Friday, December 17, 2010

Misleading Study Finds People are Misled

Apparently a new "study" from World Opinion.org came out the other day supposedly finding that Americans are highly misinformed by cable TV news, and Fox News viewers, most of all. Maybe that's all true, however... Based on the study itself, who'd know?

Like most public opinion data gathering, this survey was so poorly done and so dependent on (what I must assume are) the views of the authors of the questions that it's completely meaningless and ironically, the whole thing is probably more misleading than a lot of the ideas they are discussing.

I took a statistics class once with a professor who literally addressed the class saying "wah wah wah", like Charlie Brown's teacher. And I do mean, "literally".
I suspect this thing will be burning up teh intertubes, thanks to the magic of confirmation bias, but I really want to take a second and point out some of the intense failure that happened within the questions that were asked of participants.

They claim that various percentages of people incorrectly believe that a whole list of "obviously false" statements are in fact true. But if you actually go through each of these statements, the accuracy is extremely suspect. Few of their points have easily determinable "binary" answers, so the very idea of "true" vs. "false" is kind of silly in a lot of cases. More importantly though, in any case that is open to interpretation, the authors foolishly defer to the government-backed, statist position.

On economic issues, as many of these are, that is a recipe for failure. Check it out.

First, here are the few points excerpted by Mark Howard's article, "Study Confirms Fox News Makes You Stupid" (what was I saying about confirmation bias again?) that I'm not really going to argue with:
  • 56 percent believe Obama initiated the GM/Chrysler bailout
  • 38 percent believe that most Republicans opposed TARP
  • 63 percent believe Obama was not born in the U.S. (or that it is unclear)
  • 60 percent believe climate change is not occurring
  • 38 percent believe that most Republicans opposed TARP
These points are basically binary, so true or false more or less applies. I'll go ahead and be generous and give them a pass on things like Obama and the GM Bailouts (no, he didn't initiate them, but he certainly owned them, didn't he?), and I'll avoid getting into too much of a discussion of climate change. Though I will note that climate change is a broad idea that has come to package a bunch of bad policy and bad conclusions with highly complicated and occasionally dubious science... So I don't always blame people for answering "no" to questions like "is climate change occurring", if they believe that a yes response would be taken to mean that they also support the whole package of idiocy.

I do feel compelled to point out, however, that while Mark Howard is using the study to specifically mock Fox News viewers and "Republicans" in general, he seems to have entirely missed this whole passage:
On the other hand those who voted Democratic were more likely to incorrectly believe that: it was proven to be true that the US Chamber of Commerce was spending large amounts of foreign money to support Republican candidates (voted Democratic 57%, voted Republican 9%); Obama has not increased the level of troops in Afghanistan (51% to 39%); and Democratic legislators did not mostly vote in favor of TARP (56% to 14%).
So... Just to be clear, 38% of Republicans believed that Republicans opposed TARP, but 56% of Democrats believed Democrats opposed TARP.

Both are wrong, but as you can see, Team D was substantially more wrong. In fact, a majority of Democrats were wrong about that issue, as compared with the minority of Team R. That's an odd miss for someone writing an article about how Fox News Republicans don't know anything... Of course it really all comes down to confirmation bias. Republicans want to believe that Obama is responsible for stupid stuff that Bush did, like initiating bailouts, and Democrats want to believe that Obama is not the war-mongering jerk expanding troop levels in the Middle East that more principled people are still paying attention to.

Strange... Or not.

The study points out that viewers of the other cable networks have big problems too... Of course, I doubt we'll hear much more about that from any bloggers, journalists or pundits that pick up this story.

But there's a lot more here that's not only "debatable", but in, my opinion, just plain wrong about the so-called study. For instance - taken from the World Opinion site:
  • Though the Congressional Budget Office (CBO) concluded that the stimulus legislation has saved or created 2.0-5.2 million jobs, only 8% of voters thought most economists who had studied it concluded that the stimulus legislation had created or saved several million jobs. Most (68%) believed that economists estimate that it only created or saved a few jobs and 20% even believed that it resulted in job losses.
  • Though the CBO concluded that the health reform law would reduce the budget deficit, 53% of voters thought most economists have concluded that health reform will increase the deficit.
  • Though the Department of Commerce says that the US economy began to recover from recession in the third quarter of 2009 and has continued to grow since then, only 44% of voters thought the economy is starting to recover, while 55% thought the economy is still getting worse.
Now let's take a second and step back here...

The "proof" offered for each point comes from the government itself - and it's hardly even really proof to begin with, as measurements like "jobs saved or created" DON'T EXIST!!

The authors of the study, and idiots like Howard jumping on the bandwagon, don't seem to grasp that they're looking for proof of something that is fundamentally non-existent and intrinsically flawed. They also don't seem to grasp how the CBO actually works.

The Congressional Budget Office releases reports that virtually always come with massive disclaimers because their mandate basically forces them to analyze the economic impact of political actions based solely on the assumptions provided by the politicians themselves. They are rather inhibited in general and wind up making frequently flawed predictions because they are required to base their projections on static data, when - much like Schroedinger's Cat - manipulations of the economy by the government change the incentives people rely on to make economic decisions.

And by the way, I'm not just making this up. The CBO is staffed by some reasonably smart people after all! So they're usually pretty clear about this - for instance, from their Director's Blog last year:
CBO’s forecasting record provides a measure of the uncertainty underlying forecasts under normal circumstances. However, the current degree of economic dislocation exceeds that of any previous period in the past half-century, so the uncertainty inherent in current forecasts exceeds the historical average.
Note what he's saying here (in as diplomatic terms as I think someone in his position could possibly use) is that their predictions are filled with uncertainty in the best of times, and in the worst like now - they're positively unreliable. I mean, geesh... Let's not forget that back in 1965 when Medicare started, it was estimated (by the CBO) to cost $12 Billion by 1990.

How much did we spend on Medicare in 1990? $110 Billion!

The CBO was not just wrong... They were wrong by a factor of 10! Yet that's the evidence that the authors of the World Opinion study employ to prove that people are misinformed on stuff like health care?

Maybe, rather than viewing people as "misinformed", we should be recognizing that the results of their study show a higher level of mistrust for government, and government statistics. Blindly believing what the government claims about the economy has proven to be a gigantic mistake, time and time again.

Let's try not to forget that back in 2006, just before the housing collapse, our wonderful Federal Reserve Chairman, Ben Bernanke said:
"A leveling out or a modest softening of housing activity seems more likely than a sharp contraction…”
Yeah... Good work.

So specifically, I'm going to utterly reject the fictitious notion that government stimulus has "saved or created" any jobs. That's pure Broken Window Fallacy reasoning, and nothing more. I'm further going to reject the idea out of hand that a massive government health care mandate filled with entitlements and pay outs is going to reduce the deficit in any way, shape or form. Again - the CBO score reflected their limitations, not reality. I wrote about this a long time ago in an article titled "The Parmenides Fallacy in Health Care".

Lastly - Any projections of current "economic growth" relies solely on GDP data, and well... Guess what... GDP goes up when government spends more. Private sector job growth is non-existent, and there is no economic growth happening outside of the state-sector.

So... No. Fail.

When unemployment has hovered at 9.6% for two years (and even increased a little in the last couple months) and isn't even counting people who gave up looking for jobs, and when prices at the store are rising, and wage rates are stagnating... Looking at numbers like GDP or CPI really radically miss the point. The authors are basically suggesting that people are misinformed for being skeptical of highly dubious claims made by politicians who make careers out of screwing things up.

So are you "misinformed" for believing that the economy isn't improving, that adding a massive and impossible-to-fund Federal government program to accommodate the payment of insurance for 30+ million currently uninsured Americans will add to the deficit, and that the stimulus had negative economic effects?

