Saturday, March 26, 2016

"Politically correct" facts that Paul Joseph Watson Daren't Talk About

This post is a response to this video by British Infowars-affiliated conspiracy theorist Paul Joseph Watson, as well as this article Watson wrote for Infowars.com and which he uses as a source in his video. 
First, I will respond to his point that black people are "more likely to be victims of violent confrontations with police officers than whites because they commit more violent crimes than whites per capita." Now this hypothesis can be tested, by seeing not just if black people are more likely to be shot by police than white people, as even Watson acknowledges is the case, but also by seeing if there is a relationship between crime rates in an area of the US, whether by black people or overall, and racial bias in police shootings in that area. As it happens, a study has been done testing this hypothesis, and it found that:
  • "the probability of being {black, unarmed, and shot by police} is about 3.49 times the probability of being {white, unarmed, and shot by police} on average." Alright, so Watson seems to be right so far, but it's the next finding that seems to contradict his narrative:
  • "There is no relationship between county-level racial bias in police shootings and crime rates (even race-specific crime rates), meaning that the racial bias observed in police shootings in this data set is not explainable as a response to local-level crime rates." Well that would seem to be inconvenient for someone who wants to portray a disparity as the fault of black people. I'm sure Watson will pull some other factor out of his ass to claim that this disparity isn't the result of systemic racism, b/c he's dug himself too deep into this "politically incorrect" hole to climb out now.
Next I will address Watson's claims about blacks committing more crime than whites relative to their share of the population. His claims in this regard are as follows: 
"Despite making up just 13% of the population, blacks commit around half of homicides in the United States. DOJ statistics show that between 1980 and 2008, blacks committed 52% of homicides, compared to 45% of homicides committed by whites.
More up to date FBI statistics tell a similar story. In 2013, black criminals carried out 38% of murders, compared to 31.1% for whites, again despite the fact that there are five times more white people in the U.S.
From 2011 to 2013, 38.5 per cent of people arrested for murder, manslaughter, rape, robbery, and aggravated assault were black. This figure is three times higher than the 13% black population figure. When you account for the fact that black males aged 15-34, who account for around 3% of the population, are responsible for the vast majority of these crimes, the figures are even more staggering."

This is all true, but misleading: for one thing, black people are much more likely than white people to be the victims of murder. From 1980 to 2008, the homicide victimization rate was 6x higher for blacks than for whites. Why? Well, there is some evidence that racial disparities in poverty and the prevalence of families headed by women are a factor. However, this does not explain the entire Black-White homicide gap. Other factors proposed to explain at least part of it include that, according to one theory, "Individuals belonging to groups perceived to have low marginal penalties for killing will be feared, and they will accordingly also be killed with greater frequency. By the same token, individuals whose own deaths are less likely to be investigated and prosecuted vigorously will fear being killed, and may therefore be induced to kill preemptively. This means that social groups with high victimization rates will also have high murder rates, even if there is no racial segregation in social interactions." This theory predicts that more segregated neighborhoods will have higher racial disparities in murder rates, a theory which a number of studies have found evidence for.



Moving on:
"Despite the fact that black people commit an equal or greater number of violent crimes than whites, whites are almost TWICE as likely to be killed by police officers.
According to data from the Centers for Disease Control, between 1999 and 2011, 2,151 whites died as a result of being shot by police compared to 1,130 blacks."
The word that renders the above claims bogus is "likely". This implies that a higher percentage of white people were killed by police than black people, which, given that there are about 5x more whites than blacks in the U.S., is not the case at all. Even though almost 2x as many whites were killed by police as blacks during the above time period, when you account for their share in the general population, blacks were 3x more likely than whites to be victims of such killings. This is comparable to the 3.49x figure for being unarmed I cited above. 
Now Watson acknowledges that blacks are overrepresented among police shooting victims, but argues that they are underrepresented with respect to perpetrators of violent crime. Yet as I noted above, there is no association between race-specific crime rates in an area and the racial disparity in police shootings in that area, so this explanation doesn't hold up. Another recent study found that with regard to police shootings, "Race does matter but only insofar as it increases the level of firearm violence and, even then, only to a point." 

