[Nisupulla]

Quote:

There are three types of lies: Lies, damn lies, and statistics~Disraeli

What minimal knowledge of statistics does a lactivist need? The forbidden thread about reasons people choose not to BF got me thinking about this. More than once I've heard people say the consequences of not BF are not significant enough to justify the effort. I've also heard that the 6-8 pt IQ difference dismissed as too small to worry about.

Statistics 101

True statement: In cities where there are more churches, there are more murders.
Quiz: Why is this statement true?
Hint: If you are thinking, "Why would having more churches increase the number of murders? That doesn't make any sense." You are on the wrong track (but doing what I hoped you would). Answer's at the bottom of the post.

Of course, the lack of knowledge about statistics leads to comments such as "I was fed formula and I'm fine" or "Joey was breastfed for a year and he's obese" or "One study says BF reduces asthma and another says it increases asthma. There's no way to know what to believe, so I make my own choice." etc.

Yesterday I stumbled onto the statistics about hormone replacement therapy (HRT). Remember? That was the study that was cut short because the risks of HRT were too significant to ethically continue the study.

Then I was thinking of the similarity between HRT and formula. Both are mostly about women's lives. Both involve using something artificial to change the natural process in order to increase quality of life.

Here are the HRT statistics I found. There was a 50% higher risk of coronary artery disease events during the first year of the study. Sounds huge, doesn't it.

But the real numbers actually seem small. For every 10,000 women 8 more got breast cancer, 7 more had a heart attack, 8 more had a stroke, and 18 more had blood clots in the lungs if they were on HRT.

From my source, "While seven, or eighteen, out of ten thousand women represents only a very small increase, it was unacceptably high for doctors conducting this medical research. Understandably, the doctors could not condone knowingly endangering the health of the volunteer participants of the study."

Bringing this back to breastfeeding, I think the numbers are probably small changes for individual risk, but medically significant. I think that might be why it's so frustrating for me to hear people use the statistics as an "excuse".

Comments or suggestions on how to keep the statistics plain and simple?

Answer: In cities where there are more churches, there are also more people in general. More people more murders. The churches and murders are correlated but there is no causation. Both number of churches and number of murders increase due to higher population.

 

[annekevdbroek]

I think some of the confusion comes from misunderstanding correlation - that is something can be related (correlated - tending to occur together) without any causal relationship between the two events. Your example of churches and murders is an example of correlation. Correlation is frequently confused in the public and press as indicating causation - which it does not.

On the other hand, with scientific research you hope to find causal relationships. That is, in the case of BF and IQ you look at children who are BF and those who are not and try to control other variables that effect IQ (control staistically, not environmentally). So two groups of children, on BF and one formula fed are matched on every other possible variable - e.g. birth complications, parent's IQ and educational level, and any other pre/post natal variables that you can control. You test the IQ hopefully as early as possible and then as the children develop to see if there are differences between the two groups. One of the studies on this topic looked at infants who were premature - 50% received formula and 50% received breastmilk (donated, I think). Since physical contact from nursing was controled for - both groups of children were bottle fed - the researchers tried to make everything aside from teh BF equal between both groups. This study showed the increased in IQ points.

A difference of 6-8 points is half of one standard deviation - which is pretty large in terms of IQ. Other studies have suggested that the longer BF goes on the more IQ benefits there are....

Anyhow, the other issue you mention is that actual risk (e.g. 5 vs. 10 people out of 10,000 get some disease) vs. statistical risk (100% more people get the disease). That is a tricky statistical mess isn't it? In my experience doctors/etc. who use this information are aware of the different meaning of the numbers - and do take it into consideration. However, the more alarming statistics tend to be what shows up in the press. I think that the issue of circumcuision and increased risk of cancer is one of the issues that is effected by this statistic - huge % incresaes, but the real numbers are very small (e.g. 1 vs. 2 of every 100,000 or something).

People also confuse anecdote with data. That is, "I was fed formula and I'm fine" does not equal scientific research. Of course, many of the children on formula are fine - but the *risk* of certain problems is higher with the use of forumal, which is no guarantee that those problems will or will not occur with either BF or formula.

Sorry to be so long winded. I teach intro psychology and this issue of correlation/causation comes up frequently in class.

 

[sapphire_chan]

Offtopic: I got the church/murder question right! I'm so proud of me. My favorite correlation along those lines, is "100% of people who consume some form of dihydrogen monoxide will die."

I think newspapers and things should have regular "interpreting statistics" columns just in general. In the Washington Post they have a regular feature called "Quick Study" that looks at new medical findings, they always have a little blurb with caveats about the presented study.

 

[Clarinet]

Quote:

Originally Posted by sapphire_chan Offtopic: I got the church/murder question right! I'm so proud of me. My favorite correlation along those lines, is "100% of people who consume some form of dihydrogen monoxide will die."

My high school chemistry teacher used this on us. Gave a sheet with all the ways dihydrogen monoxide can kill you. After several minutes of discussion, some students were all for banning the killer substance. It was amusing.

\]

Back on topic, I have found that people just don't care about statistics unless it happens to them. Something like 70% of car accidents happen within 10 miles of your house. Well, my car accident was about 20 miles away so while I believe the statistic, I don't drive more carefully or pay attention more when I near my home. 50% of marriages end in divorce. I'm still married. I know the statistic is true but I don't see it as my marriage has a 50% chance of failing.

 

[nakagain]

50% marriage thing is more like 30% will divorce but remarriages have a 70% to fail..so thus %50% of marriages fail

its all about wording....

 

[Nisupulla]

So how does one go about arguing with credibility that statistics support BF?

 

[jillc512]

I think it is easy for some to dismiss 'studies' and 'statistics', but it is more difficult to dismiss a particular study showing a particular thing that has a large sample size, adequate controls, etc.

If you're trying to argue with an individual, find the study that shows BF is important because of a reason that is important to them. For example, my friend's mother had breast cancer. She would be much more interested in hearing about a study that says BF reduces the mother's risk of breast cancer by x % vs. just the general "according to recent studies, breastfeeding is healthier".