Do any of you guys at work use a lot of statistics programs to conduct data analysis, perform experiments etc? I feel like some (a few) people know it where I work but a good chunk like R&D has absolutely no clue. Because I'm not a senior engineer with a PhD I get handed what I like to say bitchwork sometimes based on conclusions from data that I don't even feel comfortable with. I feel bad for the operators sometimes simply because the decisions come from above and the people who want the experiments done for themselves don't even know how to extract data.
It's like some people just throw two means out there and conclude that this is the better method when things like spread, variation, etc aren't even looked at. We have a terrible system of experimentation done, and it's disturbing that the guy at work who pulls and presents most of the data has NEVER used statistical software like JMP or Minitab before. I've never ever seen a presentation done where a p-value is even shown. I don't see how one can conclude there's statistical significant evidence when no one even makes an effort to show it. It's sad.
Kinda makes my internship before look like a joke where while peoples backgrounds werent in industrial engineering, everyone seemed like they knew general statistics, probability and how to design experiments. Of course I'm no expert and as a newgrad I feel like I'm not really a stats guy having only taken one manufacturing course and having done one internship. So I just keep performing my experiments my own way, but whenever people hand me what I see as stupid experiments, I suffer and need to spend hours to finish (gosh haven't people heard of factorial DOEs?).
BTW, I'm posting from "training" because someone at work finally felt like we need statistics training that they brought in some consultant to train like more than half the engineers in this company.
Indstuctor says, "Area under the curve is probability" *people furiously scribble down notes. SERIOUSLY?
/RANT
It's like some people just throw two means out there and conclude that this is the better method when things like spread, variation, etc aren't even looked at. We have a terrible system of experimentation done, and it's disturbing that the guy at work who pulls and presents most of the data has NEVER used statistical software like JMP or Minitab before. I've never ever seen a presentation done where a p-value is even shown. I don't see how one can conclude there's statistical significant evidence when no one even makes an effort to show it. It's sad.
Kinda makes my internship before look like a joke where while peoples backgrounds werent in industrial engineering, everyone seemed like they knew general statistics, probability and how to design experiments. Of course I'm no expert and as a newgrad I feel like I'm not really a stats guy having only taken one manufacturing course and having done one internship. So I just keep performing my experiments my own way, but whenever people hand me what I see as stupid experiments, I suffer and need to spend hours to finish (gosh haven't people heard of factorial DOEs?).
BTW, I'm posting from "training" because someone at work finally felt like we need statistics training that they brought in some consultant to train like more than half the engineers in this company.
Indstuctor says, "Area under the curve is probability" *people furiously scribble down notes. SERIOUSLY?
/RANT
Facebook
Twitter