Watson's next argument is that "Despite being outnumbered by whites five to one, blacks commit eight times more crimes against whites than vice-versa, according to FBI statistics from 2007. A black male is 40 times as likely to assault a white person as the reverse. These figures also show that interracial rape is almost exclusively black on white."
I will tackle the first of these claims first: that there are way more whites than blacks but for some strange reason blacks commit more crimes against whites than vice versa. Yet this is misleading because it is precisely because there are more whites in the U.S. that we expect more white victims of interracial crime than black ones. As Sampson and Lauritsen put it, "In felony homicides,as in robberies, black offenders are more likely to victimize whites than white offenders are to victimize blacks.Yet this is still what we should expect because blacks are the smaller group and have more chances to interact with whites. Variations in the relative sizes of the black and white populations thus explain the patterning of interracial violence." (see page 329)
As for the interracial rape claim, this comes from a column by Pat Buchanan in which he says that "Interracial rape is almost exclusively black on white — with 14,000 assaults on white women by African Americans in 2007. Not one case of a white sexual assault on a black female was found in the FBI study." The study in question was not actually conducted by the FBI, but by the Bureau of Justice Statistics (BJS). As Media Matters has pointed out, "The BJS study is not a tally of every single violent crime committed in 2007, but, rather, an estimate based on a nationwide household survey. The study makes clear that its estimates for the overall number of blacks who were raped and for the number of whites who were raped by blacks are both based on interviews with 10 or fewer respondents.

Michael Rand, a statistician who worked on the study, has said that Buchanan is interpreting it incorrectly, noting that, according to Media Matters, "these small sample sizes [of 10 or fewer respondents]-- which result from the fact that blacks make up less than 13 percent of the overall population and from the fact that rape is a relatively infrequent crime -- prevent the BJS from accurately calculating margins of error" and that for this reason, the BJS data cited by Buchanan "are not reliable as exact point estimates of the actual levels of such crime."



Wednesday, March 9, 2016

A critical analysis of "Racism in the United States: by the Numbers" and criticisms thereof

In this post I will be looking at the issue of whether systemic racism against black people exists in the United States, as claimed by John Green in his video "Racism in the United States: By the Numbers", which was released in December 2014. I will also address both Green's claims and those made by Sean Last on the website "therightstuff.biz" in this article, (update: link is dead, replaced with archive) in which he claims to debunk every claim Green makes in his video. 

First: longer prison sentences for black men who were convicted of similar crimes as white men. To be specific, Green cites a Sentencing Commission report from 2013 that found that black men's prison sentences were "nearly 20% longer than those of white men for similar crimes in recent years." The same report stated, as Green did, that the racial sentencing gap has increased markedly in the past decade, or, more precisely, since 2005, when the Supreme Court's US v. Booker decision allowed judges to issue sentences outside of a range of guidelines set by the Commission. Yet, a study published the same year found no evidence that the Booker decision caused the racial disparity to increase or even that it had increased since then. But the question remains: do black people get longer prison sentences than white people, and if so, what is the cause of this difference? The Sentencing Commission showed that the answer to the first question is yes, but there are many other factors that must be accounted for before concluding that this difference is due to discrimination. As Last notes, "“similar” crimes are not identical crimes. It may be that the average assault committed by a black man is significantly worse than the average assault committed by a white man. " Well, perhaps a more thorough analysis better controlling for non-race factors would suit you better, no? In that case, I would like to direct you to a study published 2 years ago that found that "... initial case and defendant characteristics, including arrest offense and criminal history, can explain most of the large raw racial disparity in federal sentences, but significant gaps remain. Across the distribution, blacks receive sentences that are almost 10 percent longer than those of comparable whites arrested for the same crimes." Last also cites a very strange and under-publicized study published in 2013 that found that differences in arrest and incarceration rates between black and white males can be "completely accounted for after including covariates for self-reported lifetime violence and IQ." But Last doesn't mention that in one of the few discussions of this study I could find anywhere on the Internet, Rebecca Coffey noted that "If the data in this research prove to be sound, its conclusions about the criminal justice system are still off the mark. One needs to examine the criminal justice system itself in order to pronounce about its integrity. The real matter at hand, though, is what if the data about White vs. African Americans hold water? They suggest that, more than white men, African American men engage in violence—and, on average, their IQs are lower. Ouch. But, as the saying goes, “context is everything.” Decades of research have associated poor education with low scores on intelligence tests. Another mountain of research has shown that a lifetime of being routinely denied access to common courtesies, to just returns on one’s labors, and to whatever pot of gold may be at the end of life’s big rainbow could drive a man to drink, drugs, and maybe even violence."


