Friday, January 3, 2020

Measuring

Sometimes it is said that if something is important, it needs to be measured. Sometimes it is said that if something is measured, attention will be paid to it.  Part of my graduate education and much of my teaching was about testing students and giving grades. The story of "standardized" tests involves Alfred Binet of France.  A similar story and approach is found in "How to Measure Anything: Finding the Value of Intangibles in Business" by Douglas Hubbard.  

Whether it is intelligence or business acumen or leadership ability, measuring an intangible, especially one that matters to humans, is tricky.  I heard enough about measuring and testing to immediately be on my guard when the subject comes up.  

Here is what I wrote about this subject in 2011:
A friend wrote about intelligence.  That is a subject I used to deal with quite a bit.  My experience is that it is best to leave a blank when trying to assess someone's intelligence.  After physical science made some very impressive strides during the 1800's, psychologists wanted to do the same thing.  Alfred Binet was asked by the Paris school board to build a test that could tell whether a child was mentally capable of benefiting from normal schooling or insufficiently intelligent for it.  Louis Terman of Stanford University modified that test to create the granddaddy of American intelligence testing, the Stanford-Binet test.  

My doctorate is in statistics, experimental design and measurement, which means I had to take a grad course in individual intelligence testing.  The idea is that a piece of paper with questions and exercises on it can be printed in multiple copies and used to estimate the intelligence of each of a group of children.  Those that give the answers expected by the people or machines that mark the papers are considered intelligent. However, in certain cases, an individual test is administered.  Here, a trained person sits alone with a student and asks questions by voice and requests the student to manipulate objects.  

The problem is that the method is very, very crude.  I sometimes think of getting an umbrella down a chimney.   It will go down one way (point first and closed) but not the other (open).  In a similar way, if a child is intelligent, mostly in a verbal, logical way, the test will show that.  However, the child can be extremely intelligent but not seem so on the test. Just one of many, many obstacles is language.  If the child speaks only Spanish or Romanian, and we ask a question in English, guess what happens?

Contrast what used to be the 2 leading I.Q. tests, the Stanford-Binet and the Wechsler.  The Binet test used a conception of intelligence as logic. So, Binet asks a child, "Johnny put his pants on over his head today.  Tell me what is funny about that?" If we get an answer more or less equivalent to "well, pants are made in such a way that they cannot be donned so as to be worn in the conventional way by putting them on over the head", the child is considered intelligent.  

Wechsler used a different idea: that an intelligent mind would gather certain basic info.  He thought the child that knows that basic info is intelligent but not if the child doesn't.  So, we ask the child,"How far is it from New York to Paris?" If the child says approximately 3000 miles, we have an intelligent kid but if not, not.

There are lots of other holes, just as bad.  Beware the idea that intelligence is measurable.
Especially as time brings change, we need different kinds of intelligence.

Popular Posts

Follow @olderkirby