« Glenn Beck: Mexico Should Compensate US for Cost of Illegals | Main | Have we entered the twilight zone? »

Lies, damned lies, and statistics

A lot of the political arguing going on in the arguments here lately have taken a rather odd twist. more and more people (mainly detractors) have been citing opinion polls, surveys, and the like as "evidence" to bolster their arguments.

Let me clue everyone in to something I believe I've hinted at, but never actually said explicitly:

I don't give a rat's ass about such things.

I don't think I've ever cited, discussed, or praised a single survey in making my arguments and concocting my essays. And that's been a deliberate choice.

(This is where someone finds a single example I have forgotten.)

I have two reasons that have compelled me to this position.

1) Polls are notoriously unreliable. There are literally thousands of examples of polls that turn out to be completely wrong. Dewey did NOT defeat Truman. "Exit polls" at elections are almost never right. In politics, it's a cliche', but it's a cliche' because it's true: there is exactly one poll that matters, and that's the one done in the booths on election day. All the rest are irrelevant.

2) Polls do not reflect whether something is right or wrong, correct or incorrect, only what people THINK is so. I do not base my opinions on what anyone or everyone else thinks; they are uniquely my own, derived from my own thoughts, observations, beliefs, and prejudices. I do not present what I think is the consensus of the general populace, but what I think is true, correct, or right.

A few centuries ago, it was overwhelming public opinion that the earth was flat. Prior to World War II, many people thought that Hitler wasn't that bad, that he could be lived with -- or, in some horrid cases, was right.

As Opus the Penguin observed about 20 years ago, "if ten thousand people do a silly thing, it is still a silly thing."


Comments (41)

There's a quote attributed ... (Below threshold)
Bo:

There's a quote attributed to Will Rogers that is somewhat relevant to your point, Jay.

"The problem's not what people don't know, it's what people know that just ain't so."

Can we get a survey going o... (Below threshold)
mantis:

Can we get a survey going of how many people think Jay is correct about polls?

57% of Americans would disa... (Below threshold)
a4g:

57% of Americans would disagree, Jay.

I've noted that those who c... (Below threshold)
_Mike_:

I've noted that those who cite polls most often tend to fall into the collectivist camp (as indicated by their being an advocate of more statist/socialist policy); whereas, an individual would tend to value reason above the opinion of others.

Well, Mike, what do you exp... (Below threshold)
P. Bunyan:

Well, Mike, what do you expect? They don't have facts or morality on their side, so they have to resort to public opinion.

Good post Jay. I especially liked the "flat earth" example. That is soooo like the leftists today.

Didnt polls show that Gore ... (Below threshold)
Gianni:

Didnt polls show that Gore would win in 2000?


OOOps.

I don't believe in polls ei... (Below threshold)
Matt:

I don't believe in polls either except as an interesting window on what a block of carefully selected people's opinion is at that particuolar moment. It is interesting that media doesn't usually discuss the actual polling data, questions, demographics ect unless pushed very hard, of if the poll is counter to their opinion.

I think it was Mark Twain that said, "Figures don't lie, but liars can figure."

One of my favorite quotes o... (Below threshold)
Steve L.:

One of my favorite quotes of all time:

"Just because someone agrees with you doesn't make you right. It just means there's someone else out there as stupid as you."

Sadly, I have no idea who said it.

Polls certainly don't indic... (Below threshold)
Publicus:

Polls certainly don't indicate what's right or wrong. But, properly done, polls DO indicate what people are thinking. Polls DO show the war in Iraq is unpopular; that said, Jay is right: we shouldn't necessarily govern by polls.

I will also note that THIS website cites polls when they support the conservative viewpoint:

Rassmussen Reports: Giuliani's lead over Hillary Grows
http://wizbangblog.com/2007/02/23/rassmussen-reports-giulianis-lead-over-hillary-grows.php#comments
Poll: Americans Should Stand Behind President Bush During Wartime
http://wizbangblog.com/2007/02/20/poll-americans-should-stand-behind-president-bush-during-wartime.php#comments

In any poll, we should ask: were the questions asked somewhat neutrally, or did they "push" a particular answer? And we should certainly critically evaluate polls. But they are not without value.

I have heard the quote also... (Below threshold)
Burt:

I have heard the quote also attibuted to Mark Twain, But, the earliest reference to the "Lies, Damn Lies, and Statistics that I have found is actually Benjamin Disraeli (the only Jewish Prime Minister of Great Britain} A list of his quotes found in Google is really quite entertaining.

I have found that anyone conducting a poll already has their mind made up and are only trying to prove their point. Therefore they construct their questions to lead to what, they believe, is the correct answer. Further, I believe that anyone who actually answers the questions posed has entirely too much time on their hands. If I am approached by a pollster and I have some time to waste, I demand that they answer my questions first. Who is paying for this poll? What are they trying to prove? Then if I decide to participate, I will respond with "None of your business" to at least 90% of the questions.