Green also cites data that concludes that there is little difference between the rate at which white people and black people use illegal drugs, but that black people are 3x more likely to be arrested for drug possession.  The figure he cited came from the 2011 National Survey on Drug Use and Health. The issue, according to critics, is that surveys are unreliable when it comes to racial differences in drug use because, as Last puts it, "As difficult as this is for some to accept, the truth is that blacks are more likely to lie about using drugs than whites are. This may sound like a harsh conclusion, but many studies done over several decades have shown that this is the case." Among these studies, according to Last, is one from 2005 that tested concordance between self-reported drug use and objectively measured test results. This study concluded that "the results replicate and extend a growing body of research suggesting that African Americans underreport substance use on surveys." However, there is more to the story. The same study notes that 
"The results showed that relative to African Americans, Whites had significantly larger odds of providing concordant responses; the point estimate for the odds ratio exceeds a value of 15. This large and significant effect is diminished by the entry of a single mediating variable in the model, SES [socioeconomic status]. With the addition of SES, the odds ratio contrasting Whites with African Americans is reduced to a nonsignificant point estimate value." That is, the racial difference may not be because the underreporters are black so much as because they are of a lower socioeconomic status than whites. 


As far as police searches, Green notes that black people are more likely to be stopped and searched by police, even though certain kinds of contraband are found more often in white people than in black people. As evidence he cites a 2008 article by Ian Ayres, a Yale Law School professor,  in which he describes a study he co-wrote looking at police stops in Los Angeles. Ayres summarizes his findings as follows: 



For every 10,000 residents, about 3,400 more black people are stopped than whites, and 360 more Latinos are stopped than whites. Stopped blacks are 127% more likely to be frisked -- and stopped Latinos are 43% more likely to be frisked -- than stopped whites.

 


Stopped blacks are 76% more likely to be searched, and stopped Latinos are 16% more
          likely to be searched than stopped whites.


Stopped blacks are 29% more likely to be arrested, and stopped Latinos are 32% more likely to be arrested than stopped whites.


Now consider this: Although stopped blacks were 127% more likely to be frisked than stopped whites, they were 42.3% less likely to be found with a weapon after they were frisked, 25% less likely to be found with drugs and 33% less likely to be found with other contraband. 
Green's source for NYC is this link, which provides links to Excel files on the number of stop and frisks in NYC by race (and other factors that are irrelevant here). I downloaded the one for the last quarter of last year (the most recent available) and added up the # of stops by race. It looks like of the 3,793 people stopped in NYC in the 4th quarter of last year, 1,969, or about 51% were black, which is much higher than the share of the city's population that is black (22.8% as of 2013). 
Last's response is as follows: "The fundamental problem with this data is simple: there are lots of reasons why blacks might be getting stopped more than whites that have nothing to do with racism. The use of this statistic to prove racism assumes that none of these alternatives are true. For instance, black people are far less likely than white people to own a car. So this argument assumes that walking around town doesn’t increase your chances of getting stopped. It also assumes that blacks don’t spend more time away from home than whites do. If you’re out of the house more, your chances of getting stopped by police will go up. It assumes that blacks don’t act more suspiciously than whites do. Blacks might be more likely to wear clothing associated with criminality, to loiter for long periods of time, etc. Green’s argument makes all of these assumptions but he provides evidence for none of them." 
Now there is one problem with this first argument. It's entirely speculative, which is ironic because Last is accusing Green of making assertions without evidence, except for the one part about how black people are less likely to own cars. So what? Ayres' study looked at both traffic and pedestrian stops. 
I became more optimistic when it came to the next part in Last's post, which reads: 
It’s also important to note that these data sets come from two large cities: LA and New York. Both of these cities are disproportionately black relative to the rest of the United States. Because of this, and the relatively high violent crime rate among blacks in general, most people that commit violent crimes in these cities are probably black. And so black people will be more likely to match descriptions of perpetrators than whites will. This will also lead to blacks getting stopped by police more often.
The problem is that, according to a 2007 study, even when you account for the rates at which different races commit crimes, blacks and Hispanics are still stopped at higher rates than whites. A more recent study backs this up, along with the finding from Ayres that blacks and Hispanics are less likely to have contraband, specifically weapons: "...stopped blacks and Hispanics were...less likely than similarly situated whites to possess weapons, indicative of racial bias in stop decisions."

