Polls, Anecdotes and the People Who Misuse Them

Member Group : Jerry Shenk

Pollsters, media and politicians frequently generalize tiny numbers in order to make money, avoid work and/or manipulate public opinion. Their generalizations are often faulty, even false.

In a nation of 330 million, only 0.0001 percent equals 33,000 people. In the context of 330 million nonhuman units, though, 33,000 is insignificant. In national reports, federal agencies rarely list 0.0001 percent of anything as “data.” In most government reports, 0.0001 percent won’t even appear as footnotes.

Nevertheless, polling firms are paid to generalize surveys of only 1000-2500 – sometimes fewer – respondents to report “national” sentiments about political and cultural matters. Media can then cherry-pick polls as lazy substitutes for rigorous coverage of public attitudes and events, to reinforce media biases, and to influence their audiences. Accordingly, nearly-monolithic national media-reported polling “results” can be misleading – or dead wrong. Skeptical? Ask President Hillary Clinton.

Before and through the final days of the 2016 presidential election campaign, national media adopted the narrative that Donald Trump had no chance of winning the presidency, a conclusion based on the (near)consensus of national polls predicting Hillary’s comfortable lead through Election Day.

The problem was – and is – that national polls are merely minimal accumulations of anecdotal evidence offered by paid pollsters as genuine data rather than as the relatively few data points they actually represent. More and more people have eliminated landlines, and nearly everyone screens calls, so thousands of calls must be placed to get 1000 willing respondents. Margins of error are generally large. In fact, polling has become almost-prohibitively expensive, so very few faithfully-statistical polls are conducted anymore. Reliable “overnight” polling is virtually impossible.

Furthermore, in a “national” poll of 1000 “likely voters,” very few respondents must lie about their preferences and/or likelihood of voting to significantly skew poll results. 2016 General Election results – that is, hard data – provide evidence that thousands of  “shy Trump voters” and “unenthusiastic Hillary voters” in swing states lied to pollsters. (Yeah, I know…popular vote, blah, blah, blah… Pollsters, meet the Electoral College.)

Politicians’ intentional misuses of anecdotes are usually even more misleading. Politicians misrepresent sympathetic, heart-rending anecdotes as “data” to support their policy proposals. In fact, among ambitious, self-interested politicians, faulty/false generalizations are a stock in trade.

The first Democratic “presidential” debates were full of them. For example, generalizations were used to lament “joblessness” and people “working two jobs,” even though, according to Bureau of Labor Statistics data, American, including minority, jobless rates are at historic lows, and the percentage of people working two jobs has changed little over four decades.

Emboldened by President Trump’s widely-reported polling “negatives,” Democrats ignored positive polling. According to Gallup, 88 percent of Americans think the economy is “fair,” “good,” or “excellent,” and 72 percent believe that now’s the time to find a good job – both numbers substantially higher than four years ago under a Democratic president.

Instead, 2020 Democratic White House wannabes amusingly vowed to “fix” the best economy we’ve had since before the last time a Democrat won the presidency, maybe ever. Ironic, huh?

https://www.ldnews.com/story/opinion/2019/07/08/polls-anecdotes-and-people-who-misuse-them/1671385001/