There actually have been an... (Below threshold)

There actually have been an inordinate number of polls cited lately. And there seems to be a direct correlation between the number of polls cited and how much they are in harmony with the commenter's views.

Then if their views are represented by less than 40%, it's referred to positively as a "significant minority". If it's between 50% and 57%, it's referred to as a "majority" (which could mean anything, but sounds good). Anything beyond that the citer gets more shrill as it goes up and it becomes incontrovertable fact/right/truth and is hereby completely justified by sheer virtue of numbers.

I completely concur with Jay on this one. I've given my opinion at length about polls before here. I won't rehash it again.

And Winston Churchill's quo... (Below threshold)

And Winston Churchill's quotes are still my favorites. My most favorite being a conversation with [I don't know who the woman was]:

Woman: If you were my husband, I would poison your tea.

Churchill: If you were my wife, Madam, I would surely drink it.

I've been reading Tip O'Nei... (Below threshold)

I've been reading Tip O'Neill's biography recently and I was stunned to discover how long it took before American (majority) public opinion turned against the Vietnam war -- it wasn't until 1970!

wavemaker, read Hallin's <i... (Below threshold)
mantis:

wavemaker, read Hallin's The Uncensored War for a good look at why, as far as the media's role is concerned.

Besides, 77% of all statist... (Below threshold)
Dave A.:

Besides, 77% of all statistics are made up on the spot.

Historically, 74.7% of all ... (Below threshold)
bobdog:

Historically, 74.7% of all statistics are made up on the fly, but I put no stock in such claims.

There's a third reason, Jay Tea: Sometimes (if not frequently) the outcome of a poll reflects the outcome expected by the group paying the bill.

If a group having a strong belief in, say, global warming was looking to do a poll to advance their cause, they can preselect the outcome by finding a polling firm that agrees with them, or exclude a polling firm that shows contradictory results. I wonder how many political polls never see the light of day just for that reason alone.

Then throw in the statistical smoothing that adjusts for oversampled groups, say, Democrats, and select smoothing factors that get you closer to what you want to see.

Or if you want to skew the results, skew the sample. If you're looking to push gun control legislation, for example, do phone surveys during the day in affluent suburbs to weight the results with soccer moms who are heavily anti-gun. Or pick a city where violent crime is a hot topic like LA or New Orleans. Or Washington, DC. If I wanted to push legislation against the tobacco industry, I'd sample in Pasadena or Berkeley, or another similarly prissy area with a history of political correctness and get the results I wanted.

I don't think it's a blatantly widespread problem, but it seems to me that you can find a polling firm that will confirm almost any proposition you want to support with scientific-sounding babble.

I recall a presidential tracking poll in the last presidential election cycle by one of the major firms (Zogby or Rasmussen, I think) that proudly announced a big Democratic victory in the upcoming elections. Not only were they wrong, but their results were way out of line with every other major poll. It was so visibly out of whack that many news programs held their collective noses when they reported the survey results. Considering the shallowness of most network news, that's like a buzzard refusing a meal because it stinks too much.

One thing I've learned to watch is the confidence interval for survey results, particularly on local news programs. The confidence tells you how many people were contacted. If it's 5% or more, it's a tiny, and probably skewed, sample, and reliability is doubtful. Odds are, the sample population shared the same alleles.

With a confidence factor of 11%, more than 65.2% of the people I surveyed agreed with me, and also strongly disapproved of the conduct of the war in Iraq. Of course my survey was conducted at my local gas station at the hi-test pump while I was filling up in Hollywood, but who can question behavioral science or the truth of my findings?

Statistics don't lie, peopl... (Below threshold)
BarneyG2000:

Statistics don't lie, people do.

Jay, I hate to break it to ... (Below threshold)
Heralder:

Jay, I hate to break it to you, but I recently conducted a poll on your post about polls and got the following results:

Q: How do you view Jay's post at Wizbang! -?

Jay is wrong: 58%
Jay is not right: 71%
Jay thinks he's right but he's wrong: 94%
I like ice cream: 81%

I know my total percentage ends up being 246%...I'm trying to work out how that happened. I think it may have to do with my sampling methods or maybe my questions, or maybe my math.

Anyway, the margin for error is 3.5 percentage points....so it looks like:

a) Jay is wrong.
b) People like ice cream.

1) Polls are notoriously... (Below threshold)

1) Polls are notoriously unreliable.

The Real Clear Politics poll averages correctly predicted every one of the 2006 Senate races.

Didnt polls show that Gore would win in 2000?

Actually, they were correct.

Gore 50,999,897 48.38%
Bush 50,456,002 47.87%

The national polls didn't account for the electoral college which allows a candidate who receives a minority of popular votes to win.

