AJPS is losing all credibility.
with whom? The three anonymous high school dropouts who spend their lives in their basements screeching support for “karI’s” ranting?
Hardly of any consequence.
Enos' refusal to correct results he knows are incorrect is research misconduct
-
I don't understand why people are taking sides on this. Two things can be true simultaneously. Karl can be a failure and the Enos paper might merit retraction at the same time.
All sensible people know this. There is only one we/irdo Enos defender trying to make it all about Karl. He resembles a Kremlin propagandist.
-
I don't understand why people are taking sides on this. Two things can be true simultaneously. Karl can be a failure and the Enos paper might merit retraction at the same time.
Just because two sides can be true doesn't mean they are.
The fact is, "Karl" is an insane, juveniIe troIl who has a long history of shjtposting racjst lncel anti-intellectual MAGA spewage.
The real motivation behind all of this: Karl wants to attract more paid subscribers to his MAGA-larping blog so he can earn more than $200/year to subsidize his monthly check from Welfare Canada.
-
I don't understand why people are taking sides on this. Two things can be true simultaneously. Karl can be a failure and the Enos paper might merit retraction at the same time.
Just because two sides can be true doesn't mean they are.
The fact is, "Karl" is an insane, juveniIe troIl who has a long history of shjtposting racjst lncel anti-intellectual MAGA spewage.
The real motivation behind all of this: Karl wants to attract more paid subscribers to his MAGA-larping blog so he can earn more than $200/year to subsidize his monthly check from Welfare Canada.Wow you're dropping some real bombshells here. Karl is stirring up drama because he wants more subscribers?! You're guaranteed to get tenure with that level of reasoning.
I don't care what his reasons are. If the Enos paper is wrong then it's wrong.
-
Just because two sides can be true doesn't mean they are.
The fact is, "Karl" is an insane, juveniIe troIl who has a long history of shjtposting racjst lncel anti-intellectual MAGA spewage.
The real motivation behind all of this: Karl wants to attract more paid subscribers to his MAGA-larping blog so he can earn more than $200/year to subsidize his monthly check from Welfare Canada.
Mostly right, except it misses Karl's primary objective, which is to generate some but of culture-war red-meat that his boyfriend Greg Re finds sufficiently compelling to encourage Tµcker to give it a 5s shoutout.
Karl thinks these Tucker shout-outs increase his street cred with the American MAGA crowd whom he's seeking to monetize with subscriptions to his blog. He might be right, though he needs to find a more sustainable we to eek out a living if he ever wants to get off of welfare.
-
Could the missing data be related to matching (e.g., he threw out any precincts in which there we no matches)?
No, they could not. This has already been addressed in another thread. Hard to understand why some people come up with theories in his defense without actually going through his replication files and reading the report.
tl;dr: deletion of hundreds of precincts occurs BEFORE matching.
The relevant file to look at in this case is vote_choice.R in Enos' replication archive. He starts with two csv data files for votechoice, THEN performs EI (for which he would needed the whole dataset), then chooses a distance (1000 m from demlished projects) to divide observations into treatment (d<1000) and control (d>=1000). He then matches treated and control units by some variables. Further deletions occur.
But the important thing here is that hundreds of unfavorable observations never made it into his EI analyses or had a chance to be included in the matching, because instead of starting with the actual official data he started with an already pre-processed dataset. No code to show how that initial dataset was obtained in his replication archive. No reasons for deleting precincts from the outset given in the article.
-
Could the missing data be related to matching (e.g., he threw out any precincts in which there we no matches)?
No, they could not. This has already been addressed in another thread. Hard to understand why some people come up with theories in his defense without actually going through his replication files and reading the report.
tl;dr: deletion of hundreds of precincts occurs BEFORE matching.
The relevant file to look at in this case is vote_choice.R in Enos' replication archive. He starts with two csv data files for votechoice, THEN performs EI (for which he would needed the whole dataset), then chooses a distance (1000 m from demlished projects) to divide observations into treatment (d<1000) and control (d>=1000). He then matches treated and control units by some variables. Further deletions occur.
But the important thing here is that hundreds of unfavorable observations never made it into his EI analyses or had a chance to be included in the matching, because instead of starting with the actual official data he started with an already pre-processed dataset. No code to show how that initial dataset was obtained in his replication archive. No reasons for deleting precincts from the outset given in the article.Right. So he at first deleted a bunch of precincts because it was immediately obvious they should be excluded . Then he ran his analysis which excluded the less obvious ones that should be excluded by his criteria.
Sorry bud. This isn’t evidence of anything except someone who is so hellbent on believing his own accusations of fraud that he can’t see it’s weakness .
-
Right. So he at first deleted a bunch of precincts because it was immediately obvious they should be excluded .
Why was it "immediately obvious" they should be excluded? Again, no reasons given in the article, no code showing the criteria for the deletions. So what was the reason?
-
Could the missing data be related to matching (e.g., he threw out any precincts in which there we no matches)?
No, they could not. This has already been addressed in another thread. Hard to understand why some people come up with theories in his defense without actually going through his replication files and reading the report.
tl;dr: deletion of hundreds of precincts occurs BEFORE matching.
