Edited Date/Time:
Statistics don't lie, but people can present statistics in a misleading manner (sometimes unintentionally) to misinform their audience.
http://racerxonline.com/2017/07/04/breakdown-subtle-shades-of-gray
Before I go any further further, I want to make it clear that I agree Jason Anderson is partially at fault for the collision between himself and Baggett on the opening lap at Red Bud. The part I disagree with is JT's ability to present statistics in a scientifically correct format.
After listening to the RacerX podcast on Monday I was eager to read JT's Breakdown on RacerXOnline. He sounded very confident and repeated "Number don't lie" or some variant of that phrase while hinting that he had used some statistics to back of his difference of opinion with Grant Langston.
Full disclosure, JT is obviously a smarter and more talented racer than I could ever have dreamed of being. No doubt about that. To use a cliche, he has forgotten more about the topic than I probably will ever learn. That being said, I disagree with his use of statistics in his column this week and I even feel a bit awkward attempting to challenge the work of someone who so easily outranks me in terms of knowledge. I mean no disrespect, I only wish to open a dialogue about how I (And potentially others) see this situation.
My first concern is that there is no Inter-Rater Reliability. This basically means that because JT was the only person collecting time's for the section, his times cannot be compared to anyone elses to determine if his times are accurate. To know for sure, it is good to have another person time the section too.
Secondly, and most importantly, JT writes "After analyzing the numbers, I hypothesized that Anderson indeed slowed just a touch in order to make sure Baggett’s intersection point would be filled by a Husky 450.". and here we identify a cardinal rule being broken as JT declares his hypothesis after analyzing data.
Why is this important? When following the scientific method, or attempting to, you always declare a hypothesis before collecting and analyzing data, so as to ensure you do not bias yourself, or participate in what we call "confirmation bias" whereby someone looks for information to confirm an already held belief.
A hypothesis is declared and then tested with the data that is collected, not the other way around.
Thirdly, JT proposes these statistics in a vacuum. Where is the baseline for comparison? How many other riders lap times in this section did JT watch "hundreds" of times, measure, and analyze? None, it seems. Which goes back to the previous point, he was only looking for data to back up a preexisting belief rather than analyzing the totality of the situation (more than a single rider).
People may wonder why I would go through the trouble of posting all this if I generally agree with JT that Anderson did indeed slow slightly to cause contact with Baggett and my reason is this: he claimed "numbers don't lie" on the podcast repeatedly which fools people into thinking that so long as you have numbers to back your point you are more-correct than others.
My thought is, if you are going to go the extra mile and attempt to add statistics to the journalism, at least do it in a scientifically correct manner.
I wonder, am I totally off base and an idiot here? I could be. Does anyone have any similar or differing thoughts on the topic?
Just want to hear the thoughts of other community members.