The exit polls in Florida were also right. A majority of Florida voters intended to vote for Gore but didn't because of the butterfly ballot. Exit polls can't account for the fact that a bunch of old people accidentally voted for Buchanan when they intended to vote for Gore.

Heralder:Your numb... (Below threshold)
Hugh:

Heralder:

Your numbers about Jay aare clearly wrong, however the conclusions are not.

Now Jay, try thinking before responding. I know it's hard but......

Polls can be very accurate,... (Below threshold)
Diane:

Polls can be very accurate, but just like the news, it depends on if the question/story is "slanted" and if the procedures are handled correctly. Most polling questions are so generally stated, that the only benefit is to sway a certain group of people For example, "Do Want American Soldiers to come home from Iraq?" This question is conditional for many, but no way to give those conditions in a basic "yes" or "no" question. Answering who you want to or will vote for in an election is more "black/white".

The sad part is having our government officials rely on polls to make decisions. In some instances the (American) public is often not well informed enough to have a valid/educated opinion to be used as a source in making important judgements!


Larkin,What about th... (Below threshold)
SCSIwuzzy:

Larkin,
What about the absentee/overseas ballots? Who did they intend to vote for?

Hugh:Your... (Below threshold)
Heralder:

Hugh:

Your numbers about Jay aare clearly wrong, however the conclusions are not.

Does that mean I can claim "Fake But Accurate"?

Hugh:<br... (Below threshold)
_Mike_:

Hugh:


Now Jay, try thinking before responding. I know it's hard but......

The 'chickenhawk' argument applies to this statement.

Larkin, you've proven yours... (Below threshold)

Larkin, you've proven yourself wrong. The polls might have said that Gore would win the popular vote, but that's utterly irrelevant. I once did some math, and I figured out that it was possible for a candidate to get elected with 12 votes, while the other one gets millions and millions. It's astronomically unlikely, but it is theoretically possible.

To tout the "popular vote" as if it matters more than a fart in a hurricane is silly, at best.

J.

Heralder -Re: you... (Below threshold)
Upset Old Guy:

Heralder -

Re: your poll results.

I suspect you forgot to normalize your responses to reflect the demographics of your polling group.

Oyster -

How about this one: "If your young and you're not liberal, you have no heart. If your an adult and you're not conservative, you have no brain."

That was another Churchill-ism

Oyster ~ the woman in your ... (Below threshold)

Oyster ~ the woman in your Churchill joke was Lady Astor, with whom Churchill was known to trade barbs at every turn. Another famous encounter between them came at a dinner party when she observed Churchill was imbibing rather liberally, as usual:


Lady Astor: Why, Sir Winston, you are drunk!

Churchill: Yes, Madam, and you are ugly. But in the morning, I shall be sober . . .


It is also said that when she hosted a costume party, Churchill rang her up for suggestions on an appropriate disguise, and she replied, "Pretend to be sober - no one will recognize you!"


When Churchill described the Munich Agreement as a "total and unmitigated defeat" in the House of Commons in 1938, Lady Astor (who had taken her husband's constituency when he ascended to the House of Lords upon his father's death) shouted out, "Nonsense!"

In another Commons speech about the history and future of mankind, every time Churchill mentioned the word, "man," Lady Astor shouted out, "And woman, sir!" After several of these interruptions, he addressed it, saying, "In this context, it is understood that 'Man' EMBRACES 'Woman,' Mr. Speaker." Lady Astor was not amused.

George Bernard Shaw on Lady Astor's many social campaigns: "She has no political philosophy and dashes at any piece of kindly social work that presents itself, whether it is an installment of socialism or a relic of feudalism..."

MP Harold Nicolson said of debating her in the House of Commons, "It was like playing squash with a dish of scrambled eggs."

Not a bad looking "dish," though: Lady Astor in 1909

Surely, by now, you are sorry you asked . . .

;-)

There are literally thou... (Below threshold)
Brian:

There are literally thousands of examples of polls that turn out to be completely wrong. Dewey did NOT defeat Truman. "Exit polls" at elections are almost never right.

You're equating two entirely different things. Yes, almost all polls are completely wrong, in that the final numbers rarely match the predicted numbers. I'd be shocked to find any poll that was 100% dead-on accurate.

But outside of a vote, that's not the issue. A poll that said that Gore would beat Bush 51-49 was 100% wrong in its outcome, but pretty damn close in its numbers. In a vote, any vote, the outcome matters. But when you have a poll that says 67% of the country is against this or that, it doesn't matter that the actual number isn't exactly 67%. The point is that it's around 67%, within its official margin of error, and perhaps even a bit beyond. But it's not going to wind up being 27%, or even 47%, no matter how much one believes that "polls are inaccurate".

