how statistically valid are articles we read?
The version of the browser you are using is no longer supported. Please upgrade to a supported browser.Dismiss

View only
Article TitlePublicationDateAnnotations
The Reproductive Rights Rollback of 2015NYT Sunday Review12/19/2015Claims the trend of abortion restrictions has "accelerated in 2015", even though there were 57 restrictions in 2015 as compared to 288 since 2011, meaning that 2015 is no increase over average -- data they link to from the Guttmacher Institute also does not support their claim.
No Justification for High Drug PricesNYT Sunday Review12/19/2015cites "recent journal article" but does not link to recent journal article. Cites "a recent analysis" but does not link to analysis. "surveys have shown" ~ does not link to surveys. Very hard to check claims.
The One Question You Should Ask About Every JobNYT Sunday Review12/19/2015Relies heavily on a study which is three decades old -- and given that the study is about workplace culture, it seems like its conclusions could change profoundly (acknowledges this problem). Claims that "that some elements of cultures are unique, but those are the least important parts" but provides no real evidence to back up this claim.
Syrian Refugees: In Desperation, Obama offers statistics and liesFox News Op-Ed11/20/2015Questions Obama's refugee statistics because he does not provide sources for them. Then offers statistics of his own (72% have been men of military age) without providing any source for it. References "claims that are at best questionable and in many cases outright false" without providing any examples.
Our Obsession with Elite Colleges is Making our Kids Feel WorthlessQuartzCites study of 550 American CEOs but provides no link to study. Further, says study says that 2/3 of CEOs do not go to elite colleges -- but that's obviously true because the vast majority of people don't go to elite colleges (and what is the definition of elite anyway?) The relevant statistic (for students / parents, to whom this article is targeted) is the odds ratio -- how much more likely are graduates of elite colleges to be CEOs? And even this doesn't engage at all with the question of whether going to an elite college is causal. Also, this whole article (like the Atlantic article it cites) is about Palo Alto and links the high suicide rate to high achievement. But I've yet to see a compelling argument that the high achievement / wealth / pressure causes the high suicide rate as opposed to the Caltrain; the Atlantic article acknowledges that rich kids are no more likely to kill themselves.
Which State Has the Worst Drivers?FiveThirtyEight10/24/2014Points for using three different metrics. Points for attempting to normalize for population + miles driven. Already this article is considerably more statistically literate than most op-eds. But the first figure is very hard to make sense of -- what are the errorbars? Should I be caring about how often drivers are distracted, or the total number of collisions, or what? Bars aren't ordered so it's hard to see which states are on top -- I would probably prefer a table of numbers, not a bar graph. Why is there such a large variance in the fraction of crashes due to distracted drivers? Could that be due to different recording procedures? Similar complaints for speeding. Is it possible that different insurance rates are due to different costs of cars or amounts of driving, not different rates of crashes? Ultimately I think this piece doesn't mislead but it's also hard to make sense of -- on the one hand, it doesn't make overly strong claims from its data, but on the other hand, each metric is individually difficult to evaluate and it never really aggregates them or talks about the correlation between metrics so I can't decide if there's any consistent validity to them.
How To Predict Which Chicago Cops will Commit MisconductFiveThirtyEight12/15/2015Says: small fraction of cops are responsible for disproportionately large number of complaints. But this observation alone does not imply that some cops are bad apples -- even if cops are uniform (say each has a number of complaints drawn from the same Poisson process) some will have more complaints and these people will account for disproportionate fraction. I suspect the data cannot be explained by a plausible uniform statistical model, but author fails to refute / discuss this (although evidence presented later in article does -- under a uniform model, we would see no correlation between past + future numbers of complaints). Second: article implies that we can use bad past behavior to pick out bad cops. But there is a large multiple hypothesis problem here which it fails to discuss. Third: neighborhoods could plausibly produce larger numbers of complaints -- something author wisely attempts to control for. But there are other things which could also produce larger number of complaints besides bad actors. Maybe some cops tend to intervene in certain types of crimes, for example, and these crimes tend to produce more complaints.
The incredible unfairness of the global war on drugs, in one paragraphVox12/21/2015Pretty good article, I think. Refers to a report as a paper (implying it has been published in a peer-reviewed journal, which I don't think this has) but the report looks legit. Says that rates of illicit drug consumption are higher in the US than in Mexico -- uses two different surveys to compare, which isn't ideal, and it also seems possible that there are cultural differences in how acceptable it is to talk about drug consumption, but the differences are pretty big. It's also not clear if "illicit drugs" include marijuana (and is that part of the drug war)? Finally, fails to discuss the counterfactual -- argues convincingly that war on drugs has been harmful, but what better world are they proposing?