Green next points out that of the 336 DNA exonerations of convicted criminals nationwide, 206 convicts were black.  This works out to about 61%, which is way higher than the 13.2% of the general population of the US who are black as of July 1, 2014. Last's response:

This is a pretty weak argument. There are lots of reasons for which blacks might get falsely convicted more often than whites that have nothing to do with racism. For example, according to project innocence [sic: he means the Innocence Project], the organization that compiled the data Green referenced, not being able to afford a good lawyer is the most common cause of a false conviction. Blacks are poorer, on average, than whites are. So they probably can’t afford good lawyers. But that is no proof of racism.
Edwin Grimsley, who actually works for the Innocence Project, takes a different approach in a 2012 article on their website. He argues that "Despite numerous studies depicting similar levels of participation by Whites, African-Americans, and Hispanics in non-violent crimes – notably drugs, weapon possession, and speeding - Bureau of Justice statistics show minorities are arrested and incarcerated at higher rates for these crimes." Why does this matter in this context? Because, as Grimsley continues, "Cumulative non-violent arrests can lead to future suspicion for violent crimes, as seen in the stories of those exonerated by DNA testing. A number of our clients who have been exonerated through DNA testing had their mugshot photos entered into a photo array lineup culminating in a misidentification. For example, Rickie Johnson of Louisiana had a mugshot on file because of a misdemeanor traffic violation. The victim identified his mugshot and he was later erroneously convicted of aggravated rape." Later, Grimsley notes: "nearly 75% of the DNA exoneration cases involve an eyewitness misidentification – approximately 42% of which are cross-racial misidentifications. Research studies have shown how victims or witnesses can have difficulty identifying facial characteristics of a dissimilar race. Also, the "blacks are poorer" thing undercuts Last's "racism doesn't exist" arguments, as I will discuss below.

Because minorities are more likely to be arrested as juveniles, false confessions and fictitious incriminating statements are more prone to occur. Juveniles are especially vulnerable to giving false confessions because they’re easier to manipulate."


So it seems that at least, the data is consistent with racial bias being at least a major contributor to the exceptionally high rates of false convictions among black people.

Although his arguments don't stand up to much scrutiny, Last is certainly good at inserting irrelevant invective and speculation, like this little nugget: 
Additionally, activist groups that get cases reexamined might be more likely to help black prisoners than white ones. This too could explain why blacks are more likely than whites to be exonerated by DNA evidence. One might argue that this is a form of racism, such groups would be discriminating against white people after all. But it’s obviously not the kind of racism that Green has in mind.
No evidence provided, of course, so it can be dismissed without evidence too. 

Green's next point is that black kids are more likely to be tried as adults--and sentenced to life in prison--than their white counterparts. His source--a 2012 article in the New York Times-- doesn't say much about these disparities, except to provide a link to back up the authors' claim that 84% of juveniles sentenced to life in prison for non-homicide charges were black in 2009. The link doesn't seem to say this, at least not in the abstract, but maybe it's somewhere in the full body of the article. There is, however, more evidence that people are more likely to recommend that a juvenile be transferred to adult court if the juvenile in question is black.