Or... Are you just employing the basic critical thinking skills that the study authors have abandoned?

* * * * *

Now... Here's the thing... I don't know what this says about Fox News - and I don't care. I don't watch it. I don't watch MSNBC either... Since I don't actually own a TV, I tend to get my news directly from wire-services, the internet, and through the magic of social networking. I find the whole cable news concept to be broadly misinforming and an abysmal way to learn about the world. But this study doesn't show that. It just shows that people can use weak science, bad questions and weak statistical gathering methods to say whatever they want to say.


And oh by the way....... Total news viewer-ship is roughly around 7,000,000 per day. Want to know how big the sample size of this study was?

848 people... So they're making assumptions about a very large group of people based on around 0.15% of people who are presumed to be part of that group. Why does anyone think this kind of nonsense qualifies as valid information!?

Seriously... Stop it.

6 comments:

serket said...

I enjoyed your post; it seems like there aren't many conservatives reacting to this poll. I decided to look into the claim about a majority of Republicans supporting TARP and found that in the Senate the vote was 34-15 and in the House it was 91-108; combined that means 125 Republicans supported it and 123 opposed it during the final vote when TARP actually passed.

Ichneumon said...

That was excellent, thank you. The flaws and propagandistic sleight of hand involved in this "study" were glaringly obvious, but one would never know that from reading the self-congratulatory blogs that are gleefully spreading this "proof" of what they already believe...

Sean W. Malone said...

Just to be clear... I am not a conservative by any stretch of the imagination. I am libertarian, so I really am not writing stuff like this just to defend Fox or Republicans.

But I do loathe public opinion polling and manufactured reality. It is always something biased journalists get excited about because it confirms whatever they want it to confirm. Stuff like this creates an air of legitimacy that the ideas don't deserve and I find that to be intolerable.

Anonymous said...

Assuming the confidence level calculator I found online is correct, and the sample was chosen correctly, 848 individuals is sufficient to describe a population of 7,000,000 with a margin of error of 4% and a 95% confidence level. The researchers themselves claim 3.4%.

Potential problems with the sampling include their use of the phone book (many people are unlisted, do not answer calls from unknown numbers, do not have a land line), USPS mail (many people would view it as junk or some sort of pretext), and the use of an internet survey instrument (older respondents might not be comfortable with it).

There are also some potential flaws in the questions. Q34 asks about climate change, but does not define it. Does it refer to climate change over geological ages? Or climate change in the near future? Does it refer only to anthropogenic climate change, or climate change due to any causes? "Network TV news" is not defined in D3c. Respondents may have thought the Cable News Network and Fox News Channel were "network TV news" outlets until they saw questions D3d et seq. Were respondents able to return to D3c and revise their answer?

There are many things to be questioned here, but I don't think the sample *size* is one of them.

Sean W. Malone said...

I would argue that there is a fundamental problem with using a fraction of 1% of a given population to account for the views and beliefs of the total group.

To believe that statistical sampling like this produces accurate results in the realm of sociology is absurd. And that's without even concerning myself with the exceedingly obvious biases in the questions and framing presented to participants.

And yea, of course getting participants from a phone book in 2010 is ludicrous. I am unlisted, as are nearly all of my friends... and I am willing to bet that's true of a lot of younger, tech-savvy people like me who get news from a wide array of sources, and who aren't duped by government statements about the economy or economic outcomes of legislation.

Mostly I want to drive home the point that public opinion polling is virtually all bullshit - this example being especially awful.

Sean W. Malone said...

Ps... I wouldn't mind the sample size if the conclusions of researchers and more importantly, those of idiot reporters were actually in line with reality.

Taking a tiny group and doing some poorly constructed questionaire shouldn't lead anyone to think they have discovered some profound truth about their subject that they can scale up to encompass millions of individuals. It's an epic epistemological failure.