Sexism in paradiseThe AtlanticDecember sometime 2015"Rape is widespread in Brazil -- one is said to occur once every 10 minutes" ~ this is not the most informative way to convey this information. What fraction of the female population is that? How does it compare to other countries? Rape happens frequently everywhere. This statistic is especially problematic because the non-quantitative evidence in the article is very compelling.
Why Pot Has Overtaken Cigarette Use Among TeensThe Atlantic12/16/2015This article does at least two important things right: it links to the data source used for its analysis, which appears reliable and makes it easy for me to check its conclusions; and, while I didn't have time to check everything, the things I eyeballed looked roughly right. That said, I think the article is mistitled (which is probably not the author's fault). If you just looked at the title, you would think that pot use has gone up, which is not true; the data actually show that pot use is if anything going down, making it a little unclear why the article spends so much time talking about pot at all.
No, Suicide Rates Don't Rise During the HolidaysThe Atlantic12/9/2015I think this article is pretty good! I appreciate the attempt to debunk false and widely shared claims. I'm a little dubious of the link to pollen (the study the author links to is a real study, but having skimmed it I'm dubious of its conclusions -- the relationship between pollen count and suicide rate is non-monotonic, which doesn't make much sense), but the author notes that the science on this is "still developing". Also, the fraction of articles perpetuating the holiday suicide myth is very noisy by year, which means I don't think the article should do much analysis of what causes the myth besides saying "it is frequently perpetuated and false".
The Great Republican RevoltThe Atlantic1/1/2016Provides no links to anything, which may just be because this is a print article. Sometimes fails to contextualize its statistics. Eg, tells us about the demographics of trump voters -- "Half of Trump’s supporters within the GOP had stopped their education at or before high-school graduation" ~ how does that compare to the country as a whole? To the Republican party? "More than other Republicans, Trump supporters distrusted Barack Obama as alien and dangerous: Only 21 percent acknowledged that the president was born in the United States" ~ how much more is that than other Republicans?
Give, If You Know What's Good For YouThe New York Times12/24/2015Both pieces of research described are interesting but have sufficiently serious caveats to make me doubt the author's conclusion that their work "provides the first causal evidence that spending money on others may improve physical health". First study is observational longitudinal study where they attempt to control for covariates, look at how much people gave, show that a while later people who gave more have lower BPs. Okay, but the unobserved variables problem is pretty serious here -- maybe those people had less stressful jobs or more friends or nicer personalities or a billion other subtle factors no set of covariates could capture. The second study is an actual experimental intervention and it does indeed seem to demonstrate that being randomized to a group where you give away money is better than being randomized to a group where you do not. But that is not the same as saying choosing to give away money makes you happier than not.
The Typical American Lives Only 18 Miles from MomThe Upshot (NYT)12/23/2015I think this article is good! It talks a lot about how physical proximity predicts the probability that families will offer support to each other. I think it should've discussed the fact that causality here is complex: we could have "proximity -> support", or "parent needs support -> you move back to provide it" (reverse causality) or "you live near each other <- you like each other -> you provide support" (or other third variables). But this is not a major complaint.
The Marriages of Power Couples Reinforce Income InequalityThe Upshot (NYT)12/24/2015I like this article. I can't really find much wrong with it. It quotes the wrong figure from one of the studies it cites, but only slightly (.34 vs .35) which doesn't affect its argument.
As Graduation Rate Rises, Experts Fear Diplomas Come Up ShortNYT12/26/2015This article seems to compare apples to oranges. It contrasts the change in the graduation rate with the fraction of graduates ready for college. But either it should compare two changes or compare two fractions -- it does not make sense to me to say "the graduation rate is increasing but the fraction ready for college is low" because if the fraction ready for college is also increasing, that is not really a contrast. It doesn't seem to discuss the change in the fraction ready for college at all, which is the immediate question I have.
