Social Sharing block
Not long ago, I couldn’t abide statistics. I respected the field, in much the same way a gazelle respects a lion. Most of my early experiences with statistics indicated that close encounters resulted in pain, so I avoided further contact whenever possible.
|
ADVERTISEMENT |
So how is it that today I write about statistics? That’s simple: It merely required completely reinventing the way I thought about and approached the discipline. When I decided to approach it as a language rather than a purely mathematical set of skills, the doors opened.
Why does my experience matter to you? If you’re a statistician yourself, you know all too well the typical reactions people have when they learn we work with statistics and data analysis: blank stares, uncomfortable silence, horrible jokes, or some variant of, “Oh, how nice,” followed quickly by, “Excuse me, I’m going somewhere I don’t have to talk about statistics.”
…
Comments
Stat's
In the auto industry we started to use statistics starting in the 80's and used it as a tool or comprhension for process control and quality management, it was amazing how much it worked, but, only for a small handful of us who understood the basics. Today, I think it's gotten more complicated because the basics seem to be ignorged and in many cases, by most, someone with some kind of a 'belt' needs to expalin it? Everything under the sun is being polled and everyone seems to have a different answer. One thing is clear figures don't lie but people who figure do, so, the conversation has to continue and getting back to the basics will make for more understanding.
Similar path
My path is similar. As an engineer, I hated my probability and statistics course in undergrad. Who cares about red marbles and blue marbles in a jar, drawing with or without replacement?
Two years into my first job, I realized the beauty and power of statistical methods. Got my Applied Stats degree a few years later. Since then I have worked over 25 years teaching and utilizing industrial statistics in my everyday work in a variety of fields and applications.
What should we do? Several things.
Stop teaching probability theory. Stop teaching the formula to a t-statistic, much less talking about it. Teach the method. When should I use a two sample t? One-way ANOVA. As long as they can find and correctly interpret the p-value, they can be successful.
I teach statistical thinking as a foundation. There are always resources to help with Minitab or computations, but thinking statistically makes one more powerful.
I make a huge deal about the role of statistics in solving business problems. Convert a business problem into a statistical problem, solve the statistical problem, and convert it back into a business solution. Most classes get hung up on the language of "Reject Ho, conclude Ha with 95% confidence that blah, blah, blah". Stop it! Only the rare senior manager wants to hear statistical mumbo-jumbo. State the business solution. Part of your argument of language I agree with, but part I would say isn't "language" but "thinking".
Finally, I preface all my statistics classes by telling them I will teach them "How to Lie With Statistics". And then I do. They appreciate the humor, but in teaching how to lie and how to do it correctly, they become better at statistically thinking, they learn the right way and the wrong way, and they become better consumers of information.
Good article!
KISS (Keep It Simple)
Dr Wheeler has posted many brilliant articles here on the complexity of statistics involved in quality. His conclusion, and that of his entertaining books such as "Normality and the Process Behaviour Chart", is always that simple Shewhart Charts are all that is needed. The complexities introduced by products such as Minitab are not only superfluous, they hide the real meaning behind data. Dr Wheeler may not be the best Minitab salesman but he is the world's best quality statistician.
Add new comment