Matt Barnum and Alex Zimmerman of Chalkbeat report that a new study of Bloomberg’s stop-and-frisk policy found that it increased the dropout rate, especially for black boys, who were most likely to be stopped and frisked.
Matt Barnum and Alex Zimmerman of Chalkbeat report that a new study of Bloomberg’s stop-and-frisk policy found that it increased the dropout rate, especially for black boys, who were most likely to be stopped and frisked.
The Chalkbeat article links to the original paper, which says across the top “Preliminary Draft. Please do not Cite or Distribute.” Chalkbeat has done both. I know from past experience that the readers of this blog are extremely skeptical of any research that has not gone through a peer review publishing process, so I am sure they will not assume that the authors of this article are correct.
Bloomberg is proof of what Elizabeth Warren has been trying to get through people’s thick heads: our politics are for sale.
And the DNC, which changed its own rules so that Bloomberg could participate in tonight’s debate, will do anything to ensure that a candidate acceptable to the oligarchical donor class wins the nomination.
Note to DNC: “Democracy” means “rule by the people,” not rule by the wealthy people.
Have either Bernie Sanders or Elizabeth Warren gone on record objecting to Bloomberg being in the debate tonight? How do you know that the DNC didn’t include the surging Bloomberg so that he FINALLY has to answer directly to his critics?
I really don’t understand why supposedly the very same supposedly corrupt and terrible DNC that supposedly spent all of 2019 working corruptly to insure Biden won the primary and then worked corruptly to make sure Buttigeig won the primary supposedly now has decided to work AGAINST those two candidates and is suddenly working secretly to insure that the NEWEST candidate of the corrupt oligarchs who every single person at the DNC supposedly takes their marching orders wins the primary.
So far, every one of the candidates that the supposedly corrupt people at the DNC have been illegally and improperly aiding and abetting has done poorly. Maybe they are just bad at “helping” their preferred candidate?
The person getting shafted has been Elizabeth Warren, who might as well have dropped out of the race given how she is being treated by the oligarch-run media. Not Bernie Sanders.
If you ask me, there is far more evidence that the DNC secretly wants Bernie Sanders to be the candidate since each candidate the DNC is supposedly acting secretly to promote starts losing support.
There has been one nearly all white caucus and one nearly all white primary and already the DNC is throwing Biden under the bus by promoting the candidate most likely to take votes from him and not Bernie? Seems they are putting their feet on the scales for Bernie. At least I would certainly have sympathy for any Biden supporter who made that case as it has a lot more credibility than the DNC acting against Bernie’s interests when the person the DNC is screwing over is Joe Biden.
I’m supporting Bernie but I really object to conspiracy theories that make no logical sense except to help the right wing Trumpers encourage everyone to vote AGAINST the supposedly “corrupt” Dems.
Correlation does not imply causality. This effect was conveniently skimmed over on the article. What we’re numbers before stop and frisk was enforced? Also the graphs are misleading: the scale of the bars look larger than the percentages they represent. A difference of .5% is made to look enormous. How was “statistically significant” determined? Bad stats shouldn’t support an argument.
Dawn,
A coefficient is statistically significant if it sufficiently unlikely that the true value of the coefficient in the population is zero, in this case that it is unlikely that stop and frisk had no impact on the school dropout rate. It is used in all statistical analysis, and is not “bad stats”.
If you want to learn about the technicalities of determining whether an estimate is statistically significant google will show you many places that offer written and video explanations.
I don’t think that you need worry that any regular reader of this blog will take this paper seriously. It has not been peer reviewed, and in my experience, any paper that I brought up that was not peer reviewed was immediately deemed unreliable by other commenters. In addition, this paper relied on an estimation model “in the spirit of Chetty, Friedman, and Rockoff (2014).” It is the opinion of people on this blog that Chetty, Friedman, and Rockoff (2014) as methodologically flawed and complete trash, so they must view any other papers that use the same methodology must be complete trash as well.
I thought VAManujan’s VAM paper was actually pee-er reviewed like most economic papers.
I learn something new every day.
Thanks!
