockingbird reported in August the rate of police killing of civilians has increased since Ferguson. This summer's surge of police murders and protests against them would also seem to be at odds with this poll and would cause any reasonable editor to check such a poll before assigning a reporter to wrap a flag and a speech around it. As it turns out Business Insider does not have reasonable editors.
The poll itself was conducted by Gallup and the very first batch of numbers set off an alarm bell. The findings claim that 80% of white Americans and 67% percent of “non-white” Americans have “great respect” for police. 80+67 = 147. 147 divided by 2 = 73.5%. Since every respondent had to be either white or non-white one wonders where the additional 2.5% came from. Neither Business Insider, nor www.Mic.com, which reported the poll originally, explained or even noted this. Gallup did claim on their website that the margin of error was +/- 4%. Gallup took what numbers they had and made an adjustment upward rather than downward high enough to make the numbers look good but low enough to make it newsworthy as a “near record high.”
This was the first way that Gallup skewed their results for propaganda purposes. The poll had a confidence level of 95%. A 95% confidence level does not mean the pollsters are 95% confident that the respondents are telling the truth. A 95% confidence level is a range of values that you can be 95% certain contains the true mean of the population. This means that answers outside of the range of values can be discarded before computations are made. This is a huge factor with this particular poll.
The actual question was the sixteenth asked in a long survey. The other questions in the call were not noted in the report. The question read as follows “Turning to the police, How much respect do you have for the police in your area – a great deal, some, or hardly any?” This is not rating respect for police on a scale of one to ten or even one to five. This is rating respect for police on a scale from zero to two. With a 95% confidence level and a mean of 1 this allows the surveyor to discard as many
“hardly any” answers as they need to in order to produce the desired number.
Not that many answers needed to be discarded because of the way the sample set was constructed. 1017 people across all 50 states and DC were polled. Race, gender, income and population density were factored into who was polled. 60% of the phones called were cellphones and 40% were landlines. When more than one person is present the person in the home with the next birthday was the respondent if they were available.
This method instantly adjusts towards a wealthier demographic. Retired Americans with comparatively greater wealth might have a somewhat lower income than people still in the work force. Although 41% of Americans still have landlines in their home, that percentage is skewed towards homeowners, which the less wealthy majority of Americans are not. Combine this with an ability to drop people outside of the 95% confidence level and you can move those percentage points a few more steps in a conservative direction.
Further, in an age of data mining, police murder and NSA surveillance, imagine what a person might think if they were randomly called at home and asked how much they respect the police. Would they be afraid to say they do not? The phone call is recorded for quality assurance and records are kept. Police retaliation is now a reasonable fear. This is not accounted for in the Gallup Poll. This potential bias would have a great effect on a poll with a sample size around one thousand. A mere ten people frightened enough to say “some respect” rather than “hardly any” move the survey results by a point.
Business Insider reprinted the survey results because they serve an agenda. That agenda is stability for a certain class of people in the face of mounting public anger at the increase violence of police on the civilian population. As the saying popularized by Mark Twain goes “There are three kinds of lies: lies, damned lies, and statistics."