A Year Of Reckoning: Police Fatally Shoot 1,000Washington Post12/26/2015This is a really good idea and the Post deserves credit for putting in the effort. Kudos also for coming up with 6 key takeaways although I think the article should rely less on anecdote and more on analysis of their unique database. And kudos for making the data available. Overall this article really impresses me. "In the majority of cases in which police shot and killed a person who had attacked someone with a weapon or brandished a gun, the person who was shot was white. But a hugely disproportionate number — 3 in 5 — of those killed after exhibiting less threatening behavior were black or Hispanic" ~ it's not clear how to interpret this. One possible cause: police are racist, and only shoot white people if they really have to while they shoot minorities for less threatening behavior. Another possible cause: white people have a disproportionate number of interactions with police when they're brandishing guns, but not when they're not. Really the statistic I am interested in is this: if you are brandishing a gun at police, what is probability that you are killed if you are white vs if you are not? But the data doesn't quite give us this. "The Post found that the great majority of people who died at the hands of the police fit at least one of three categories: they were wielding weapons, they were suicidal or mentally troubled, or they ran when officers told them to halt" ~ what fraction in each category! "Although more officers were indicted in shooting cases this year, the outcome of such cases improved for officers. Five of the seven cases tried this year ended with the officer acquitted or with a mistrial. In two cases, charges were dismissed. Over the previous decade, one-third of officers charged in shooting cases were convicted of crimes ranging from misdemeanor reckless discharge of a firearm to felony murder" ~ this is not statistically significant. "About 6 percent of fatal shootings this year were captured by body cameras, according to The Post’s database" ~ this is interesting. 6% is small. "fewer than half of the nation’s 18,000 police departments report their incidents to the agency.The Post documented well more than twice as many fatal shootings this year as the average annual tally reported by the FBI over the past decade" ~ this is also good.
Why Police So Often See Unarmed Black Men as ThreatsVox12/29/2015
The Unseen Victims of Traumatic Brain InjuryThe New Yorker12/30/2015This is a very compelling article that makes me want to kill someone, but this is not good: “every jurisdiction that has prosecuted strangulation as a felony with a multidisciplinary team has seen a drop in homicides" -- homicides in general? homicides of domestic violence victims specifically? how many jurisdictions? and most importantly: what about the jurisdictions that didn't prosecute strangulation as a felony? Strangulation "dramatically increase the chances of domestic-violence homicide" -- ie, by strangulation itself? or does it just predict that they will be killed in a different way (ie, it is a marker of intent)?
Thanks, Obama: Highest Earners Tax Rates Rose Sharply in 2013The Upshot (NYT)12/30/2015I can't find anything wrong with this piece, although it implies that the other piece of reporting the NYT did should've been more thorough in its analysis (since as far as I remember they neglected to mention that Obama was cracking down on tax shelters).
Happy New Year: A Time To CelebrateThe Cato Institute12/31/2015I think the general point of this article is pretty clearly true, but it doesn't seriously engage with any of its statistics. Eg, global years of life lost due to a disease is not very interesting unless you control for things correlated with having that disease (population size; population age). I don't think it misleads, though.
To Win Iowa or New Hampshire it May Be Better to Poll Worse NationallyFiveThirtyEight1/4/2016Interesting piece. Regression is intriguing but as far as I can tell doesn't cluster standard errors by election + this is pretty important to do -- results for different candidates in the same primary are very obviously negatively correlated with each other, meaning that the true number of datapoints is arguably closer to 12 than to 167. While the piece acknowledges the small sample, it doesn't actually use the proper statistical tools to compute p-values in light of it. I don't read that much of FiveThirtyEight's political coverage, but this is something which is generally important to do.
How To Make Yourself Go to the GymThe Upshot (NYT)1/10/2016I have no complaints about this piece and thought it was very interesting.
How Survivor Bias Distorts RealityScientific American9/1/2014The basic point about business books is a good one and I think this piece is basically okay. I think this piece becomes a bit confusing, though, because the point about the playing cards is not illuminating; it's not really clear how it links to the business books. Also, really what is happening here is not best described as a bias (although I guess that is the standard term) but overfitting to the data.
Privilege, Pathology, and PowerNYT1/1/2016The obvious question on reading this first paragraph is "well does wealth actually make you an asshole, or do assholes just tend to get wealthy?" And Krugman doesn't bother clarifying or even addressing this obvious question, even though at least some of the social science work he cites implies that wealth is causal. This does a disservice to his readers + to the scientists whose work he cites and is especially irritating because I know he knows better; no Nobel-prize winning economist would be able to overlook the obvious correlation / causation problem. This social science work is far more compelling than the anecdotes that dominate the piece, and he should take the time to explain it properly.