Speaking of statistical significance
Chetty et al are obviously confused about statistical significance. If the signal does not rise above the noise at age 30 for their original dataset, according to Chetty et al’s own admission (“one cannot reject the hypothesis that the effect [at age 30] is zero”), how can they claim (in the very same paragraph!) that “impact is actually larger” [at age 30 than at 28]?
https://nepc.colorado.edu/thinktank/review-measuring-impact-of-teachers
They can’t. That’s just total nonsense.*
Chetty needs to take a very basic stats course (assuming he is capable of understanding the math)
And anyone who supports such nonsense has absolutely no clue about statistics.
Just to be crystal clear: that a hypothesized effect (or impact as Chetty calls it) lacks statistical significance ( referred to by Chetty et al as “statistically insignificant”) does NOT mean the effect does not exist. That much Chetty et al get right. But it also does NOT mean it DOES exist! One can not legitimately draw a conclusion either way. Because one can not even conclude from the data used in the analysis that the effect is real, one is certainly NOT warranted in claiming that the impact is larger at age 30 than at age 28 as Chetty et Al do.
In the words of Chetty et al:
”
The estimated impact at age 30 is $2,058, larger than the estimated impact at age 28 ($1,815).”
“The standard error of the estimate at age 30 is $1,953, and hence is statistically insignificant (i.e., one cannot reject the hypothesis that the effect is zero. However,this does not mean that there is no effect at age 30; **rather, it means that one has insufficient data to measure earnings impacts accurately at age 30. **”
// End of Chetty et Al quotes
That last bolded sentence is not necessarily true.
What it actually means is just what Chetty et Al said in the immediately preceding sentence!! No more. No less: one can not conclude the earnings impacts at age 30 are different from zero!
The underlying reason for the lack of statistical significance — eg, insufficient data — does not affect the conclusions one is warranted in drawing (is, the “meaning”). It MIGHt be that the effect lacks statistical significance simply because the data set is too small. On the other hand, it MIGHT be because the effect is not even real (does not even exist). But the MOST one can say based on the data used to do the analysis is ” one can not conclude the earnings impact at age 30 is different from zero.” Period. That’s ALL one can legitimately conclude.
If one can not conclude from a given data set that a hypothesized effect is even different from zero at age 30, one can most certainly NOT claim “the estimated impact at age 30 …is larger than the estimated impact at age 28″ but Chetty et Al apparently believe that a value that lies in an interval that includes zero (for age 30) is somehow necessarily larger than a positive value (for age 28).
The brilliance of VAManujan on display for the whole world to be awed by.
I could go on and on about Chetty’s BS stats, but I would be wasting my time because anyone who can claim two contradictory things in a single paragraph as Chetty et Al did is utterly clueless about statistics.
And by the way, don’t try to change the focus to Adler because I am quoting Chetty himself here. Hoisted by his own petard!
SDP, since I write about Chetty in my new book, I can’t help asking, have you read it?
🙂
Not yet, Diane but I look forward to doing so.
Let me guess: Chetty is David, right?
Dawn,
I hope you see now that your worries are unfounded. Researchers who use these methods are routinely ridiculed by the most prolific commenters on this blog. No orthodox reader will take this as evidence that intensified stop and frisk policies cause more students to drop out of high school. In fact, given that the authors are at Harvard, another group that is routinely ridiculed on this blog, it is likely that the orthodox reader will take this paper as evidence that the plutocrats that control Harvard are trying to smear Bloomberg because he is interfering in their plans for the election.
Dawn
I hope you can now see that Teaching Economist doesn’t address any of the statistical argument , but instead completely avoids it.
And I don’t recall saying anything at all about “stop and frisk.” I specifically addressed economic “pee-er review” and “statistical significance (and Chetty’s ignorance of what it), which TE (not I) had brought up.
But maybe TE does not know the difference between the words “statistical significance” and “stop and frisk”. They do kind of sound the same.
I think we can agree that the stop and frisk –> higher drop out rate argument is bogus. It should not have been posted on the blog thereby giving it merit.
Sent from Yahoo Mail on Android
Who agrees? Do you have different data?
The interpretation of the data is flawed. Have you not read our comments? This study was not peer reviewed, for one.
Dawn
You are right that correlation does not mean or even imply causation.
Unfortunately, pretty much all studies in the social sciences suffer from this serious (if not fatal) “issue” (Raj Chetty’s study of the impact of “effective” teachers among them).
