Being published doesn’t make it true

Lay people are often bewildered by the fact that doctors and scientists disagree with each other. The Wax study says homebirth increases the risk of neonatal death. The Johnson and Daviss BMJ 2005 paper says homebirth is as safe as hospital birth. What’s the truth?

When faced with conflicting scientific claims, lay people often conclude that the truth is simply a matter of what you prefer to believe. Even worse, they occasionally conclude that there is no truth or that the truth is unknowable. It might help, though, to consider a real life example. We know that there are newspapers and news organizations will often report conflicting accounts of political disagreements. And we know that just because we read something in the newspaper, it is not necessarily so.

Reading a scientific paper is similar to reading a newspaper article. A Democratic leaning newspaper may have an article with the headline that Obama was born in Hawaii. A radical Republican newspaper may have an article with the headline that Obama was born in Africa. That does NOT mean that Obama’s place of birth is indeterminate or that we cannot know where Obama was born.

The abstract of a scientific paper is the equivalent of the headline in a newspaper. It tell you the conclusion that the author wants you to draw. It does NOT mean that the conclusion is true, anymore than a newspaper headline means that the article underneath it is true.

The body of the scientific paper is the equivalent of the body of the newspaper article. It offers facts and draws conclusions based on those facts. Even articles with false claims will offer facts. The radical Republicans offer facts for their claim that Obama was born in Africa: his middle name is “Hussein;” his father was born in Africa; there are not many black people in Hawaii. The Democratic newspaper offers facts: it might show a picture of Obama’s Hawaii birth certificate with the official seal; it may have obtained access to Obama’s hospital record from the day he was born.

So we have two articles with two different conclusions and two different sets of facts. Does that mean that we cannot know where Obama was born? Of course not. It is a fact that Obama’s middle name is “Hussein” and it is a fact that his father was born in Africa, but that is actually irrelevant in determining where Obama was born. The birth certificate and the hospital record prove that Obama was born in Hawaii.

In other words, knowing which type of data is most important makes all the difference in determining what is a reasonable conclusion.

Similarly, an abstract of one paper may say that homebirth increases the risk of perinatal death, and the abstract of another paper, like the Johnson and Daviss BMJ 2005 paper, may say that homebirth has the same rate of perinatal death as hospital birth. That does NOT mean that abstract accurately reflects the data in the paper or that the claims made in the abstract are true. It is ONLY by READING the entire paper and applying the principles of statistics that we can know which paper is true. There is NO other way.

When you read the Johnson and Daviss BMJ 2005 paper, you learn that Johnson and Daviss never compared homebirth with a CPM in 2000 to low risk hospital birth in 2000. They compared homebirth with a bunch of out of date hospital statistics extending back to 1969. Yes, it is a fact that homebirth was safer than those hospital statistics, but the conclusion of the paper (that homebirth is as safe as hospital birth) is not supported by that fact. Moreover, a little independent research will show that the comparable hospital death rate in 2000 was actual 1/3 of the death rate for homebirth. So Johnson and Daviss NEVER showed anything about the safety of homebirth regardless of what their abstract says.

Contrary to the beliefs of many lay people, you don’t pick your conclusion and choose your experts accordingly. And just because experts disagree does not mean that there is no way to know whether homebirth increases the rate of perinatal death. But one point is incontrovertible. In order to interpret scientific evidence, you MUST read the actual scientific papers and you must analyze the data within them. Just because someone writes a scientific paper that supports your preferred conclusion does not make it true.

Trackbacks/Pingbacks

  1. cheap rolex watches for sale china - July 28, 2021

    … [Trackback]

    […] There you can find 35250 additional Info to that Topic: skepticalob.com/2011/04/being-published-doesnt-make-it-true.html […]

Leave a Reply

You must be logged in to post a comment.