White Female Republicans are the Angriest RepublicansJezebel1/4/20161. White female republicans are the angriest republicans? You talk about whites and republicans and females, but not about the intersection, and it's entirely possible that the intersection is different. This is literally the definition of intersectionality -- an ideology it's apparently okay to drop when talking about Republicans. 2. Self-reported anger may not have much to do with actual anger -- it may have to do with cultural acceptability of reporting anger / emotion. 3. The piece switches constantly and confusingly between "how angry are you" and "are you angrier now than a year ago". These are non-equivalent things. I have no idea which the piece is referring to half the time.
All-Male Fox News Panel Gets to the Bottom of this 'Women's Voters' ThingJezebel12/??/2015Criticizes Fox News for showing poll in which favorability numbers don't add up to 100 -- but they aren't supposed to add up to 100. This is a very, very bad mistake because it indicates the author has no understanding of the numbers being analyzed. Then critiques this guy's claim that "women care about same things as men" without providing any evidence that this is not the case. It may well be that women do rank the same issues as most important (and just place more weight on other issues) -- or it might not be, but this article gives me no idea.
Happy New Year! Tomorrow is the day you're most likely to dieJezebel12/31/2015This article is better than the other two Jezebel pieces. I have only two complaints. 1. It says that "general sadness" could drive the deaths over the holidays, but I don't think that's right -- the holiday suicide spike is a myth (according to the Atlantic article discussed above). 2. It implies that the whole thing is one big mystery, but the evidence it presents actually implies at least one explanation -- medical care is worse on the holidays.
Racial prejudice is driving opposition to paying college athletes. Here's the evidence. Washington Post12/30/2015This is an interesting piece and is written by the authors of the study and I appreciate the attempt to bolster a strong causal claim with an experiment (as opposed to just controlling for things in a survey). That said, I have a few pretty serious complaints. 1. There is no link to the study, so I can't evaluate the study itself. Nor can I find the study when I look for it. Maybe there is no study? Or what? But that seems extremely unlikely -- they did a survey and an experiment, they're planning to publish this somewhere. Regardless, a lot of the terms in the article are not sufficiently well-defined for me to evaluate them. How did they control for things? How did they define racial resentment? Is anything statistically significant? Do I have to worry about garden-of-forking-paths problems? Etc. 2. If racial prejudice is driving opposition, why are so called "racially resentful whites" just as likely as whites overall to support paying athletes? This seems like a huge problem. 3. The effect sizes they show for priming are not large enough for me to believe that racism accounts for all the gap (or even a large part of it?) making the title somewhat misleading. 4. Did they evaluate the effect of showing whites pictures of whites? Or blacks pictures of blacks? Or whites? There are various controls here which seem important to rule out alternate explanations.
Depression and Anxiety in Women Linked to Male-Female Pay GapThe Guardian1/6/2016You know how sometimes you can just eyeball a headline and go "they're going to conflate correlation and causation, aren't they?" This article did not disappoint. It also committed a multitude of other sins. It doesn't link to its study, which I'm not going to bother to find -- I'm too irritated at this point with journalistic coverage that does that -- and it doesn't discuss a) the problem of reverse causation (women with depression are paid less) b) the unobserved variables problem (controlling for education and work experience does not account for all intangibles that affect work performance -- and what about work hours) c) the fact that its "79 cents on the dollar" statistic is totally irrelevant to what's being discussed (since that doesn't compare women to their "equivalent male counterparts") and d) basic details like which industries were examined and how they got the data and who the male counterparts were.
Powerball Drawing: Clues from Previous WinnersABC News1/7/2016This is so silly I refuse to comment on it.
The Economics of Prettiness: More Attractive Women Get Higher GradesVox1/8/2016Fails to discuss third variable / reverse causality problems. One can imagine biases going in both directions: women who are prettier are spending more time on appearance (exercise, makeup) and less on studies -> lower grades. (This bias is the reverse of the one reported.) Or: women who are prettier have been getting many advantages (better social life -> more confident -> speak up in class more -> higher grades). This is interesting: "In online classes, attractive female students performed worse than expected given their other grades, suggesting that their looks were giving them a lift in regular classes" ~ this still doesn't deal with all third variable problems. Article also doesn't mention other obvious confounders like race (correlated w both appearance + grades) -- original study attempts to control for this, which article doesn't mention -- or fact that controlling for it is imperfect because just having a race variable won't capture things like skin tone, which correlates with life outcomes.
