Mike Baker Authenticated

[Snowman]

Originally Posted by Dead-Ball-Hitter

I don’t mean to offend, but I have to call bull on this one. I worked previously with the BLS, that’s the Bureau of Labor Statistics, who’s tag line literally is, “We have a stat for that.” A guy with a spreadsheet recording a few sales and various odd pieces of descriptive data does not a data model make. You won’t convince others that this path is meaningful. For that matter, No need to justify to others who don’t agree that these “certificates” hold any special value. You like them, that’s fine. Seriously, enjoy.

Nice try with the name drop. Let's not pretend like you know what you're talking about here. You don't build statistical models.

I'm a professional statistician/data scientist. Building models like this is what I do for work every day. The primary work of a statistician is to perform hypothesis tests using data and mathematics. Just because you don't understand how it works doesn't mean that it in fact does not work.

This isn't a difficult problem to understand or to solve for. If you want to know whether or not an MBA sticker has a statistically significant impact on selling prices, then all you have to do is gather enough data (a few hundred sales is more than sufficient) and build a statistical model to perform any number of hypothesis tests. You can control for other variables by including them in the model. Variables you might want to control for could be: Which grading company graded the card, the grade of the card, which card it is, centering, pop count, sticker or no sticker, date of sale, a market index to tether to the date, and numerous other things you might want to track like stains, creases, focus/registration, etc. Then you can just build a simple (or complex if you want) regression model or an ANOVA/ANCOVA model, or some machine learning models, or a multilevel mixed effects model, or whatever other model you want to build that tests hypotheses. This is pretty elementary stuff in the world of statistics. It's really not that complicated. Models will estimate the impact of each variable included in the model, and it will also output a P-value that informs you whether or not that impact is statistically significant (the more sales data you have in your model, the more power the model has in determining statistical significance).

But I'm sure you know all this stuff because you worked "with" the Bureau of Labor Statistics lol.