“The fact is that, despite its mathematical base, statistics is as much an art as it is a science.”
Data. Numbers. Thinking about them.
Being the Director of Research & Outcomes, this is my world. Numbers are concrete and irrefutable, on the surface. But they are also subject to context and interpretation when one thinks about them. Where did the numbers come from? How was that group defined? How widely was this tracked?
When you hear that the average grades for children in your school district is a 3.0 or a B, is that good enough? If you think about this, it could mean that ½ of the students earned Cs and ½ the students earned As – making B the average. It could also mean that 1/3 of the students scored Ds and 2/3 scored As. There are many combinations of grades that could equal a district’s average grades of 3.0. Most people use the average, or mean, to report data about groups. But it may be more accurate to report a different measure of central tendency. Might it be more useful to report the median grade (the grade in the middle of all the grades)? Or the mode (most frequent grade?). All are measures of average. Quickly, a simple statement becomes a cluster of questions for someone with a brain like mine.
There is a favorite classic book, How to Lie with Statistics by Darrell Huff. It was first published in 1954, and despite some old-fashioned examples, offers cautions about data that is timeless. The book remains a priceless resource for anyone unaccustomed to looking carefully at the volumes of numbers reported by advertisers, financial firms, retailers and researchers. It’s a fun book to explore.
Most data has bias. Or a strong possibility of it. Consider that it is humans that are collecting and collating this data for reporting. And by definition, humans are human (read imperfect).
Seth Grodin remarks wisely in his April 20th 2016 blog, Numbers & the Magic of Measuring the Right Thing that when one measures the wrong thing, one gets the wrong thing. Is the #1 music hit always the best song? Precise measurement is not necessarily significant. It’s only significant if what you’ve measured matters.
Recall from an earlier Spurwink blog that the prevalence rate for children with autism is 1 in 68. This is an incredibly powerful statement. It reflects lives of people all over the country and impacts federal funding spent on research and intervention. This is SO important. This makes the precision of any reported data about Autism’s prevalence important as well.
If it is worth measuring, it is worth measuring well. Decide first if you even care about the number being reported.
Blind acceptance or stubborn refusal of data are both equally shortsighted. Healthy skepticism is fine. Asking questions is even better. Curiosity is what drives inventions – electricity, the telephone, the computer, life-saving medicines. It moves our human culture forward. Join in the process.
by Linda S. Butler, Ph.D., LCSW
Director of Research & Outcomes