LOL statistics? She doesn't even understand algebra. She once seemed to think that b/SE = t implies b = t*SE! Can you imagine what
Oh wait
Dang
Her claim was that if b/SE ~ t then b ~ t*SE.
LOL statistics? She doesn't even understand algebra. She once seemed to think that b/SE = t implies b = t*SE! Can you imagine what
Oh wait
Dang
OH right. How could I be so dumb?
I mean, it's correct either way, but you're right.
So the beta coefficient follows the t*SE distribution?
Her claim was that if b/SE ~ t then b ~ t*SE.
LOL statistics? She doesn't even understand algebra. She once seemed to think that b/SE = t implies b = t*SE! Can you imagine what
Oh wait
DangOH right. How could I be so dumb?
I mean, it's correct either way, but you're right.
Hint question for you, CG:
How do you know in the first place that b/SE follows a t distribution (so you can rearrange terms and get that scaled t, as you claim)? What assumptions about the distribution of the beta coefficient do you need to make in order to arrive at that result? Think hard about it. Maybe you'll finally understand why you're so wrong it's ridiculous.
In 4 out of the 5 articles she had pre-tenure (essentially all her work other than the logically inconsistent EI article that was thoroughly debunked at the 2002 Polmeth), CG's only contribution to the literature she borrows heavily from is to test an interaction effect of neighborhood (based on a book by someone else). In her analyses she shows an utter lack of knowledge on how to test interaction effects, making her conclusions unreliable. To defend her articles, she claimed that her test statistic for the interaction effect was actually the coefficient of the neighborhood variable. When pointed out that coefficients can't act as test statistics (with detailed explanations by several people), she started inventing this far-fetched theory on how coefficients can be test statistics, which is full of contradictions and ridiculous claims, and only further confirmed her complete ignorance of basic statistics.
Where does she make this mistake? In what article?
I have no dog in this fight, but I don't understand why people are even debating the statistics of it all. The claim that beta is a test statistic because t is a test statistic and beta can be written as a function of t is just a logically invalid argument. It would be like saying t is a regression coefficient because it can be written as a function of beta.
Yes, it's logically invalid. But CG insists she can use the coefficient as a test statistic because it follows the t*se distribution. The only other person who defended her is some dimwitted King student (possibly Matty B). No one else considers this debatable.
I have no dog in this fight, but I don't understand why people are even debating the statistics of it all. The claim that beta is a test statistic because t is a test statistic and beta can be written as a function of t is just a logically invalid argument. It would be like saying t is a regression coefficient because it can be written as a function of beta.
I have no dog in this fight, but I don't understand why people are even debating the statistics of it all. The claim that beta is a test statistic because t is a test statistic and beta can be written as a function of t is just a logically invalid argument. It would be like saying t is a regression coefficient because it can be written as a function of beta.
You're the one using terrible logic. The definition of test statistic is sufficiently broad that a coefficient or sample mean can qualify. Not only was this proved a hundred times in the previous thread, it was also an example problem in Casella & Berger, to the endless embarrassment of the stats buffoons.
Yes, it's logically invalid. But CG insists she can use the coefficient as a test statistic because it follows the t*se distribution. The only other person who defended her is some dimwitted King student (possibly Matty B). No one else considers this debatable.
I have no dog in this fight, but I don't understand why people are even debating the statistics of it all. The claim that beta is a test statistic because t is a test statistic and beta can be written as a function of t is just a logically invalid argument. It would be like saying t is a regression coefficient because it can be written as a function of beta.
Total lie. Most of the stats bros admitted they were wrong, especially when the kill shot evidence proving me right was found in Casella & Berger. The fact that you're too dim to get it is not a defense.
Again, if you weren't embarrassed before, then why bring up a silly thread from weeks ago? Do you jump awake at night remembering how you got destroyed? Bet you do!
In 4 out of the 5 articles she had pre-tenure (essentially all her work other than the logically inconsistent EI article that was thoroughly debunked at the 2002 Polmeth), CG's only contribution to the literature she borrows heavily from is to test an interaction effect of neighborhood (based on a book by someone else). In her analyses she shows an utter lack of knowledge on how to test interaction effects, making her conclusions unreliable. To defend her articles, she claimed that her test statistic for the interaction effect was actually the coefficient of the neighborhood variable. When pointed out that coefficients can't act as test statistics (with detailed explanations by several people), she started inventing this far-fetched theory on how coefficients can be test statistics, which is full of contradictions and ridiculous claims, and only further confirmed her complete ignorance of basic statistics.
Where does she make this mistake? In what article?
Every single word of this is a lie. The racist stats blobs are angry that CG published 5 solo top-3s and they can't get a job as a Wal-Mart greeter. Makes them feel inferior. Because they are!