Campus Sex...With a SyllabusNYT Sunday Review1/8/2016"No, most rape is not the result of a misunderstanding. To the contrary, one-fourth to two-thirds of rapists are serial attackers, studies show." a) this is a huge range -- 1/4 is super different than 2/3 -- article needs to reconcile these results. b) "most rape" is not the same as "most rapists". long-tail effect is important here. c) if only 1/4 of rapes were committed by repeat rapists, than could 75% be the result of a misunderstanding? this is an incorrect interpretation, but someone could reasonably make it from article's statement d) if most rape is committed by serial rapists, the article needs to talk about that a lot more, because why would we expect these rapists to be swayed by consent training? e) "More than 1/5 women will be victims of campus sexual assault" ~ cites WaPo study on this, which has a severe low response rate problem.
When Teamwork Doesn't Work for WomenThe Upshot (NYT)1/8/2016I'm generally persuaded that this is a problem. But he doesn't discuss the third variable problems here at all. Eg, maybe women who work alone are more likely to be unmarried. Even the fact that gender of coauthor makes a difference to how much credit the women gets -- maybe this is because gender of coauthor correlates with seniority in some weird way. The author attempts to control for this, but I'm not sure I agree with the way she did it -- she seems to control for coauthor seniority, but not with whether women, when they coauthor with men, tend to coauthor with senior men.
I quizzed dozens of Silicon Valley elites about inequality. Here's what they told me.
Vox1/9/2016Overall pretty interesting + looks like it required a fair bit of effort. "A plurality of founders agree: Among 33 founders I surveyed, 48 percent said that mediocre growth was more problematic than financial inequality, while 42 percent believed the opposite. Among the general population (as represented by 595 people polled on SurveyMonkey), 59 percent of people believe inequality is more important." ~ this is not a statistically significant difference. A bunch of other gaps the author points out are not statistically significant either. The author addresses this concern in the supplementary methods but a) this response seems dubious to me and b) it should be in the main methods. Should also probably discuss biased sample due to low survey response rate.
You Can't Trust What You Read About NutritionFiveThirtyEight1/6/2016Nice piece. i think it's solid, although I'm of course biased towards anything that complains about lots of other studies -- this piece makes a convincing case that you should not take nutrition research very seriously. I'm not sure taking 3 538 writers and asking them to keep a food diary is the most rigorous way to study food diaries, but it's entertaining and it gives us an idea of the objective problems. Why does it take them hours to run 27,000 regressions? It seems like it should be faster than that. Whatever. They should maybe mention that there are standard ways of correcting for multiple-hypothesis problems, but given that most of the studies they're critiquing probably don't do this, that's not a serious omission.
Masters of LoveThe Atlantic6/12/2014Overstates causality for results that (as it describes them) are correlational. How do we know that kindness causes couples to stay together? Perhaps it's that if you're kind to someone, it's a sign you love them more; or are under less financial pressure; or don't have a mental disorder. The article advises us to be nice to our partners, which is of course intuitive advice, but the results don't actually show that being nice to your partner will make you stay together. Also, what is this casual reference to the claim that "being mean to your partner gives them cancer"?! I'm not going to read that study in detail because it needs to be fleshed out or dropped, but it has all the hallmarks of a correlational result that they're overstating as causal. (also, the study they link to doesn't mention cancer in the abstract? It talks about hormones? what?)
Powerball's $1.3 billion swindle of AmericansThink Progress1/11/2016Mostly I like this article. But it implies that the lack of jackpot causes people to not play, which is a plausible thesis, but doesn't advance any real evidence in support of this claim -- a million things besides the lack of jackpot might've caused last year's decline in ticket sales.
Black Hospital Patients Given Cold Shoulder in Disturbing New StudyHuffington Post1/11/2016Simulations may be very different from real behavior. Study was not blind: coders of behavior were aware of race of patient (and presumably of hypothesis), which could influence coding of subjective behaviors.
A New Study Confirms Anti-Vaxxers Are Fueling The Rise Of Measles And Whooping CoughBuzzfeed3/22/2016Sloppy about causal language. The JAMA study it describes is very careful to say there is only an association. Buzzfeed ramps up this language. I think this is a pretty minor sin, though, because causal claim is plausible given what we know about how disease spreads.
A New Study Shows 92% of College Students Would Prefer to Read Paper BooksBuzzfeed2/9/2016In general pretty innocuous. Should link to study, not to NBC report. Possibility of sampling bias? (how was survey conducted?)
The "Gender Tax" study shows that women are charged more for being womenBuzzfeed12/22/2015The title is not an accurate statement of the findings. "items targeting women cost more than the male versions 42% of the time" ~ what fraction of the time was the reverse true?
