Source of the post
Scientists use statistics all the time to back up their research, yet few have any academic background in statistics to speak of. This may be a serious problem (I have to say "may be", as I never did statistics at university myself and I regret it, so I can't really back up that claim.
If you like, consider checking out the text Introduction to Error Analysis
by John Taylor, which is written at the undergraduate level to explain principles of statistics and their application to data collection and model testing. It's quite compact and readable for a textbook, and I still reference it frequently.
Source of the post
Statistics need to be viewed with severe skepticism. ... Statistics has its uses but finding a mechanism that causes the action is more important, because that's actual science.
Statistics is not the problem.
Statistics is a tool, and all science depends on it. Demonstrating causation requires it! You take some model describing "how A causes B", make predictions with it, compare them to the data, and show that they fit the data while other sensible models or the null hypothesis do not. Showing that the model fits the data means plotting data points and showing they are consistent with the model predictions within error. Being even more rigorous, we'd then do a chi-square test or something similar to say, using statistics, how well does the model fit the data. What is the probability that the agreement is not a statistical artifact?
What must be looked at carefully are not the principles of statistics, but rather the methods, assumptions, and the presentation of the data. Usually "lying with statistics" means being sly about how you present your case. The earlier example of p-hacking is a good example.
Source of the post
JD and Wat, did you read about the research from the four universities (Columbia, Harvard, UC-Berkeley, UC-Davis)? It's pretty damning.
Of course. I think you mistake me for thinking these companies treatment of waste disposal is not disgusting. I'm not saying anything of the sort. I'm not advocating that anyone ignore it or pass it off as "probably not a problem." What I'm advocating is caution about what specific conclusions we reach from various studies, or how strongly we believe them.
I actually used to love statistics and started studying it on my own in high school and even bought myself a statistics calculator (a Texas Instruments machine) back then to ease the process. When I saw some of the tainting being done in some industries to skew the numbers, thats what got me down on statistics.
I also saw some "scientists" use it to defame minorities; intelligence "The IQ Cult"- which still seems to be going on in some circles, unfortunately, it was a throwback to "The Bell Curve." The reason I got so interested in that is that my middle school back then had a gifted program for which everyone was tested to get into. I somehow didn't make it in and yet I saw many of my friends were in that program, and yet I KNEW I was smarter than most of them (I was tutoring many of them in science and math.) So I asked my mother to call the school and was told my score was "131" and that I had missed the mark by one point. I knew something was fishy here because NO IQ test is sufficiently accurate to be able to dismiss someone based on 1 IQ point. I convinced my mother to pay $20 or so to get the official Mensa test (even though it was for adults) and I took that and scored a 158 on it. We photocopied the results and sent them to the school but they said they wouldn't accept the results of an "adult intelligence test" LOL, they only used the results of their CAT test. That's how I knew that statistics have been used by white "scientists" to discriminate against minorities just like they are used by polluting industries to discriminate against the people who suffer from the toxins they dump. Looking back at history, the whole neonazi movement actually began in America, and that is where Hitler got his inspiration from.
PS to this- on every IQ test I've since taken, I've scored 142 or higher, and averaged around 149 (including the Mega Test, which was made for "one in a million"), which is what got me into Triple Nine Society. I dont care much for such societies, but I guess I was always a bit miffed at what happened to me when I was in middle school. It just made me very skeptical of statistics as prejudiced people use them to further their prejudices.
PPS I LOVE the chi square test, it is both simple and yet elegant. We used that in genetics all the time!
PPPS Sadly, there is still IQ-based bias going on in the world today, I saw a world map earlier with the "average IQ" of people in each nation, and some nations supposedly have an "average IQ" of under 70?! If that were actually true, those nations would be nonfunctional.....
PPPPS you're completely right about individual studies, which is why we need meta-analyses using multiple studies.