The relevant file to look at in this case is vote_choice.R in Enos' replication archive. He starts with two csv data files for votechoice, THEN performs EI (for which he would needed the whole dataset), then chooses a distance (1000 m from demlished projects) to divide observations into treatment (d<1000) and control (d>=1000). He then matches treated and control units by some variables. Further deletions occur.
But the important thing here is that hundreds of unfavorable observations never made it into his EI analyses or had a chance to be included in the matching, because instead of starting with the actual official data he started with an already pre-processed dataset. No code to show how that initial dataset was obtained in his replication archive. No reasons for deleting precincts from the outset given in the article.Right. So he at first deleted a bunch of precincts because it was immediately obvious they should be excluded . Then he ran his analysis which excluded the less obvious ones that should be excluded by his criteria.
Sorry bud. This isn’t evidence of anything except someone who is so hellbent on believing his own accusations of fraud that he can’t see it’s weakness .Did you read the 2018 paper that wreks Enos?
-
The most benign explanation is that Enos made a long series of severe data management errors. The onus is on him to demonstrate that they were honest errors. Further, being aware of the errors and refusing to publish a corrigendum is research misconduct. He is knowingly allowing incorrect results to remain in print under his name.
-
⣿⣿⣿⣿⣿⣿⡿⠿⠿⢿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿
⣿⣿⣿⣿⣿⣿⣿ DC ⢽⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿
⣿⣿⣿⣿⣿⠟⢰⣾⢛⣃⣾⣿⣿⣿⣿⠟⠋⠙⠛⢿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿
⣿⣿⣿⣿⠡⣶⡾⢿⡄⠻⣿⣿⣿⣿⡟Burnet ⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿
⣿⣿⣿⣿⠄⢸⣇⢸⡇⠱⣌⣛⣛⡛⣉⣤⣤⡤⠤⠭⠭⠉⠻⠿⠿⠿⢿⣿⣿⣿
⣿⣿⣿⣿⡆⡄⢻⡀⠳⣷⣶⣷⠂⣰⣿⣿⡀⣶⠄⠄⠄⠄⠄⠄⠄⠄⠄⢸⣿⣿
⣿⣿⣿⡿⣡⣿⣦⣌⠢⡙⠋⠄⢺⡿⠿⣩⣴⣿⠠⣤⣤⣤⣤⣤⣤⣤⣤⢀⣿⣿
⣿⣿⣿⡇⢿⣿⣿⡏⢠⣄⣙⠂⢈⣴⣾⣿⣿⣿⢸⣿⣿⣿⣿⣿⣿⣿⣿⢸⣿⣿
⣿⣿⣿⣿⣦⠻⣿⣷⠈⢿⣿⡐⣿⣿⣿⣿⣿⣿⢸⣿⣿⣿⣿⣿⣿⣿⣿⢸⣿⣿
⣿⣿⣿⣿⣿⣆⠹⣿⡇⠈⢻⣧⢻⣿⣿⣿⣿⣿⢸⣿⣿⣿⣿⣿⣿⣿⣿⢸⣿⣿
⣿⣿⣿⣿⣿⠟⣠⡿⢁⣾⠄⣿⢸⣿⣿⣿⣿⣿⢸⣿⣿⣿⣿⣿⣿⣿⣿⢸⣿⣿
⣿⣿⣿⣿⡏⢸⡿⢁⣾⣿⣶⡇⣿⣿⣿⣿⣿⣿⢸⣿⣿⣿⣿⣿⣿⣿⣿⢸⣿⣿
⣿⣿⣿⣿⡇⡾⠄⠬⠍⡙⠄⠑⠿⠿⣿⣿⣿⣿⢸⣿⣿⣿⣿⣿⣿⣿⣿⢸⣿⣿
⣿⣿⣿⣿⡐⠧⣜⡻⢿⣏⡘⠂⠄⠐⣿⣿⣿⣿⠸⢛⣛⣛⣛⣛⣛⣛⠻⢸⣿⣿
⣿⣿⣿⣿⣿⣷⣶⣶⣿⣿⣿⣿⣿⣿⣿⣿⣿⣿⣶⣾⣿⣿⣿⣿⣿⣿⣷⣾⣿ -
"The destruction of research records, absence of research records, or respondent’s failure to provide research records adequately documenting the questioned research is evidence of research misconduct where the institution establishes by a preponderance of the evidence that the respondent intentionally, knowingly, or recklessly had research records and destroyed them, had the opportunity to maintain the records but did not do so, or maintained the records and failed to produce them in a timely manner and that the respondent’s conduct constitutes a significant departure from accepted practices of the relevant research community...
A respondent has the burden of proving, by a preponderance of the evidence, any and all affirmative defenses raised (such as honest error)."
-
Yes, but whatever happened to Brunet's aspirations of becoming a "bonified" full-stack developer?
https://www.peopleperhour.com/freelancer/christopher-brunet-aspiring-full-stack-developer-yqwvjzj
https://research.fas.harvard.edu/policies/procedures-responding-allegations-misconduct-research