I honestly don’t know anything about the stop and frisk study in question so would not draw ANY conclusions about it — not even if it had been peer reviewed (which is why I did not do so above). Unfortunately, peer review is “not always” (rarely?) what it is cracked up to be. In many cases, it means little or nothing (and TE would never admit that his own field of economics is actually rife with peer reviewed rubbish).
Having said that, it is also clear to me (and anyone else with a knowledge of statistics) from what you wrote above that YOU are not sufficiently knowledgeable (eg, about statistics) to be drawing the conclusion that the study in question is “bogus” and that “the interpretation of the data is flawed”.
It works both ways. One can not claim to know that a study is legitimate OR bogus without having sufficient knowledge to make that assessment. Unfortunately, that is also true in the case of studies that actually have been peer reviewed.
You know nothing about me or my knowledge of statistics. I do not claim to be an expert, but I am not ignorant either. Take your arrogance and shove it up your ____.
Dawn, we don’t use that language here. You have been warned.
It won’t happen again. Not interested in reading a blog where questionable “research” is presented as fact, not even vetted before posting. And a response of “boring” about an article discussing the need to understand statistics is very sad. Very mathphobic. Reading it, you would understand why this stop and frisk study should not have been given attention.
Chalkbeat is a reputable publication. I don’t agree with everything they publish but I respect their integrity.
I don’t know whether stop and frisk increased the dropout rate. I do know that it’s wrong, deeply demoralizing and racist to stop young black males, throw them up against a car hood and search them. The Police Department’s own statistics said this was highly ineffective. The overwhelming majority of boys who were treated like criminals carried no weapons. The practice was sheer racial profiling and was found to be unconstitutional by the courts. Since the NYPD stopped the practice, major felonies have declined.
Dawn
You asked above
“How was “statistically significant” determined?” And said” Bad stats shouldn’t support an argument.”
TE pointed out the standard meaning of “statistical significance” and your very own words indicate that you did not know this.
So yes, we do know something about your knowledge of statistics from what you wrote above. It is close to nil.
There is nothing wrong with not knowing perse, but if you don’t know what “statistically significant” means, you are certainly not in any position to decide whether a study is legitimate OR bogus.
Pointing that out has nothing to do with arrogance but with reality.
Dr. Ravitch,
I do not believe that Dawn is relying on new data to reject this paper’s conclusion. Instead Dawn is relying on the collective wisdom of you and the posters here on this blog. The collective wisdom of this blog is that Chetty’s methodology is garbage, so the conclusions of the paper are also garbage. The paper in your post uses Chetty’s methodology, so the methodology in the paper is also garbage, so the conclusions of the paper are also garbage. It would be illogical to reject one yet accept the other.
Boring.
TE
You still haven’t addressed the points I raised above about Chetty’s very own claims (in Chetty’s own words).
In other words, you didn’t (won’t?) address the actual substance of a criticism of Chetty.
Instead you fall back on your hackneyed and meaningless “collective wisdom of this blog” verbiage.
Why is that?
Still only crickets from TE on substance.
I am bookmarking this exchange and intend to link to it EVERY time Teaching Economist utters the words “collective wisdom of this blog”, ” orthodoxy of this blog” or something similar.
I’m curious if TE uses such “powerful” arguments when he peer reviews the work of other economists — arguments like “The collective wisdom of these authors is that X’s methodology is garbage, so the conclusions of the paper are also garbage. ”
Powerful stuff indeed.
Since TE is so contemptuous of this blog, I wonder why he continues to read it. He is always negative and often condescending.
While i do lament the narrowing of opinion that I have seen over the nearly decade that I have commented on this blog, I do see it as important because many people with diverse opinions read it, though of course they do not dare to comment. I hope that my comments can help this blog be something more than a conformation bias machine.
Mostly you are not informative, just negative and condescending
The crickets are deafening.
And a note to the crickets: it’s “confirmation bias” , not “conformation bias”
There is a technical name for what we are witnessing here: cricket projection — when crickets project their own shortcomings on others.
William Black talks about the central problem of economics
“Economics could be a Science if More Economists were Scientists”
https://neweconomicperspectives.org/2013/10/economics-science-economists-scientists.html
Thanks for the article.
I could not watch the debate last night but from what I heard Elizabeth Warren beat Bloomberg hands down, savaged him. The CNN moderator said she could, from what she did last night, do the same with “Trump – as if it would make a difference to his fans.