Study: Minimum Wage Increases don't Hurt the Restaurant IndustryGawker1/15/2016I am not sure this study has been peer-reviewed? Gawker also doesn't describe the methodology of the study in anything like enough detail for me to know if it's valid. Once I look at the study, I don't know that their results support their claim. For one thing, they do cite a meta-analysis which says that effect sizes (for effect of minimum wage on employment) skew negative -- they attribute this to "biased economists". Then they do their own analysis of data, but it's not at all clear why we can get causal conclusions from their regression. Given the vast amount of research which has been done on this question, it's not clear to me why this study is worth reporting on.
Study: Dumb People Think Dumb Things are DeepGawker12/4/2015This is a fun study, but the way they generated "bullshit" is not really generic -- it's biased towards "new-agey" sounding stuff -- so sure, you may target people who like new-age stuff. If you generated bullshit using financial buzzwords you might find that mathy / finance people were more receptive to it. Also, their coefficients (Table 1) change sign dramatically and still remain highly significant?? That seems bad (indicative that their variables are highly correlated and it's not at all clear what is causal). What on earth is a pearson product-moment correlation? That's just the correlation, right? They also run literally hundreds of tests on small samples, which makes me very nervous about garden-of-forking paths problems. Table 3: did they actually look at the correlation between...a variable X and Y - X? Of course that is going to be negative? Also, they re-analyze everything using "outliers removed" to get nicer conclusions which again makes me paranoid. From Table 3: it seems like people who like bullshit also like motivational statements in general, so maybe the paper should really be "who likes motivational statements". Similarly, in Table 4, when they actually look at the difference between BS and motivational statements, that difference is uncorrelated with many of the variables they claim are causal -- not at all clear what they're identifying is bullshit specific.
Study: More Useless Liberal Arts Majors Could Destroy ISISGawker12/7/2015Egregiously click-baity title which I guess is supposed to be tongue-in-cheek...except they're making jokes about an actual attempt to stop a terrorist organization and, also, you know this article is going to be reshared by people who don't take it as a joke? More importantly, the article itself is full of bad statistical reasoning. How are you concluding studying engineering CAUSES people to join ISIS? Engineering is plausibly correlated with 8 gajillion other things and yet you're advocating “humanising” the teaching of scientific and technical subjects. Even more egregiously: you haven't even shown a correlation! You say: lots of educated jihadis studied engineering. But maybe lots of people from countries that tend to produce jihadis study engineering? What is the base rate of studying engineering?
New Study: Ebola Virus Could Reach US by "Late September"Gawker9/4/2014Again, title is bad -- anything could happen, the question is what is likely to, and 18% is pretty unlikely. Also, the article needs to tell me where that number came from beyond "air traffic patterns". What assumptions were made? Do independent researchers find the methodology reliable? Simulating spread of an unprecedented outbreak like this one seems super-hard. Also, this reporting shows the typical short-term, Western-centric bias which was so irritating about the reporting of this outbreak: if you look at the actual study, they focus not on the 18% to America in the next month but on "The extension of the outbreak is more likely occurring in African countries, increasing the risk of international dissemination on a longer time scale".
Study: Gen Z Thinks Their Peers Are Completely Insane About the Reality of Girls and BoysBreitbart3/15/2016Okay, ignoring the trans / homophobia in this article -- is this study peer-reviewed? Who is this advertising giant and how did they select people to survey? I started by assuming this article would competently describe the study results, which are really straightforward (bar graphs), but it actually doesn't. "All the rest portray themselves as somewhat bi-sexual" ~ false; some said they were asexual, and some chose not to answer. "When shopping for shoes, 39% of boys might buy high-heels, or maybe even push-up bras" -- a bra is not a shoe? Also, study says nothing about either of these products -- what it actually asks about is whether you "always buy products which are geared specifically towards my own gender" so eg if a guy buys non-gendered FLIP FLOPS, that would count. I could keep looking at this article, but suffice to say it's bad.
Study: female genital mutiliation now blights every corner of EnglandBreitbart7/22/2015Misdefines FGM -- there are different types / severities. To focus on statistics -- the article emphasizes repeatedly that places in England with more immigrants have more FGM. What they don't mention is the way the study estimated the incidence of FGM: they literally took the number of people who immigrated from countries with FGM + multiplied it by the incidence of FGM in those countries. So the correlation between FGM + immigrants has to emerge as a consequence of the study methodology; it's not a correlation which emerges independently. I suspect the author of the article either didn't understand this or didn't care, either of which is bad.
Main menu