Last's response is to trot out the old (and true) talking point about how black adults commit more violent crime than do their white counterparts, arguing that for this reason, we should expect that "black juveniles [are] simply more likely than white juveniles to commit serious crimes or to have histories of crime." Maybe, but what if you were to look only at black and white juveniles who were convicted of murder, so they committed the same crime, and then see if there are racial differences between the black and white members of this group? As it happens, no surprise, this has been done, and 1/8 black juveniles convicted of murder will be sentenced to life in prison, as compared with 1/13 white kids convicted thereof. Also backing up claims of racial bias on this point is the fact that, according to a 2000 study, "African American, Latino and Asian American youth are significantly more likely to be transferred to adult court and sentenced to incarceration than white youths who 

commit comparable crimes. Compared to white youths, minority youths are 2.8 times more likely to be arrested for a violent crime, 6.2 times more likely to wind up in adult court, and 7 times more likely to be sent to prison by adult court." 

Another 2000 study showed that in California, "Hispanic youth were 6 times more likely, and African American youth were 12 times more likely than White youth to be transferred to adult court. Minority youth did have higher arrest rates, but examination of the arrest categories suggests little reason (such as greater number of violent crimes) for the racial disparity in terms of transfers to adult court."


Last also argues that black children mature faster than white ones do, citing multiple sources, the first of which is a 1967 study by Arthur Jensen. I looked everywhere in that paper for mentions of anything Last attributed to it (which is that "Even as infants blacks posses [sic] brain waves and skeletal features which suggest greater physical maturity.") and couldn't find anything that matches what he claims. But he then makes up for this deficit and cites a story about this study, which found that black people age faster biologically than white people do, and this story which says that black girls on average go through puberty .9 years earlier than do their white counterparts. I personally consider this less relevant than the actual criminal history or characteristics of the crime for which a given juvenile was sentenced, because this is what should, ideally, determine the way in which minorities are treated. 

I will close this section by quoting from a 1996 study on this subject: 
"Nonwhite youths referred for delinquent acts are more likely than comparable white youths to be recommended for petition to court, to be held in pre-adjudicatory detention, to be formally processed in juvenile court, and to receive the most formal or the most restrictive judicial dispositions."
Later, the authors state:
"Our qualitative findings support several interpretations. Intentional race discrimination does not appear to play a major role in accounting for racial disparities in processing. Although some officials whom we interviewed believed that some justice officials were motivated by prejudicial attitudes, few recounted specific instances of racially motivated actions. Without question, there are some justice officials who hold and act upon racially prejudicial attitudes. As long as race bias exists in the general culture, it would be surprising indeed if it did not operate through individuals in the juvenile justice system as well. However, we are not inclined to conclude that the disparities we observed are largely attributable to intentional race discrimination. Instead, we see much evidence of institutional racism. This is evident both in criteria for diversion and pre-trial release that focus on family support and cooperation, and in efforts to provide the economically disadvantaged with resources at state expense that the more affluent can purchase on their own. Obtaining these resources exacts a price in terms of adjudications of delinquency and sentences to confinement."

With respect to the job market, Green cites two studies showing that black people are less likely to get job callbacks than comparable white people--in one case, when their resumes were otherwise identical except for having black/white sounding names. The other study showed that black former inmates were less likely than white ones to get job callbacks. This latter study reported a surprisingly large race effect, as the author put it: "The effect of race was very large, equal to or greater than the effect of a criminal record. Only 14 percent of black men without criminal records were called back, a proportion equal to or less than even than the number of whites with a criminal background." 


Last's response begins with citing this ridiculously long report that says a lot about race, but after combing through it for a few minutes I found a quote that is directly relevant to what he is claiming, namely that "whites out-preformed blacks at every level of education." The quote in question is: "The average differences in prose, document, and quantitative proficiencies between White and Black adults are 49, 50, and 63 points, respectively. When level of education is taken into account, the average proficiency differences across the nine levels of education decrease to 36, 37, and 48 points, respectively. The remaining disparities in performance between White and Black adults may be the result of numerous factors. One plausible explanation is the variation in the quality of education available to these two populations. Differences in socioeconomic status are also likely to be a factor." Also worth pointing out is that the report cited by Last didn't distinguish between which kind of degree earned or where it was earned, whereas the study by 
Marianne Bertrand and Sendhil Mullainathan used resumes that were identical not just in the level of degree earned but regarding what kind of degree was earned from which university.