Right on, Jay Tea!! Adding ... (Below threshold)
Wil:

Right on, Jay Tea!! Adding Wizbang to my blogroll.

well, polls may not be comp... (Below threshold)
slingshot:

well, polls may not be completley accurate and reliable, but in their absence, how do you propose to get the general opinion of millions of people in an entire country? apparently jay thinks the way to do it is just think about what he thinks it should be, and then conclcude on that basis. i'll take polls, mainly for the reasons explained by brian at 6:03.

Brian, this could become a ... (Below threshold)

Brian, this could become a great discussion about Burke's theories on representation. Instead, I'll simply summarize it:

We had a president who lived and died by the poll in Bill Clinton. We have a president who seems to believe that he was elected not to parrot the mood of the moment, but to stand on his own principles and beliefs. I prefer the latter.

J.

I'll tell you how, Slingsho... (Below threshold)
Ryan:

I'll tell you how, Slingshot. You let them vote and not try to shape their opinion through push polling, designed to elicit specific responses through carefully worded questions.

To elucidate: There have b... (Below threshold)
Ryan:

To elucidate: There have been test polls done where they ask the SAME question and change ONE word to a different synonym and get opposite results. Polls are so easy to manipulate that in general they are, while not meaningless, less than useful for actually judging people's opinions> Furthermore, they usually neglect the one thing I want to see: "Do you the respondent actually know anything substantial about the matter at hand?"

i get the point that polls ... (Below threshold)
slingshot:

i get the point that polls can be manipulated, and i agree. but, when multiple, different polls taken by different polling organizations come out with the same conclusions, or close, i think that means something. the fact that people are uninformed or simply dumb does not mean that when that is reflected in the polling, the polling is wrong.

That's true sling -- and wh... (Below threshold)

That's true sling -- and what they frequently reveal is how useless uninformed opinions are.

Actually, the 2000 polls pr... (Below threshold)
MikeSC:

Actually, the 2000 polls predicted a scenario where Gore wins the election and Bush wins the popular vote. Which is why liberals right before the 2000 election kept mentioning the Electoral College on TV.
-=Mike

I was thinking of rewriting... (Below threshold)
epador:

I was thinking of rewriting my Pollish jokes, but since JT refuses to Poll Dance, I think we should just all be thankful and move on.

epador, puns like that just... (Below threshold)

epador, puns like that just might get you poll-axed out of here. They (poll) tax my patience that much.

J.

When most of the polling or... (Below threshold)
Ryan:

When most of the polling organizations share a similar bias it proves little. And when most of the news only publishes the parts of poills that says what they want. . .and almsot never releases the actual questions. . then they are meaningless. A long, well constructed poll, given over a long period of time vcan be meaningful> These crap short period snap polls are worthless.

We had a president who l... (Below threshold)
Brian:

We had a president who lived and died by the poll in Bill Clinton. We have a president who seems to believe that he was elected not to parrot the mood of the moment, but to stand on his own principles and beliefs. I prefer the latter.

Hmm, interesting. I think most people prefer a President who turns out to be right a majority of the time. I guess that quality apparently has no place in your wholly subjective assessment.

History is filled with visionaries and mavericks, as is the corporate world. But standing on your own principles and beliefs doesn't in and of itself automatically make you a visionary, nor does it make you right. There are a lot of wrong, dead, and broke people who stood on their principles and beliefs, and kept standing on them even as they spiraled down and took others with them.

Does a measure of the opini... (Below threshold)
Ryan:

Does a measure of the opinion of a group of people the majority of which do NOT know the facts of the case prove what 'right' is?




Advertisements









rightads.gif

beltwaybloggers.gif

insiderslogo.jpg

mba_blue.gif

Follow Wizbang

Follow Wizbang on FacebookFollow Wizbang on TwitterSubscribe to Wizbang feedWizbang Mobile

Contact

Send e-mail tips to us:

[email protected]

Fresh Links

Credits

Section Editor: Maggie Whitton

Editors: Jay Tea, Lorie Byrd, Kim Priestap, DJ Drummond, Michael Laprarie, Baron Von Ottomatic, Shawn Mallow, Rick, Dan Karipides, Michael Avitablile, Charlie Quidnunc, Steve Schippert

Emeritus: Paul, Mary Katherine Ham, Jim Addison, Alexander K. McClure, Cassy Fiano, Bill Jempty, John Stansbury, Rob Port

In Memorium: HughS

All original content copyright © 2003-2010 by Wizbang®, LLC. All rights reserved. Wizbang® is a registered service mark.

Powered by Movable Type Pro 4.361

Hosting by ServInt

Ratings on this site are powered by the Ajax Ratings Pro plugin for Movable Type.

Search on this site is powered by the FastSearch plugin for Movable Type.

Blogrolls on this site are powered by the MT-Blogroll.

Temporary site design is based on Cutline and Cutline for MT. Graphics by Apothegm Designs.

Author Login



Terms Of Service

DCMA Compliance Notice

Privacy Policy