|
Post by jaffinator on Aug 16, 2021 16:20:56 GMT -5
So for the study, .3 is statistically significant - therefore sufficiently outside of the sampling distribution under the modelling done in the paper. The issue is that that's not really enough evidence on its own to just say "this analysis shows that this is a real relationship" anymore. Yes, classical statistics a la Pearson and Fisher might more or less say that increasing sample sizes are not inherently an issue, but (relatively) recent developments in meta-research have shown that things like researcher degrees of freedom and p-hacking are real issues such that simply saying "well the coefficients are statistically significant" without doing anything else leads to bad, irreproducible science.
Moreover, it is not the case that more sample size inherently leads to false positive results. But it is the case that it dramatically increases your chances of finding minute differences that are not truly meaningful. This leads to cases in which the test does not actually answer the research question at all; it is not clear from reading this paper (to me at least) if umpire racial biases really to meaningfully alter ball/strike calls. Large sample sizes also amplify any issues there might have been in data collection or sampling, though it's not completely clear how much of that is going on here.
|
|
|
Post by kingofthetrill on Aug 16, 2021 17:30:53 GMT -5
I'd be more curious if race effects other calls rather than balls/strikes. Like ejections.
|
|
|
Post by kcbosox on Aug 16, 2021 18:16:07 GMT -5
I have briefly read the paper. First, when looking at std devs over six decimals long, there are possibly issues. Second, what I did not find was data interpreted or presented braking down race of umpires compared with race of BOTH pitcher and batter. To only interpret the data on individual dyads leaves out a huge interaction effect. Analyzing simply same race batter-pitcher dyads, and mixed race batter-pitcher dyads does not sufficiently break down the variables. Also, a sure fire sign of incomplete data evaluation can be seen in the limitation sections, suggesting, in this paper, that attendance could be a relevant variable, is ridiculous, as if the were serious, they would have suggested more meaningful limitations. Lastly I am concerned about the sample sizes of black and white pitch counts. The number of difference in total data points demonstrates a weakness in the data.
|
|
|
Post by jl1947 on Aug 16, 2021 20:40:15 GMT -5
Also, just because correlation does not equal causation, doesn't mean it is not the source of the cause. You can say hitting yourself in the head with a hammer leads to head injuries and someone can say that correlation does not equal causation. But sometimes it is. Right. If a and b are correlated that does not inherently mean a causes b. But if a does cause b, it should be the case that under correct model specifications a and b are correlated. Yes. The ice cream analogy makes no sense. The Hypothesis must be well constructed, measuring things that may have a causal effect, or may have a significant relationship, an alternate hypothesis, or things that prove no significant relationship, or possible cause, null hypothesis. Hypothesis testing is the beginning of the inquiry. Further study is almost always the next phase, and that's what this alternate hypothesis seems to suggest. The difference is not huge, but seems to be persistent enough to suggest causation maybe be at play. Unless, that is, anyone is suggesting selection bias. If the data is contiguous, and of large enough sample or population, or even if it is not contiguous but of random samples of sufficient size that are controlled for homogeneity of means, and the alternate hypothesis proves significant, then we should not dismiss it out of hand. As for the "woke" reference used by one of the posters, the manner that term is used by some on the right is as a new kind of dog whistle, is getting tiresome, and mostly unhelpful; a little bit intellectually lazy, or purposely used to obfustcate a matter of serious import. Similar to calling all Trump supporters fascists. Some are welcoming of fascist behavior by their leader if it meets their desired results, many, if not most, are not. Studying race and its effect in our society is not a new or "woke" phenomenon. Determining how race affects outcomes in America is not a trivial matter. Racial bias in America does not always present itself in a blatant way. Bias is often manifest in subtle ways. No matter how subtle, the effect is real for those on the wrong end of it. Conversely, for those whose knee-jerk reaction is to dismiss any hint of bias or racism, the mere mention of the reality of their existence is offensive and inviting of derision.
|
|
|
Post by julyanmorley on Aug 30, 2021 16:08:51 GMT -5
|
|