Now what about Last's next claim that "blacks with graduate degrees performed worse on tests of functional literacy than whites who had completed some college but hadn’t gotten a 2 year degree. In quantitative reasoning whites who had graduated high-school and had never even been to college did just as well as blacks who had 4 year degrees"? Looking at Figure 1.7 in the report, the score for blacks with graduate degrees for prose (298, there is no category called "functional literacy", contrary to Last's assertion) is indeed lower than that for whites in the some college group for this metric (302). Looking at the same chart for "
quantitative reasoning" shows that, as Last says, whites with high school diplomas only got a score of 279, while blacks with 4 year degrees got 280, which, though not exactly "just as well", is close enough for me. Nevertheless, the text bolded in the above paragraph undercuts Last's argument that racial differences in literacy within educational levels must reflect differences in inherent intelligence.


Green finishes by noting:

1. that schools attended by mostly minority students are less likely than those attended by mostly white students to offer Algebra 2 or chemistry. 
2. that, as the ACP pointed out in 2010, "Overwhelming evidence shows that racial and ethnic minorities are prone to poorer quality health care than white Americans, even when factors such as insurance status are controlled."
3. that white people are much more likely to inherit money and land than black people are, which is a major contributor to racial disparities in wealth in the United States.
Last ends by trying to dismiss the three lines of evidence above by saying that "all three [above] data points might be explained by whites having a lower time preference than blacks. This means that they are more willing to put off immediate gratification to pursue long term rewards. There are several sources of evidence, including replicated experimental evidence, which indicate that such differences exist which I have discussed elsewhereThis could lead to whites making better students, and thus causing their schools to offer better classes, to whites being more likely to purchase insurance and engage in healthy behavior, and to whites saving more and so leaving behind a greater inheritance. I’m not saying that this is necessarily the cause of group differences. The point is that I can easily think of explanations, for which there exist some evidence, that could account for these data points and which have nothing to do with racism. And so inferring racism from them is not a valid argument." The blog post he links to cites studies showing that black people are, among other things, less likely to save money than white people, but one study he cites in the other post also says that "Mexican[s] as much as African Americans have lower saving rates than Whites, even after controlling for income and socio-demographic factors. Whereas these differences for Mexicans are due to lower flows of money into assets, they are explained by lower capital gains for Blacks."

Aside from that, the differences in quality of healthcare received have nothing to do with blacks allegedly making worse lifestyle choices than whites. And I should also note that with respect to schools, "black and Latino students are significantly more likely to have teachers with less experience who aren't paid as much as their colleagues in other schools." Research also shows that regardless of poverty level, schools (at least in Pennsylvania) with more white students get more funding than do schools with more black students. So the evidence suggests that black students are treated unfairly ona large scale b/c of their race, which contradicts Last's arguments about how they are more impulsive and that's as valid an explanation as racism. Also, poverty seems to be a factor here, and the reason for that is discussed in this above-linked PDF about the racial wealth gap.

In short, Green is much more right here than Last.




No correlation apparent between strict state gun laws and homicide rates

I tried to see, as Eugene Volokh once did, whether strict state gun laws are associated with lower homicide rates. Overall homicide rates, not, as has been done before, gun homicides only. Below is a chart of my results. On the x-axis is how strong each state's gun laws are according to the Law Center to Prevent Gun Violence, and on the y-axis is the average homicide rate of all the states with each gun law score (of which there were 9 possible ones, ranging from A- to F, as detailed on the Law Center to Prevent Gun Violence's website linked above. I converted these scores into numbers from 1, corresponding to F, to 9, corresponding to A-.) The homicide rates were taken from the Death Penalty Information Center. As you can see there is no positive correlation at all between these two variables. Looking at them using Excel and typing in "=correl" in an adjacent cell, it seems like there is, if anything, a slight negative correlation here. This peculiar result seems to be in line with previous analyses on this subject, e.g. by Factcheck.org, although not with respect to suicide, where stricter laws are associated with lower overall rates.

Saturday, March 5, 2016

The myth of the University of Virginia "Medicaid is worse than nothing" study: or, why you should ignore Medicaid bashers.

In my last post on Medicaid expansion, I mentioned the claim, frequently made by opponents of Medicaid expansion, that Medicaid doesn't improve health outcomes. In this post I will look at one of the more negative versions of this claim: that it actually leads to worse health outcomes. Examples of this claim being made abound, such as this little nugget from the right-wing Galen Institute: "Mountains of clinical literature show that patients on Medicaid have, on average, poorer access to care and poorer health outcomes than those with no insurance at all. The largest national study, conducted by the University of Virginia, examined outcomes for 893,658 individuals undergoing major surgical operations from 2003 to 2007. It found that patients on Medicaid were 13 percent more likely to die in the hospital after surgery than those with no insurance, even when adjusting for age, gender, income, region, and health status. Medicaid patients were 97 percent more likely to die than those with private insurance."
There are innumerable instances of conservatives uttering variations of this claim, whether it's Florida senator Richard Corcoran, Virginia lieutenant gubernatorial candidate Pete Snyder, or Tim Phillips at Townhall.com. So what is this "University of Virginia study" and does it say what these politicians and pundits claim?
The answer to the first question is: "
Primary Payer Status Affects Mortality for Major Surgical Operations", published in 2010 in the Annals of Surgery. As for the second question, the study states that "Unadjusted mortality for Medicare (4.4%; odds ratio [OR], 3.51), Medicaid (3.7%; OR, 2.86), and Uninsured (3.2%; OR, 2.51) patient groups were higher compared to Private Insurance groups (1.3%, P < 0.001)." This basically means that people on Medicare, Medicaid, and no insurance, were, respectively, 3.51, 2.86, and 2.51 times more likely to die than were people on private insurance. Divide 2.86 by 2.51 and you get about 1.14 (if you round up), which means according to this study, people on Medicaid were 14% more likely to die than uninsured people. Right? 

Well, yes, but this doesn't mean Medicaid was totally or even partly the reason for this pattern. As Politifact has put it, "Medicaid recipients are the poorest, sickest and least educated group of patients. They are the least likely group to seek preventive health care. As a result, they are more likely to enter hospitals in dire conditions that require emergency surgery.
"Medicaid patients had the highest incidence of acquired immunodeficiency syndrome, depression, liver disease, neurologic disorders and psychoses," the study said. "Furthermore, Medicaid patients had the highest incidence of metastatic cancer." In other words, to quote Irving Kron, another co-author of the study, "The reality is it’s apples and oranges. The problem with Medicaid is there’s more emergencies, because they’re sicker than most people. … They wait for care and unfortunately, emergent patients don’t do as well as elective patients." So this confounding factor could account for the reason Medicaid patients did worse than uninsured ones, rather than Medicaid being an inferior system.
The researchers behind the study have also said that "uninsured patients have similar characteristics to Medicaid recipients and that it is "plausible" that both groups may suffer from a "system bias" that limits their access to private hospitals and top physicians." 
If you don't find these arguments compelling, consider that there are other possible confounders that may well explain a lot of the mortality differences found in this study: "Not all of the uninsured patients in the study were low-income people who would qualify for Medicaid if it is expanded. Thirty-one percent of them lived in zip codes where the average household income was greater than $45,000. Many of these people may have been uninsured by choice. "The uninsured population in our study included a subgroup that may have had a high income," Ailawadi [one of the study's authors] said." Also worth noting is that Ailawadi has also said that "I don’t think we’re able to say a government-sponsored system is beneficial or not." If he, one of the study's authors, doesn't believe his study proves which type of insurance is the best in terms of health, then who are politicians, pundits, and other non-experts to suggest otherwise?  Similarly to the above confounders, other experts have noted that "...uninsured patients may elect to go without coverage and end up on Medicaid when they become sick, or that insured patients in the study may have reached lifetime policy limits and had to resort to Medicaid."
And as though this all weren't enough to debunk the right-wing interpretation of this study, consider that it wasn't even supposed to study which system worked better! Kron himself has said that the study he co-authored "focused on whether socioeconomic status was a factor in medical treatment, which it clearly was, and not on the quality of the systems paying for health care."
In short, don't believe this bullshit about the "13 percent worse" claim about Medicaid, which totally 

misses the point of the study on which it is based and lots of other factors that could account for this 

difference.