Dear all:

In 2007 I was motivated to conduct a little study on my own because of a controversy that erupted on Lactnet about the "accuracy" of test-weighing.  I went to the original article which concluded that the "accuracy" was just fine it was the "precision" that was off.  In reading the details at the time, it became very clear that they have NEVER TESTED PRECISION.  Precision is tested by repeating the same estimates over and over again with the SAME instrument.  There are different types of precision-- repeatability:  which is when everything is kept constant and you repeat the measures within a short interval; reproducibility:  which is when you may change the observers or the instruments or conduct the experiment over a longer duration.  When you change the observer -- you are actually assessing inter-observer error, when you change the instrument, you are changing inter-instrument error and when you wait to repeat the measurement over a long interval you are testing dependability.

Now, having reviewed quite a bit of evidence on this topic, NO ONE HAS REALLY TESTED PRECISION except for dependability.  Dependability plagues all assessments.  If you assess a baby using any tool included your eyeballs only once -- you have not really covered the dependability issue -- be it swallows, diapers, behavior, or weight.  It is dangerous practice not to actively listen to what the mother is telling you and follow up.  

Embedded in the original article, after four years of contemplating the fact that the education in mathematics and statistics is so abysmal that even a peer reviewed journal cannot even spot a GLARING error such as mistaking what actually was a very poor test of accuracy for a test of precision -- I found a very telling one line sentence in the original article.  

The authors THOUGHT they were testing evaporative water loss.  Evaporative water loss is basically sweat loss.  They actually did repeat their measurements of test weighing.  They concluded that evaporative water loss was unimportant.  What they actually conducted was a test of precision.  The median and the interquartile range of the differences between these estimates was ---- get this 1 g.  Now PLEASE tell me how one can conclude that a repeat measure with only 1 g of difference even with evaporative water loss is IMPRECISE?   By their own study results their repeatability was at least as good a within 1 g of the original estimate.  I have found repeatability of eyeballing the volume of milk in a bottle to be off by much much more.  

I have been getting lots of articles sent to me lately about how evidence-based medicine is the downfall of clinical practices.  I would argue that lack of critical thinking on even very basic issues of math and science have led to lack of an evidence base.  It does not matter whether or not you are practicing based on experience or belief or evidence, if you do NOT engage in critical thinking. 

This means that when you think you have the world's best cure for whatever ails breastfeeding, that you put on your critical thinking cap and come up with all the reasons you might be wrong and then test out all those reasons.  You do not collect only the information that confirms your own personal experience, belief system, or abstracts of the research that you did not read in full detail. That is known as confirmation bias.  As humans it is always validating to find others who think like we do, but that can lead to dangerous complacency.  

PS.  This issue is not really about test-weighing -- it IS about critical thinking.

Sincerely,

Susan E. Burger, MHS, PhD, IBCLC

             ***********************************************

Archives: http://community.lsoft.com/archives/LACTNET.html
To reach list owners: [log in to unmask]
Mail all list management commands to: [log in to unmask]
COMMANDS:
1. To temporarily stop your subscription write in the body of an email: set lactnet nomail
2. To start it again: set lactnet mail
3. To unsubscribe: unsubscribe lactnet
4. To get a comprehensive list of rules and directions: get lactnet welcome