stat counnnter

Tuesday, August 18, 2009

The Value of Null Findings

Over at The New Clarion two days ago I posted on the fact that when statistical studies result in a null finding they are rarely published and how misguided this practice really is. I reprint that post below.

The Value of Null Findings

"Sandy Szwarc at JunkfoodScience has an in-depth look at a health science topic. Although it's titled "The Myth About Unhealthy Belly Fat" the article's theme is, 'the importance of null findings', and properly laments the fact that the media seldom reports them. This is very true and also very important. That's because:
"Null findings enable true scientists to know they’re looking in the wrong direction and that it’s time to go back to the drawing board and develop a different hypothesis. They also enable us to stop needlessly worrying about something that doesn’t matter."
Also, many if not most studies that purport to show a health problem actually turn out to be false and these revelations are often not published as well.

Just as knowing that something is a threat to health and life is important, so is knowing when something is not a significant threat, especially when alleged professionals are telling us it is a threat. However, we also know that the media often prefers hype and sensationalism vs boring news about non-threats. So the media has a built in bias that is heavily weighted in favor of fear-hype while a study showing that last month's health scare is actually false will tend not to get printed. This will mislead thousands if not millions of people. As the saying goes 'you can't unring an alarm bell.'

But this situation has other consequences. It creates a fearful citizenry eager to donate their money to NGOs (Non Governmental Organizations, mostly non-profit foundations) who promise to do research (more statistical studies) to find ways to protect them from all the alleged hazards of a modern, technological society.

It sometimes works like this: NGOs fund studies that try to determine what is bad for us. When a study is completed a press release is announced. The media picks up the release and reports the findings which are usually that X (say, fast foods) is bad for us. Politicians then stand on the floor of their chambers, news reports and studies in hand, announcing that congress must address this threat to public health by passing legislation more heavily regulating the use of X. They then stand before us proclaiming how good they are at protecting us from all the harmful Xs out there and therefore we should reelect them. That is what we see.

What we don't readily see is that the funding NGO has been getting large donations from companies that produce products Y and Z (say, treadmills and diet books). Also mostly out of view is the donations by companies Y and Z to the politicians' reelection campaigns. And even further out of sight is the fact that no one looked at the study, not the reporter, the editor nor the politician, to see if the study had any statistical significance or was a null finding being passed off as having significance. This last happens often.

We have to educate ourselves to what is valid science and what is pseudo science because the press isn't doing it. Pseudo science isn't just coming from NGOs either. Government agencies fund lots of it. I recommend reading the entire article and visiting JunkfoodScience often. Her blogroll is a good source for others who actually read these so called scientific studies and then examines them for our benefit."

Health care has become so politicized that one literally has to fact check every alleged harm for oneself. The problem is that grant money is not earned money. It is handed out to applicants on the basis of the applicant's desire to see if there is a problem regarding food X's effect on say, disease Y. The granting agencies are virtually buying problems the government can try to fix with more regulations over you and me. Are we to believe that the applicant's supply of grant money will continue if they produced studies showing 'no problem here' i.e. a null finding?

The solution is to remove government encouragement from the market place especially from education and science. Government control of education is the open door to government control of everything else.

No comments: