I am someone who is coming to statistical literacy fairly late in life, or late in education, anyway. I was a humanities guy who is becoming a social science hybrid, and learning the last five years about research methodology, quantitative research, and similar topics has been a challenging and satisfying endeavor. But I’m still very green, so I’d like a little help from the BJ community on a statistical question.
This piece on angry unemployed and underemployed Millennials from the Atlantic drove me crazy, even though I’m predisposed to be sympathetic to their plight. One of the emailers writes, as part of an exhausting whinge,
Much of my rage is reserved for a predatory system of higher education and the failures of a generation that came before. I’m angry that a “state” university costs as much as it does. That many, if not most of the students who attend, treat the experience like a 4-year version of MTV’s Spring Break. Massive grade inflation means one less standard deviation between myself and those who don’t try. Lax entrance standards means that even in smaller classes, half of the students do as little as possible, have nothing to contribute, and see learning as a necessary evil, if even that. These “state” universities are more interested in funding nice football stadiums than maintaining up-to-date libraries or modern classrooms. They are more interested in your tuition than your education. And will continue to hound you for Alumni contributions long after graduation….
I pursued a “Liberal Arts Degree” in communications rather than a B.S. in engineering or computer science. I spent all four years at a state university rather than the first two at a community college. I worked in the summer instead of getting an internship. I worked harder at my classes than making contacts and networking with professionals. Not everyone is suffering in this economy, and if I were going to college for the first time this fall I’d know how to prepare. But I didn’t at the time and now I’m left to face the consequences. I want to blame the universities and “grown-ups” who I feel should have known better. They were the ones, after all, peddling the mantra of “go to college, study hard, get a job.”
Instead, egotistical like the rest of my me-first, entitlement ridden generation, I blame myself. (emphasis mine)
Aside from the sense of entitlement, the paranoid style that assumes forces are working against this relatively privileged person, the non sequitur about his hard partying peers, the reliance on the commonly assumed but unsupportable claim that liberal arts students are disadvantaged economically, the odd use of scare quotes around “state” (is the idea that, secretly, his college wasn’t a public university?), his assumption that he would have gotten that computer science or engineering degree had he tried (those are tough fields!), the lack of evidence that his particular school has high grade inflation, the mathematical evidence that even high grade inflation doesn’t prevent discrimination between similar students, and worst of all, the fact that he makes a show of blaming himself when his email makes clear that he most emphatically does not really blame himself….
Isn’t his claim about grade inflation and standard deviation exactly wrong? Wouldn’t grade inflation, as it is usually understood, compress the scores themselves but leave the distribution in standard deviations unaffected? I mean, as I understand it, that’s part of the point of using standard deviations to evaluate a distribution. I’m asking genuinely here, and I could easily be wrong, so pipe up in comments. I would love an explanation from someone who really knows their stuff.
If I’m wrong, I’ll buy this poor kid a steak.
Jenny
Ironically, you just described the typical manic progressive blogger, especially the elite front page set.
Little Boots
yeah, this goober sucks. but they’re a small part of the problem.
and Freddie De Boer is here now? that’s awesome.
Adam Hyland
The statement is a little non-sensical. The distribution will be tighter so the empirical value of the standard deviation (just the sqrt of the variance) will be smaller.
TomH
I expect it WOULD affect the standard deviation. The simple explanation is that let’s say all the F’s become D’s, all the D’s become C’s and so on. Then there is less variation about the mean; previously F was far away from the mean, now D is less far away from the mean.
Presumably, colleges may still provide some evidence like a rank of a student in their class that helps one evaluate just how they performed relative to their peers, grade inflation or not.
Doctor Science
Nope, in this respect, at least, he’s right. Or at least he makes sense.
He’s saying that absent grade inflation, the grades would be spread out to make a nice, wide normal distribution, such that there are two standard deviations between him (a putative A) and those C-earning slackers. But because of grade inflation, the jokers are getting Bs; because grades are stepwise and have a top bound, both students 1 SD above the C students (=should be B students) and the 2-SD good students like him are getting As.
Ron
You’re wrong because there is a maximum possible grade possible. So when the grades get shifted upwards it compresses the the range of the grades.
cleek
does grade inflation (as it is usually understood) push all grades up equally? because if it only compresses the high end, and doesn’t affect the low end, you the std dev of the new graph will not be the same.
Freddie deBoer
Hmmm. I’m still confused by how he gets to a standard deviation in reduced distribution range.
But I will buy him that steak. Email me!
TheF79
Hmm I’d say Freddie is right – the number of standard deviations between two hypothetical students would stay the same under the grade inflation, but the size of the standard deviation would get smaller. A 95% student is still 2 standard deviations above the 50% student, no matter the reference point.
statzilla
the range of possible grades remains the same, but that’s not the same as the standard deviation. they’re both measures of spread or dispersion, but the standard deviation is the spread about the mean whereas the range is just all possible scores. Grade inflation aside, you know that students still fail. Maybe as many as before but there are fewer C’s and B’s and more A’s. That would then concentrate more observations on the right side of the distribution, pulling the mean up in the process, affecting the SD in the process.
As to how grade inflation affects his position relative to others in terms of a standard deviation is beyond me at this point after a night of drinking.
Pavlov's Dog
I think what he is trying to say is that the difference in grading between top performers and bottom feeders has narrowed to the the point that differentiation is negligible.
I have mentored a number of of people in MBA programs in the recent past. People that should have flunked in a class are given a B, or horror, a C. As they say, the problem is not graduating from Havard, but getting into Havard. This seems to have spread to even low level graduate programs. You would have to be almost a corpse to get a C.
J. Michael Neal
Yes. The distribution of the grades almost certainly distributed normally[fn1] (the normal distribution is the classic bell-shaped curve). The normal (like most distributions) is defined by its mean and its variance. I won’t go through the math, because the mathematical function describing it is seriously badass. For all normal distributions, about 95% of the population will be within 1.96 SDs of the mean.
Now what can be a problem is that a compressed distribution can lead to a lot of uncertainty. That’s because your GPA is not an accurate indication of your underlying abilities. Your GPA is going to fall into a distribution with its own mean and variance. It’s just an estimate of anything useful. If the distribution of the GPAs of all students is compressed, this uncertainty about real ability will cause there to be more cases of the GPAs not being in the same order as actual ability. However, given all of the other ways that prospective employers fuck up the process, I submit that this won’t have a huge affect.
[fn1] Yeah, yeah. The fact that there’s an upper bound on GPA that this clown thinks he’s bumping up against means that the distribution won’t really be normal. It’ll be close enough for this sort of thing, so leave me alone.
Earl
Freddie,
You’re right — standard deviation or variance are measures of the average distance from the mean of the data, or the square root thereof. So as you have grade inflation, presumably the mean increases but the variance has to collapse, at least on the right, presuming the maximum grade is 100.
For the normal distribution, about 68% of the data will be within 1 standard deviation of the mean and about 95% within two.
Ron
Ugh, I didn’t read carefully. So yes, you would get smaller standard deviations, not fewer. (all this assuming that the grades are still essentially normally distributed).
Capri
I work at one of those “state” universities and I have to say the guy’s complaints don’t make any sense.
Is his point is that community college is cheaper than a 4 years at a university and that someone with an engineering degree is going to earn more with a BS than someone with a non-math degree.
Well, that’s a big fat duh. Neither is a secret. Both have been widely known and publicized since the mid seventies if not before.
If he graduated from college and just realized that fact -well he is a grade A dumb shit. He should be mad at himself.
At the institution at which I work (and the others I’m familiar with), the kids who party hearty don’t float through getting inflated A’s and B’s – they flunk out.
Retention – i.e. keeping kids in college that either aren’t prepared or flake out – is a huge emphasis in higher education right now.
J. Michael Neal
@Pavlov’s Dog: I got a C- in Management Communication. That’s the professor telling me that I failed without actually failing me. Fuck her. I worked my ass off in that class, and the fact that I was only two years removed from a complete nervous breakdown and would start twitching and stuttering when I get up in front of a group to speak should have been given more consideration.
Fortunately, I don’t give a rat’s ass what my GPA was, so it’s all good. Just so long as I got my C- and didn’t have to take the class again.
doofus
There will likely be some variation depending on the grade scale being used, but your student is at least operationally correct about the standard deviation. This is because of the same upper bound reasons mentioned by others. If you move the center closer to the horizon, your standard deviation covers less ground.
Hunter Gathers
Whining. Little. Bitch. Sounds like Grover Norquist in his larval stage.
Freddie deBoer
the number of standard deviations between two hypothetical students would stay the same under the grade inflation, but the size of the standard deviation would get smaller. A 95% student is still 2 standard deviations above the 50% student, no matter the reference point.
This was my thinking– that while the range would shrink, the standard deviations would correspondingly shrink as well, meaning that while he could argue that the empirical value of the measurement is reduced, there is not in fact one less standard deviation between him and his undeserving peers. But like I said, I’m still largely a novice.
Incidentally, if you’re interested in this stuff, do check out that Slate link about grade inflation. It’s fascinating.
Martin
@cleek:
More or less. Proper grade inflation is usually defined as an increase in the mean along with a decrease in the standard deviation. So I think that’s the gist of what you’re saying.
@J. Michael Neal: But it’s not a normal distribution. Not even close to one. As DS points out, it’s a stepwise bounded function. As you compress the grades, you lose granularity. You can’t get a grade between an A- and an A, and you can’t get higher than an A. There’s only 12 grades normally available in a +- grading system. In a full-letter grading system, there’s only 5, so the grades don’t compress well. In fact, in a grading system, it’s possible to have 0 students one standard deviation above the median grade or 0 students one standard deviation below the median.
arguingwithsignposts
I’m not calling her, but expect her.
Omnes Omnibus
@arguingwithsignposts: She is a self-proclaimed mathematician, you know.
duck-billed placelot
I think you owe the kid a steak. If there are no F’s anymore, and half-assing it gets you A’s or B’s, then a high grade-point average is no longer an indicator of hard work/intelligence above the norm of college students.
Also, sure, kid might be a bit self-centered and pretentious, but this is a hard time to be going after work. And student debt is a problem. I mean, a paranoid style that feels forces are working against him…he’s a BJ-er in the making.
Freddie deBoer
@duck-billed placelot: Yeah, I don’t mean to be too harsh. I had just read ABL’s post on Somalia when I was reading that Atlantic piece, so you can imagine my mindset, but that’s no excuse for me to be callous.
rb
The student is semantically incorrect but correct in spirit.
Due to the ceiling effect, grade inflation implies that grades will be closer together; 60% becomes, say, 70%, and 70% becomes 78%, whereas the best possible score of 100% remains fixed. Thus if the original standard deviation was 8% or 10%, the grades of those who ‘don’t try’ have gotten a good deal closer to the writer’s putative 100% in terms of the original SD.
Of course, such a reading defeats the whole purpose of ‘standard deviation’ – in terms of the (newly reduced) standard deviation of the shifted grade scheme, the ‘good’ and ‘bad’ students remain as far apart as ever. If after inflation the scoring mechanism retains its discriminatory power, and as long as the grades are interpreted in light of the known inflation, no harm done.
However, the student is correct in the spirit of things in that grades are typically NOT interpreted in terms of known inflation. A 4.0 gpa is considered a 4.0 gpa, era notwithstading. If it becomes easy for ‘just anybody’ to get a 4.0, then having a 4.0 isn’t terribly helpful for one’s job prospects.
Odie Hugh Manatee
My specialty with numbers is rolling them, I’m an artist at it.
@arguingwithsignposts:
Yup. Like a crow to roadkill.
Martin
@Capri: Well, one complaint that I have is that public universities do not make adequate numbers of seats available to all the students that desire programs like engineering, so blaming them for not being engineers when they may have applied to engineering and then gotten offered Reverse Polish Beekeeping instead, a program with no real demand, no real employment prospects, and seems to exist solely to create unemployable students that can be enlisted into unwanted graduate programs.
If we were looking at a proper supply/demand situation, a third of the existing programs at any given university would be unsustainable, and another third would grow so large as to effectively rebrand the institution.
magurakurin
reading through that article and some of the other emails, it strikes me that the main point is less the failings of the university system, but rather a bunch of young people discovering that the pathway to happiness that they were presented in high school is, in fact, a fucking lie.
Luckily some of them found out when they are in their early twenties. Still time to choose another route and find a full life rather than the sterile, mundane package of “get a 4.0 in high school, go to a good college, get a 4.0, get a job, work, die.” It never appealed to me as early as the 10th grade.
some of the complaints made me think of the last scene in Falling Down. Hopefully they will make a course correction before they get to where the Michael Douglas character got to.
Freddie deBoer
If we were looking at a proper supply/demand situation, a third of the existing programs at any given university would be unsustainable, and another third would grow so large as to effectively rebrand the institution.
Although we should be careful about assuming which is which. For instance, I often surprise people when I point out that English departments are typically big moneymakers for universities. (Of course, this gets at the labor issues within the university, but that’s a separate topic.)
Pavlov's Dog
@J. Michael Neal: The best advice I ever received about fear of public speaking was to picture your audience all sitting on a toilet with food poisoning. The second was to arrive slightly early and banter with people in the first couple rows, and then start your delivery as they are the only people in the auditorium. I also never worried about my GPA, as nobody cares except in your first job. Most of my mentoring/tutoring has been around classes that students are deathly afraid of. Public speaking is way up there, as is statistics, accounting, and finance.
Roger Moore
The width of the curve in standard deviations will remain constant only if the grade inflation leaves the overall shape of the curve the same. If there are constraints on the curve that cause its shape to change, that assumption doesn’t hold. Notably, if the grade inflation results in more students with nearly perfect GPAs, it will reduce the differentiation of students who would have had near perfect GPAs without grade inflation.
TheF79
“The student is semantically incorrect but correct in spirit.”
I think that’s about the long and short of it – though as someone who uses statistical techniques on a daily basis, there is some value in making sure that terminology isn’t being abused.
Richard S
This is the whine of a standard deviant who’ll never get the concept of upward bound and does not care about maintaining relative rank (where most transformations can restore normality).
I’m a BA, MA who is usually treated as a doctoral equivalent becuse my refereed publications are in better journals and at a higher level than my ‘peers’. I’ve gone toe to toe with and have been trained by some of the best inferential thinkers of the past fourty years. And my education was in ‘average’ institutions. Am I happy with my life – Big Yes! And this guy wouldn’t last long at Balloon-Juice. Thank FSM!
PGfan
I used to work in Placement – first placing recent college grads into entry-level career positions; later placing MBA’s (with a stint in the middle doing outplacement) and I actually have a lot of sympathy for the kid. He fell for the widely held belief that college is a guarantee; his folks believed it, his friends believed it, and folks working at colleges give lip service to it. What he didn’t understand is that college isn’t like highschool. Up through highschool you are rewarded for taking classes and getting good grades. You get extra points for additional activities. But basically the grades ARE the achievement — no one expects you to be able to use your knowledge in practical ways, you’re just supposed to have a group of grades testifying to your ability to absorb a specific array of knowledge sets.
At the end of college, otoh, you are supposed to have mastered a series of knowledge sets that can be practically translated into needed services or profit making activities for companies/organizations. It is no longer enough to have done well in an array of classes. Most Liberal Arts degrees, certainly 4-year degrees, don’t meet that criteria. Neither do most business degrees, with the exception of Accounting, which is very targeted. This is partly because the job market in general is tight and partly because companies that need people don’t want to have to train them anymore. This is not to say people with liberal arts degrees or general business degrees don’t get jobs, but they do have a much tougher time, all things being equal.
This fellow compounded his difficulties by not seeking internships (which may have been hard to come by); by not thinking strategically about jobhunting (he’s scornful about networking, etc.), and wrapped it up by going to a State School instead of being rich or connected and attending an Ivy League School, where numerous doors would automatically have opened for him. Coz that’s the other thing people don’t understand about college – it’s very much about social class. State schools produce the middling folks who will work for the people out of the Ivy Leagues. Ivy Leaguers can get Liberal Arts Degrees with average grades and still get prestigious internships and jobs. Students at State Schools have to do it all: good grades, extra-curriculars, internships, etc. If they don’t, once again, they will probably work, sooner or later, and many will, over time, build satisfying and reasonably financially rewarding careers, but they will struggle to put it together. It won’t be automatic, it won’t be easy.
Obviously I’m speaking in broad generalities, but I ran into kids like this all the time. And he shares a problem that afflicts most jobhunters: he looks at the world through the lens of what he wants or thinks he deserves. Meanwhile the job market looks at jobseekers through the lens of “what can you do for me?” and “why should I hire you over the umpty-hundreds of other applicants that have the same or similar credentials and experience?”
If people started college with those questions always before them: “what can you do for me?” and “why should I hire you, and not him, her, her or him…” many of them would make very different choices.
Lit3Bolt
@Freddie deBoer:
Would it be helpful to point out that you’re humanities major cum graduate student whinging about another liberal arts undergraduate’s whinging? That’s a very meta white whine you got going on there, Freddie. =P
Thanks for the new word, btw. Whinge. That’s a fun one.
Freddie deBoer
Would it be helpful to point out that you’re humanities major cum graduate student whinging about another liberal arts undergraduate’s whinging? That’s a very meta white whine you got going on there, Freddie. =P
Heh, well, I was genuinely asking! I needed the wisdom of the crowd.
rb
More precisely, the student doesn’t have the language to express the idea s/he wants to express. His/her point is that grade inflation destroys the discriminatory power of the grading scheme – if there is variation in ability but everyone gets an A, then by definition the grading scheme fails to measure ability; in fact it is statistically independent of ability.
The point about SDs is flat wrong, but is a roundabout way of suggesting the idea above.
As noted by Freddie, however, most research indicates that while grade inflation is annoying, shifted grade schemes retain validity in aggregate, i.e. the concern above is over-stated.
Martin
@Freddie deBoer: At public universities? No. The money makers are the ones that bring in non-resident tuition or professional fees and large contracts and grants. English programs are critical, and they can be structured to bring in some decent money, but compared to thousands of foreign and professional graduate students, tens of millions in extramural funding, and a steady stream of patent revenue, engineering programs are nearly impossible to beat outside of medicine, law, business.
Hell, even relatively unknown public university engineering programs can bring in hundreds of millions of dollars. Now, educating engineering students is wicked expensive, so in terms of getting the most bang for a subsidized dollar, english is pretty much king.
Richard S
Hasn’t ‘Whinging and Whining’ been around for a long time? Or is it just my spelling?
srv
I was reading the LA Times about some new UCLA or USC dorm, $55M for 180 posh units with hot tubs, subzeros, and all that. Something along the lines of “kids paying this much for school expect more than cinderblock dorms.”
Perhaps all the unemployed here should get a dozen credit cards and do their own MFA MTV dorm show.
Roger Moore
@Freddie deBoer:
I assume this is because they do really well on the controlling costs side of things. People tend to forget that STEM classes require lots of expensive equipment instead of relatively cheap books. Provided you have enough grad students that actual professors don’t have to grade all the papers, humanities should be among the cheapest courses to teach.
Big Baby DougJ
It cannot affect how many standard deviations you are from the mean.
What is does is lower the standard deviation.
Say that under the old model for GPA you had
Median = 2.8, SD = 0.8
Under the new one it is
Median = 3.2, SD = 0.6
Then if you took the same person with the same test scores. Let’s say he was a good student like this glibertarian genius so he was at the 86th percentile. That is 1 SD above average, so he had a 3.6.
Now he’s at the 86th percentile. That is still 1 SD above average. Now his GPA is 0.6 + 3.2 = 3.8
But he’s still the same number of SDs above the mean.
Freddie deBoer
@Martin: Hmmm, I said that very poorly. English classes don’t do tons of business, they just tend to have a very low per student cost, because they require so little in the way of facilities and infrastructure and so much of the teaching comes from poorly paid adjuncts and grad students. But yeah, I expressed that very poorly.
Little Boots
damn, did this explode or what?
jayackroyd
@J. Michael Neal:
Grades aren’t distributed normally, which makes this whole discussion pointless. There aren’t as many Fs as there are As, not as many Ds as there are Bs. The distribution of grades in the real world is skewed.
Big Baby DougJ
Yes, it is true that raising the GPA makes for less of a bell curve, but not much. My SDs were too low for GPA, they should be more like 0.4, now, I think and maybe 0.5 before.
So now he’s 2SD above with a 4.0, he used to be 2.4
Big whoop. His argument still makes no sense.
Lyrebird
@rb: The person is correct in spirit not only because of a ceiling but because of the *change* in whether the standard deviation is on a scale where differences can be detected in the letter-grade system. Employers look at the thus-filtered GPA, not the SD’s themselves.
All leading to what @Doctor Science said.
Bully for you, kid.
(I’m personally thinking that the Age of Entitlement is gonna fade as more of this cohort meets the Realities of the Job Market.)
rb
@PGfan: This. Every word of it.
The rant sounds like whining, but there’s a lot of truth to it. For many students, post-secondary education is a straight-up swindle, made possible by their following the formula they were told to follow.
I say this as someone who stumbled into a relatively employable postgraduate concentration through a combination of luck and vague (in retrospect, EXTREMELY vague) noises about the advisibility of majoring in something ‘marketable.’ Looking around, it’s very easy to see that ‘there but for the grace of FSM …’
Little Boots
in all honesty, the economics of college is a delicate subject. I benefit from this weird economy, and so do a lot of people who come to sites like this, I imagine. the whole thing needs to be rethought, like the rest of this country, but it will be painful, I imagine, for a lot of people here.
Big Baby DougJ
@jayackroyd:
Doesn’t mean it’s not normal!!!!!
It’s just not centered at C. Grades are more or less normally centered at around 3.0 to 3.3. It’s not perfect because you have more of a left tail, but it’s close enough.
James
The other stats folks have it covered, but I seriously challenge the assertion that there is anything to the “grade inflation” horse puckey. There was “grade inflation” back in the ’60’s for a while during the Viet Nam War, when students were desperate for deferments, but that pretty much stopped after the draft was abolished.
Since then, there has not been grade inflation, that I can see. People still get flunked. People still get Ds. People still get Cs. Where is the evidence of grade inflation? Seriously, present your trend data demonstrating grade inflation for major public universities.
You can’t compare grad school grading with undergrad. Grad school has ALWAYS had As and Bs — a C is a failing grade in grad school, and in a lot of professional schools. Let’s keep to undergrad, which is what this whining crybaby is about. Show me the data.
I feel bad for the so-called Millenials, but jeebus, they aren’t the first graduating class to meet with massive unemployment. Join the crowd. It’s a cyclical thing, more prevalent under Republican administrations than the Democrats. Under Nixon, graduating engineers couldn’t get jobs. In the late 1980’s and early 90’s, no one could find a job. Then you had Clinton years, with bigtime job creation. Not so much under Bush, except if you went to Liberty University.
Whining as hard as those little brats are doing, isn’t going to help.
rb
@Lyrebird: If the letter grade system is percentile based and the numerical grading scheme remains sensitive, then the former will have no difficulty picking up the effect of inflation on the latter.
The difficulty is strictly in terms of whether the numerical grading scheme loses validity (an actual numerical problem), or if the translation to letter grades is not adjusted for inflation (a problem of laziness).
Freddie deBoer
Since then, there has not been grade inflation, that I can see. People still get flunked. People still get Ds. People still get Cs. Where is the evidence of grade inflation? Seriously, present your trend data demonstrating grade inflation for major public universities.
This has been my consistent beef with talk of grade inflation myself. I’m actually preparing a lit review on the topic now and I’m finding little in the way of proof, although I’m still in the early going.
Little Boots
@Freddie deBoer:
that ain’t the problem people, not even close.
Martin
@rb:
Grade inflation really isn’t the sole problem there, though. Curving is every bit as much of a problem by destroying the discriminatory power of the test itself. The physics program I was in had a grading guideline of unbounded grade distributions. Basically, tests were designed to be impossible for students to complete in the allotted time – the instructor would be asked to design the test so they couldn’t complete it in the allotted time. That eliminated the boundedness problem, then the test was curved to set a mean at a B-, and each SD was one letter grade. That solved the inflation problem.
That worked great until I showed up. My math skills were very strong, and I’m a very good test taker. In my first semester I completed every exam and aced at least one of them. The result was that the mean didn’t move very far from where it was, but the SD exploded. On one test, the students that got 0 would still have earned a passing grade. After many demonstrations that I wasn’t in fact cheating, they struggled to find a middle ground on the grading policy.
Fast forward to my job, and I’ve been pushing through a different approach from my experiences. Basically, we do an independent learning assessment and use that to gauge whether there is grade inflation or not (statistically, not case by case). If everyone met the learning outcomes, and everyone got As, then okay. But in that case we go back and expand the learning outcomes. So grades always chases achievement. We’re not trying to flunk students, just expand the distribution while also increasing the expected achievement.
It’s a bit atypical though.
seabe
Every time I read about grade inflation, I’m always left wondering, “So where was it at my school?” Of course my area of study was aerospace engineering, so perhaps that’s why it differs with a lot of these more national averages for the percentage of A’s. A typical class throughout my entire time in college was like this:
10-15% A’s, 20-25% B’s, 30-35% C’s, 10-15% D’s, 5% F’s. Sometimes you’d have that one class where everyone takes it for the freebie and it would be 30% A’s, 40% B’s, and 30% C’s…but they were rare.
edit: The way Martin described how our tests were prepared is about right, too. And I’m a horrible test-taker, but a good researcher and technical writer. So I was good in the lab, good on the homework and projects, horrible on the tests. Thus my mediocre GPA (meh, 50th percentile).
Sly
She’ll be a member of the most overworked and underpaid age cohort in a hundred years. The rage is understandable, though the direction in which she channels it is massively wrong-headed. Not only should she blame her parents’ generation, but I’d throw her grandparents’ generation in for good measure.
@duck-billed placelot:
It’s not only a problem, it is more often than not the biggest problem. Tuition costs have grown at a faster rate than housing and health care combined over the past few decades, with a corresponding increase in student loan debt. The Fed estimates that cumulative student loan growth hit 500% in 2011, using only 1999 as a baseline.
suzanne
@Martin:
I taught architecture and landscape architecture undergraduates for three years. On the first day of class for the freshmen, the Director stood up and told them, “Ninety percent of you will be gone next year. We don’t have enough room for all of you.”
Of course, architecture and design schools, knowing that everyone and their brother thinks that being a designer is the coolest job EVAR, are a lot more wary of admitting more students than the market can bear. Unlike, say, law schools.
Kilen
First things first, we’re talking about a multinomial distribution that is unlikely to resemble anything close to a bell-shaped curve. So, standard deviations really don’t tell us anything useful. If the argument was made in terms of percentiles, then it might make sense.
Secondly, Freddie, how can you find very little in the way of proof for grade inflation? Head over to jstor.org, type in “grade inflation”, and you’ll get plenty of peer-reviewed research on the topic. I’m not active in that field, so I don’t know which journals are at the top of the field, but it shouldn’t be hard to check impact factors for a crude guess.
Nylund
I think the simple answer is that you (Freddie) would be correct if you were working with a simple distribution (eg. normal, t-distribution, or whatever), the distribution gets compressed, but the SD would similarly also shrink such that anyone at the 95 percentile is still the same number of standard deviations away from the mean, its just the size of the SD that changes. But since this is a weird distribution (a bounded step function), the kid is likely right.
But, it’s a good lesson. When you first learn stats, you learn it using all sorts of distributional assumptions to make the concepts easier to understand, but in the real world, distributions aren’t always as simple or well-behaved as the ones you first work with.
Distributions can be really weird. A Cauchy distribution has the same basic bell-curve shape as a normal or a t-distribution, but it’s moments don’t exist, ie, the mean, variance, etc. are all infinity, always.
In short, always be aware of your implicit assumptions. Assuming normality (or assuming that the normal distribution is a decent enough approximation) when thinks aren’t normal can really mess you up (and, in things like finance, can lead to really bad things).
Little Boots
student loan is a big problem. we need to figure out how we are going to let people get an education in this country. we really do.
suzanne
Between the two of us, my husband and I have a hundred and fifty grand in student debt. And that’s with my full academic scholarship for undergrad, and us both going to state schools.
The only saving grace is that neither of us took out any private bank loans. Thank FSM.
Little Boots
@suzanne:
see, that should not be. and yeah, never, ever take private loans if you can avoid it, but still, 150,000. jeebus!
Larkspur
All I know is that my college career was short and disastrous. If I had it to do over, I’d have learned a trade, and I would now be the most sought-after installer of (a) fine paving stones or (b) drought-resistant landscaping, or (c) spectacularly excellent kitchen and bathroom tile that my part of the world has ever known. I’d be employing hundreds of people. I’d be teaching other folks trades. And yes, I’d be kind of terrified at the prospect of my client base disintegrating because nobody has the money to buy my work any more.
Crap. It started out to be such a delightful fantasy.
Possibly my worst mistake was to assume that statistics was a thing, or a set of things, that I couldn’t possibly learn. Well, duh. Some things you just have to hunker down and accomplish even though it hurts and is totally psychologically unnatural. Even paving stones or tile don’t fall into place without some intentionality.
Also, whinge is kind of British. I know this from Buffy the Vampire Slayer, and yes, there is another career path I failed to pursue.
Walker
@Martin:
All grading should be on learning outcomes. The problem is that your average faculty member has no idea what that means. I can count on one hand the number of people in my department that know the difference between summative and formative assessment.
Little Boots
honestly, I think everyone should have a year between high school and college. just a year. of living in the non academic world. i think it would help.
prufrock
@James: I also believe that grade inflation is horse hockey. My basis for this is the fact that only 55% of freshmen earn a bachelors degree within six years. If grades are so inflated, why aren’t more kids walking away with diplomas?
rageahol
Only assholes make a big deal about grade inflation. I mean, I haven’t seen it, really, but i was trained in the sciences. I don’t give a wet warty fuck what happens in the liberal arts because it’s conspicuously subjective and i dont have a model for that.
with that out of the way, i ran up 85k in student loan debt just trying to get into med school, and i will run up another 250k, maybe 300k, just getting into residency. my family didnt do shit for me, but here i am, trying to do my best for people who have it worse than me. so the original whiner can kiss my ever-loving cracker ass.
student debt is Not A Problem, because of the following fact, which is not disseminated widely enough: Obama signed a bill about a year ago that says income-based repayment can only take 10% of your income. so go get that art history degree, little cracker, and if you wind up as a manager at mcdonalds, you will only be paying 10% of whatever that income is….$2000/year? who gives a shit, really. I mean i know that $2000/yr is kind of a decent amount of money if you’re only making 20k/year, but hell, GO BACK TO SCHOOL FOR A FUCKING GRADUATE DEGREE, and if you wind up in the same shithole fast food position you’re still only paying $2k.year. i dont know what more you could ask for in this fucking bullshit capitalist paradigm to foster learning for learning’s sake.
i lost my train of thought. i’m sure i’ll have more to whinge about later.
seabe
@suzanne: About private loans…assuming we still enjoy low interest rates throughout my payment period — and I expect that to be the case given our weak recovery and everyone else in the world blowing up — they’re going to be cheaper for me than my Federal ones. My interest rates for my private loans were either substantially lower (between 3-6%), or the exact same as the federal (6.8%).
Walker
@prufrock:
Well it is true that at some institutions, they encourage instructors to drop students rather than fail them. It looks better on the books. When I was at a SLAC, the registrar asked us to enforce attendance for precisely this reason (because absences are a good predictor of failing).
Little Boots
@prufrock:
six years now? is that really the average?
MattR
I don’t know much about grade inflation but it seems like it is entirely dependent on how the course is graded. For example, my freshman Intro Chem class was graded with the mean receiving a C. IIRC, the mean for the first exam was somewhere about 30 out of 85 (about 35%). OTOH, I took a probability course where the professor guaranteed an A to anyone who got over a certain % on each exam (based on historical results) and roughly 2/3 of the class ended up getting A’s or A-‘s.
(EDIT: I guess I should mention that the vast majority of my classes were math or hard sciences (or engineering) and I didn’t take any true liberl arts classes)
@Walker: Heh. My policy as a student was to drop any class that took attendence.
Martin
@Sly:
Which is why I think the test proposed for for-profit universities to be eligible for federal financial aid should apply to non-profit as well. If you can’t get your students jobs that pay enough to cover the cost of the loan, you lose access to the aid pool. It’s a good idea that should be more broadly applied.
Walker
I will say that I have seen documents circulated in the mathematics department of my current institution that outline “recommended means” for various courses (as the courses become more advanced, the recommended means go up). This is not necessarily evidence of inflation, as I have no idea how long this has been in practice.
Martin
@Walker:
Talk to the engineers. They’re learning. It’s a major part of their accreditation process. What’s sad is that they know it better than the social sciences folks.
rageahol
@seabe:
well you can thank Dubya for that, because before his second term, student loans were still tied to the prime rate.
in his second term, they passed a bill that put a fucking floor in the interest rate. so even if the prime rate is 0.025%, the interest rate on federal student loans is 6.8%. even if private loans are more like 3.5% because theyre rapacious fuckwads.
but again, the fact that the fed now administers federal student loans directly (because, again, OBAMA YAY) should give some leeway in determining that interest rate. if the independent loan issuers (insured by the fed!) were still making scads of cash off low income college students, then that wouldnt have a snowballs chance in hell of changing.
Little Boots
@Martin:
this actually seems like a good start. match aid to actual performance.
Chris
I just want to make one point that others aren’t so much making: In my experience, the more you emphasize grades, the more unpleasant the enterprise of education becomes, and the less the students learn. Real learning comes from intrinsic motivation, not from extrinsic rewards like grades. Any “rigor” that comes from giving more and harsher grades is a false rigor and is likely to be counterproductive.
Walker
@Martin:
I am in an engineering department. Those people I can count on one hand are the people involved with accreditation. There are many others who just don’t bother.
rageahol
@Chris: @Chris:
+1
Little Boots
@Chris:
I think you’re right. in fact from talking to teachers I know you’re right. but when you get into government funding, you know that you are going to need some standards. what do you do about that?
MattR
@Walker: I can see how that makes sense. As the class become more advanced, the number of students decreases and the quality of the students remaining increases (generally speaking). So somebody who scores above the mean in a lower level class and gets a B is probably showing the same level of understanding the material as when he scores at the mean in an upper level class.
Richard S
NYLUND – Thank you. I wish I’d said it as well.
seabe
@rageahol: Heh, as if I needed another reason to hate Republicans.
Little Boots
we need a much better system.
we meaning liberal progressives who actually give a damn about academia. we don’t know what we’re dealing with, yet.
Tyro
I’m finding it pretty difficult to believe that someone with a B.S. in Communications was a really hard worker and intense studier who thinks that all the professional rewards are going to come to him/her. This isn’t a major known for its academic rigor.
The thing is about grade inflation is that is obscures the rank within an individual class, but it serves just fine to differentiate students in the aggregate. After taking 30 classes, even when the grades are more-or-less guaranteed to be B and above, will give you a pretty good fine-grain idea of who’s on top and who’s on bottom.
And in any case, outside of graduate and professional schools, no one cares that much about your GPA.
Little Boots
WE saddle people with enormous debt. WE gotta think about that. it’s difficult, but we have to, anyway.
Roger Moore
@Martin:
This seems like the right approach to me. You set achievement targets that correspond to specific letter grades, and let the actual grades fall where they may. If the distribution skews toward high achievement, that’s a sign that you’ve underestimated your students and should teach them more. If it skews low, that means you’ve made the objectives too challenging and you need to dial things back a bit. I assume that people are reluctant to take this approach because it’s hard. It requires that you be constantly changing not just your tests but also your syllabuses and lectures, and that’s a lot of work.
Martin
@Walker:
Oh dear.
Well, the earliest lesson we learned was to not tie assessment to accreditation. That just created resentment. Instead, construct a system where the faculty are the first and primary beneficiaries of the assessment effort. ABET won’t give a shit – they just want a process that everyone has bought into and actually functions. And then bury any and all talk of accreditation – it should just become an administrative byproduct if you’ve got something that runs on its own.
Little Boots
so nobody wants to think about it. well, I don’t blame any of you.
Big Baby DougJ
I’d be surprised if average GPA over four years wasn’t fairly close to a bell curve centered in the low 3s, even if individual classes don’t have that much of a bell curve to the grades. You add up a 32 random variables (I know these aren’t really random variables) and you get something like a bell curve, usually.
Little Boots
doug? come on. go deeper. you think about stuff like this, i know you do.
sven
I don’t understand why this young man chose to use the term standard deviation at all. It would have been more natural to simply say ‘massive grade inflation makes it harder to distinguish myself from students who don’t try as hard’. If he had been a finance, stat, or econ major I might have believed the phrase was used incidentally. Instead he comes off as desperate to prove that his superiority over the other students is real. I don’t quite believe he is as smart as he thinks he is…
Little Boots
gpa. okay, let’s all pretend that’s the issue here.
cthulhu
As statistician and part time instructor, it is not worth getting caught up in the mathematics of it. There’s a lot of issues at work here, some objective but far more subjectivity/individual case effects. Moreover, the mechanics of grading are not necessarily a normal distribution-based theory; there are several alternatives out there. In a related area, the whole IQ thing continues to be a big annoyance to me as the original derivation by Binet was all about identifying weaknesses for remediation. That more people don’t question what the upper end of the distribution really represents is a sorry state of affairs.
For all his whining, there’s simply no way a 3.8 from Cal State Bakersfield, from UCLA, and from Stanford are treated equally by either grad schools or employers. Hell, I doubt that a full point GPA shift hurts you that much (3.0 at Stanford vs 4.0 at Bakersfield). But obviously there’s more to it than just grades. What with employers considering credit numbers, GPA is becoming ever less predictive.
All this reminds me of the time in high school when my friends and I were sitting around, under varying forms of intoxication, pondering as to what we would answer if a really hot woman asked us what we thought our best quality was. My friend answered “My grades” and I think we all laughed for about 20 minutes and I still snicker about it to this day.
Martin
@Big Baby DougJ: Well, yeah, but there’s a few considerations to that. You have minimum grade or minimum GPA requirements to earn the degree, so after 4 years you’ll have zero students < 2.0 because you've booted them all out. Add them back in, and your mean GPA will drop from the low 3s to the middish 2s. A typical public university profile will show a 2.5-2.6 as the mean before you've kicked everyone out, and that will steadily climb to about a 3.1-3.2 as you thin the herd. So, you are always measuring an unrepresentative population of your starting student body – you shouldn't overlook those students that you've omitted.
So now you're bounded between 4.0 and 2.0. I don't know how widely this is seen, but for the universities I've looked at, what you end up with is quite close to a truncated normal distribution but asymptotic to a fixed value, about 1-2% of the population size. There's always about 1-2% right at the boundaries.
In a grade inflation scenario, you’ll start to see significantly more students piled up against the upper bound than the lower.
Big Baby DougJ
@Martin:
Interesting…but still the 80th percentile is still probably the around the same number of SDs above the average under grade inflation as before. I mean, clearly at the very highest end 3.9-4.0 it might not be, but other than that, I can’t see how this changes things much in terms of how many SDs above the average a certain percentile is.
There’s one other thing too. That if grade inflation makes fewer people drop out, it increases your percentile ranking among those who do graduate.
FlipYrWhig
@Roger Moore:
The problem with this is that at virtually no point in your career as a college professor will your future be contingent upon how well your students learned things, much less how equitably you graded them. Your future is determined by your publications and your grants-person-ship. Grading is the bane of every professor’s existence, and the goal of a grading system is to be able to stop grading as fast as possible so that you can get back to the stuff you need to do to get tenure. There is zero incentive to derive a grading system that has any other purpose.
I’m in the humanities, and I grade a lot of essays, and give extensive suggestions for how to revise, so that way if the student revises he learns something, and even if he doesn’t he knows what he _could_ have done. Even with small class sizes, that consumes hours, and pages, that you never get back, and for which you are never rewarded. In terms of words, it’s like writing a book a year that no one will ever see.
(I do think that there is an issue with compression at the low end of grades, in the humanities at least, because even if you work to keep A’s and A-minuses hard to get, a LOT of people end up in that B-minus to B range, and students respond psychologically to a B-minus as though it were a D anyway.)
Little Boots
doug, what do you think of the cost of college?
Brachiator
Don’t worry about it. The article is little more than an example of the use and abuse of statistics. Grade inflation is irrelevant to the idea that some college students have an infantile expectation that going to college and getting high grades, or even just graduating, automatically entitles them to a lifetime career, high salary and the spouse of their dreams.
It is also irrelevant to the sad idea that simply working hard and getting good grades is the same thing as learning, or makes you superior to those who share an equal disdain for education and achievement, but who are less hypocritical about it and look at college as a place to party hard without worrying about parental interference.
jayackroyd
@Big Baby DougJ: Indeed. Thanks.
RickD
@jayackroyd:
Congratulations, Jay. You’ve proved that they are not distributed normally around a mean of C.
That hardly means that grades are not distributed normally.
At most schools, at least in the classes I’ve taught, grades are distributed normally. Some universities (such as MIT) explicitly scale their grades to have the desired mean and standard deviation. Most do not.
Different courses have different mean grades. This fact is widely known to people who teach. First year Calculus has a very low mean – it is typical for half the class to get a D or F. OTOH, upper-level courses designed for majors are likely to have a mean close to B+.
I can scale the grades so that the mean is wherever I want and the standard deviation is wherever I want. The point of the original complaint is that changing the mean grade does not necessarily change the standard deviation.
The amount of innumeracy here, even in the comments section, is astonishing.
TheMightyTrowel
OT: I am a warped and twisted person but one of my best moments of the last few years was getting a letter from the govt in late 2008 explaining that [big corporate bank] which held my student loans had sold them to the govt as toxic assets under TARP. I don’t think it was phrased quite that way, but that was the translation. I AM a Toxic Asset. Also 4 months later a job I’d done in 2007 finally paid out and due to exchange rate wobbles paid out 3 times as much as expected and I got to pay off all my grad school debt at once. To the government, no [big corporate bank].
Padraig
Standard deviations would be affected by grade inflation, since there’s a hard ceiling (A) that doesn’t get inflated. So, if grades inflate by one whole grade, then those who were getting Cs are now getting Bs, and are still discriminable against those who were getting Bs and are now getting As – there you’re right. But those who were getting As are now indistinguishable from those who were getting Bs, and I suspect that’s what he’s angry about.
Now go buy the poor humanities grad a steak.
Eunoia
Grade Inflation? FWIW, almost all people have more than the average number of legs.
Seriously though, if grades are ordinalised (A..F), then we’d need to know the histograms in each year and in which year the student was graded, if we are to compare across the years.
Samara Morgan
In typical de Bore fashion, freddie is completely clueless about the real problem, the colonization of America’s educational system by the “freed” market.
That is what the glibertarians at the atlantic are really whining about.
See Educational Mortgage.
@Big Baby DougJ: well you can also have skewness and kurtosis shift the shape of the distribution.
Like in grad school where a C becomes an F.
Samara Morgan
@RickD:
sure, it just possibly changes the shape of the distribution.
And i think this only happens in the humanities and lib arts where you can have subjective grading.
Harder to do in the hard sciences.
But then again, the glibertarians think a lib arts/humanities curriculumn is the only curriculumn there is.
Samara Morgan
@Big Baby DougJ: wouldnt that be a Poisson distribution? if A’s and B’s are the only passing grades?
look at what de Bore teaches…rhetoric, amirite freddie?
that would be highly subjective i think– not that i’ve ever taken a rhetoric class, lol.
Samara Morgan
snore city, Freddie. You teach fucking “rhetoric”.
You ARE them.
Samara Morgan
@Doctor Science: no he doesnt make sense.
shifting the mean changes the shape of the distribution with skewness and kurtosis.
freddie teaches ;;sneer;; rhetoric.
Simonee
I’m surprised that so many folks here are focusing on the writer’s tone instead of the meat of the argument that he/she is making. The entire article is gut-wrenching: these people are our future. And they sound hopeless. Helpless and docile.
I’m sure racking up tens of thousands of dollars in debt (that cannot even be discharged through bankruptcy) while receiving collection letters during one of the most horrendous job markets in a generation would turn a lot of you into WATBs, too. It’s not exactly easy to owe $40k (or more likely, a whole lot more) before you’ve had the opportunity to work and earn a single paycheck. These things factor into family planning as well.
These are people who grew up during the 90’s, when there was a pervasive sense, particularly among the middle class, that there was a fairness to the system.
Of course they’re bitter.
David Fud
One thing I haven’t seen mentioned is Restriction of Range. In the data set we are talking about, two direct restrictions have happened. First, the restriction upon entering college, and second, the restriction on a large percentage of failing students. Both of these undermine the predictive validity of GPA for any criterion. It also modifies the standard deviation of the restricted sample. See here for a basic example.
Lee Hartmann
Sorry to be late to this party, but:
a) the fact that the student clearly doesn’t understand what a standard deviation means (“uses the wrong semantics”) clearly indicates to me that he would not, actually, have easily become a computer scientist or engineer.
b) Assuming a normal distribution, Freddie is correct (and Doug J, etc)
c) In the intro classes I have taught, I have never seen a normal distribution, especially at the low end – a few students appear to fall asleep for the first couple of months, etc.
Cheryl from Maryland
I feel the student’s biggest mistake is thinking that grades alone should make the difference. Once you’ve got that degree, all are equal as far as school work. What the hell else has the kid been doing? As an owner of an art history degree, job hunting during St. Ronnie’s recession, I have been gainfully employed at a museum for close to 30 years. What helped me move forward wasn’t my grades or stats. It was that I studied skills outside of classes that museums needed.
Marc
@J. Michael Neal:
Or maybe, just maybe, that’s the grade you earned. In other words: no, fuck you.
FlipYrWhig
@Simonee:
I guess it depends on your meaning of “grew up.” Someone who graduated from college last year would have been born in 1990, and probably didn’t get much of a chance to internalize the idea of the relative fairness of The System in the ’90s.
Cermet
@PGfan: I am copying your post and letting my daughter read it – while she is over-the-top in academics in HS (4.0, taking AP physics, AP multi-variable calculus in 10th grade) I having been explaining to her the issue of college and jobs and the need to ‘get into’ a top college. She understands and is focused on this goal. This will very much help her realize the truth behind this goal … OF COURSE I have explained to her that any field, even no college is fine if that is what she really wants and has a realistic goal to live the related life/job. Also, college majors in art, music, or (I hate to say this) medical are all fine, too. I do explain you have to both enjoy and like your major. (aside – yes, she has a great social life/friends and time for outside activities. I press for fun, too. She is also earning her pilot’s license so I do feel she has a life and is exposed to other ideas besides mine.)
The choice is her’s but I also want her to be fully informed (unlike this persons’ approach and mistake – I do feel for them but I understood this issue when I was in college thirty some years ago … its not new or shouldn’t be to anyone) and I tell her that she needs to understand both the truth about our f’ed up system (and entitled elites) and the real world job market.
Thanks for the good insight!
Jado
I think the issue may be one of “actual standard deviation” in a grade-inflation situation versus “expected standard deviation” from the viewpoint of the prospective employer.
I.e., if an employer interviews a “B” student from Giant Football Team State University where there is a problem with grade inflation, and that “B” student is a drooling moron, then how do you convince the interviewer that your “A” is actually a real “A”, instead of just a small step up from drooling moron?
I saw this in my academic career, when the same useless idiots would show up for the next semester’s class and my friends and I would wonder how they passed the previous class, considering their obvious lack of ability or work ethic. And engineering colleges are not immune to grade inflation, as evidenced by the progress of the useless idiots I saw in my engineering classes. Thank god I didn’t have to interview anywhere those useless morons interviewed (different geographic bases for the job searches).
Cermet
@Padraig: Don’t forget what they do in HS – an A in a course is not the same for all courses with the same name: you have a bio course standard, Honors, (this next name is special to some systems) GT (gifted and talented) and, of course, AP course. Each figures in over all GPA in a different manner (so some 4.0’s are far ‘higher’ than others. Level of a course matters more than the content: Geometry Honors << Art GT. This, of course, is proper but the kids don't like it who take the math. Funny how they (including parents) lose sight of the value of non-technical courses and their value to the person. WE are becoming the chinese in some ways – as the elite press the middle class downward, education is one area that is still (partly) open to advance.
jayackroyd
@RickD:
DougJ already corrected my error, but thanks to you as well.
Upon reflection, though, I’m a little surprised, because my recollection from my college and graduate school education is that, with a very few exceptions (organic chemistry, for instance or, in your example, first year calculus) grades were not distributed normally, but skewed upwards–that is, a B- was the attendance and all assignments completed grade, with C and below reserved for just not doing the work.
And in any graduate level class, the minimum grade was generally a B, even for an undergrad attending a grad level course. I don’t see how such a grading “curve” can end up being normally distributed–the left hand tail would have to be significantly fatter than the right.
Of course, the college could make a grading policy that enforces normality–which is what you say happens where you’ve taught. Just like the IQ test, where normality is imposed on the raw scores. (Yes, the Bell Curve is an assumption, not a research result.) That has not been my experience, either at my east coast elite undergraduate school, or at my midwestern land grant graduate institution.
Mark
@Nylund: I disagree. I think you (and others) are focusing on the outcomes of a single class – and assuming that there aren’t underlying percentage grades that get converted to letters.
For a university of any significant size, and for students taking 15+ classes (I took 45 classes, 5-6 per semester), you have, for all intents and purposes, a normal distribution of GPAs. Even if the mean moved up a full letter grade in the last 50 years, it went from 2.0 to 3.0, so there’s still enough space between 3.0 and 4.0 to maintain the shape of the curve – he’s talking about differentiating himself from the average, not from the failures. His z-score is going to be the same under the new system and the old system, so standard deviations is the wrong language.
If I gave him a lot of credit, one point I would make is that grade inflation sometimes results in the ceiling being set fairly low, particularly in state schools. So too many good students get As, and so in, say a math class, the differentiator can be who makes an arithmetic error as opposed to who understands the material.
Samara Morgan
@jayackroyd: the assumption of normality mirrors the IQ bellcurve, ie test results and course work mirror the innate ability of the student.
But there is nothing to prevent a lib arts teacher in a subjective humanities class from giving As to 75% of the students.
Then the distribution of grades is no longer normal.
This is typical of graduate classes where anything less than a B is a fail.
But it isnt grade inflation.
Grade inflation is a pretty meaningless term in lib arts humanities classes that have subjective grading.
Glen Tomkins
Well, one lesson we can take from the original quote and the comments here is the home truth that in statistics, it isn’t usually the math itself that trips people up, it’s the application of the mathematical model to the real world. I don’t think that it’s so much a question of anybody being wrong, as it is of misunderstanding the application.
The author of the quote isn’t so much wrong on the statistics itself, it’s just that he’s muddle-headed about what grades mean, and what everybody understands grades mean. In particualr, he confuses the grading of aptitude vs achievement testing.
Assigning grades by normal distribution only makes sense in aptitude testing, and even there, only makes sense if it is assumed tha the aptitude being tested distributes normally. The assumption is that the degree of innate aptitude or ability is present randomly within the tested population, and so a successful test will give each subject a grade based on where they fall in that presumably normal distribution.
On an achievement test, which is usually what is going on in school, you certainly hope that the achieved learning doesn’t distribute randomly on a bell curve. If the teaching is doing anything worthwhile, it is pushing all the students towards the upper limit. And since the purpose isn’t to rank order the students’ aptitude, but rather to document their success or failure at learning certain material, you have limits and gradations set internally by the content, and not set by their place on a curve. It would be perfectly reasonable, and not indicate grade inflation, for all the students in a class to get As — if they all achieved to the level of A work. And it would not prove undue harshness if every single one of the lazy, stupid bastards in another class got the Fs their work deserves.
You choose the statistic appropriate to what you’re trying to quantify in the real world. You don’t edit reality in an insistence that it meet the conditions necessary for the application of some statistic.
Barry
@J. Michael Neal: “Just so long as I got my C- and didn’t have to take the class again.”
I sympathize, but you really need to look at remedial work. Difficulty with public speaking will hammer you in applying for and working in a management position.
David in NY
I sense (though I could be wrong), that many of the people who answered above never studied or used statistics in their lives, and are just humanities students who think, as usual, that they can bluff their way through the question.
Samara Morgan
@jayackroyd:
not exactly– normality of test scores is derived from the assumed normality of the bellcurve.
uniform grade inflation is just going to flatten the curve, giving it fatter tails.
the relative distances between individual scores are still going to be the same.
and like you say, this is less of a problem in the hard sciences, math, physics, chemistry, because grading is less subjective.
But giving As to 75% percent of the class would change the shape of the distribution from Gaussian to likely a Poisson form.
Some Guy
@Martin: That is actually really dependent on the institution and the relative size of the programs. You need to look at cost ratios. English departments, even large ones, are very cheap to run because salaries are often half of engineering and medical fields. And the overheard is minimal. The best way to look at it is dollars earned for the investment and English is very high in that ratio typically.
And you have to figure the return on indirect percentage at any given institution, the cost of building relations with community and industry partners, which eats up a huge amount of the soft money, and it gets pretty complicated.
Bottom line: large humanities programs with heavy teaching loads make shit-tons of money for their cost but seldom are invested in proportionally.
Samara Morgan
@Glen Tomkins: agree.
however, the “cut” classes like freshman chemistry and physics can be assumed to model the bellcurve because of sheer numbers.
My chem 101 teacher at a big midwestern semi-private uni said on the first day..
“my purpose here is to keep as many of you as possible out of med school”, and laughed sardonically.”
Samara Morgan
@David in NY: this thread has a definite humanities bias.
Freddie himself teaches “rhetoric” i believe.
Barry
@James: “I feel bad for the so-called Millenials, but jeebus, they aren’t the first graduating class to meet with massive unemployment. Join the crowd. ”
I don’t have a link handy, but there’s a chart showing percentage job losses over time, comparing all post-WII recessions. One measure of the overall effect is the area above the curve (a function of the percentage losses x time). Using that measure, the current crash has left *all* post-WWII recessions in the dust by a long shot, and is heading towards having an area larger than the rest put together.
Also, the percent of the working population which has been unemployed for 99 weeks or more is a record, going back to whenever they first started tracking this.
And as has been pointed out by some other bloggers (for whom I don’t have links) that for an ordinary recession, the economy would be in recovery mode, with a high level of GDP growth, and massive (re-)hiring.
We aren’t seeing any of that, and the GOP is making d*mn sure that we won’t.
It’s not as bad as the Great Depression, but it’s far worse than anything that the USA experienced after that.
David in NY
@Samara Morgan:
I’m more a humanities type myself, but I know enough to shut up when I don’t have an answer to a technical question.
BTW, I think that @Glen Tomkins:’s comment
is a really good point.
Samara Morgan
@David in NY: And look at the source of all the whining– the Atlantic.
i think freddie would like to work there someday with Young Conor and McMegan.
thus the link. :)
Samara Morgan
@David in NY:
and trust me, its the math at least half the time. humanities students are required one stat course if that. look at the exemplars at the Atlantic– How many posts has Levenson made on McMegan’s lack of math skills? and she’s supposed to be an economist?
/rolls eyes
All statistics is simply applied mathematics. its finding the curve in the data and fitting a model and predictive trend analysis.
Glen Tomkins
@Samara Morgan: I certainly agree that many university classes are graded as if for aptitude, and not achievement. But the role of the large numbers enrolled in such classes is not mainly that large numbers allows the use of a curve, it’s that the large numbers means that there is no effective teaching, students are pretty much on their own, and the learning outcomes depend almost entirely on their individual aptitudes. So, sure, using a normal distribution is fine in that situation, it accurately describes the reality being modeled. It’s just that the reality being modeled is pretty sad.
The other part of why mathematics that fits aptitude testing is appropriate for such classes is suggested by the pre-med angle you mention. There are subject that have been turned into tests of aptitude for purposes quite foreign to their content. Organic Chemistry, for example, is almost completely irrelevant to medicine (I put in that qualifier because I’m an Internist, so knowledge of anything under the sun — baseball, for example — could conceivably help me figure out what the hell is wrong with my patient.), but it’s on the MCAT. The MCAT is itself mostly an aptitude test misunderstood as an achievement test. It tests your aptitude for digesting vast unwieldy masses of unpalatable raw trivia, and nothing is better than Organic Chemistry, precisely because it has nothing to do with medicine, at testing the willingness to lay siege to immediately irrelevant and inherently uninteresting fortresses of knowledge. So you have Organic Chemistry classes overrun with premies who have absolutely zero interest in the subject, but must throw themselves at it anyway. Whatever idea that you could set content-based levels of achievement in mastering Organic — which would be a very unimportant, low pressure, intro course if it were taken mainly by students with an interest in being Organic Chemists — goes out the window when most of the students are only interested in the subject as a way to prove their aptitude by doing better than than everybody else.
Barry
@Jado: Seconding Jado here – the problem is not really standard deviation, but what does an ‘A’ mean? What does a ‘B’ mean – worked hard in a tough program, or drifted in an easy one?
And the discrimination problem becomes more acute if one assumes that people tend to focus on the letter, and not the +/-; e.g. A+,A,A- are all scanned as ‘A’.
Barry
@Samara Morgan: “How many posts has Levenson made on McMegan’s lack of math skills? and she’s supposed to be an economist?”
She’s got an MBA, and poses as an economist – well, actually she poses as somebody who knows her *ss from a hole in the ground on a lot of topics, none of which is true.
And I have not yet heard of her making ‘math errors’ in a direction which is not Chicago Glibertarian BS Artist.
Barry
@Samara Morgan: “All statistics is simply applied mathematics. its finding the curve in the data and fitting a model and predictive trend analysis.”
I hate to be harsh, but as an actual working statistician, if that were true my life would be much less complicated.
Tell that to any statistician, and see what their reaction is.
Samara Morgan
@Glen Tomkins: agree totally on O-chem.
:)
but in the hard sciences testing can model aptitude, especially in very large classes because of the law of large numbers.
The glibertarian position at the Atlantic is a four-year humanities degree is the only degree out there, because they are all first culture intellectuals or wannabe first culture intellectuals.
I think grade inflation in a humanities curriculumn is pretty irrelevent. like jackroyd (i think) pointed out the Z-scores remain the same, the relative distance between students.
Now, if i saw the guys at Cosmic Variance and Bad Astronomy complaining about grade inflation in physics, i would think of it as a real problem.
Coming from the Atlantic?
Not so much.
Samara Morgan
@Barry: haha, pardon. when i was a statistics TA i told that to my class to ramp down the fear factor.
when i was a calculus TA i told my students do your homework the day it is assigned and you will pass.
you are correct, a working statistician has much more to contend with.
Samara Morgan
@Barry: let me clarify my position.
none of my students ever aspired to be working statisticians– they were humanities or biology majors that needed the stat credit to graduate and hadn’t had math since ninth grade….stat majors went to a different class that had math prerequisites.
And when i was a calculus TA i told my students do your homework the day it is assigned and you will pass. Both statements were true in the context of the uni.
:)
Samara Morgan
@Barry: The point i was making is that the Atlantic is staffed by first culture intellectuals wholly lacking degrees in the hard sciences. I doubt any of the “angy young millenials” freddie is talking about has anything but a four year lib arts degree.
And i was thinking of the gastritic calculator and other epic McMegan math fails, but w/e. you can google if you are really interested.
its interesting that she just has an MBA. isnt she the economics editor, blogger, pundit…w/e she is?
Samara Morgan
for the newbies, this is a classic de Boer post, flogging one of his pet stalking horses.
Recent history, where freddie staunchly defends the value of a four-year-lib-arts degree, with YES! you guessed it– Bad Statistics.
freddie, why dont you take a stat class? it would do you a world of good.
Samara Morgan
@Glen Tomkins:
its Darwinian. Survival of the fittest.
VAdem
This student may want to look at what employees need, and see if the local Community College offers courses in it and take a course in it while networking. Community Colleges are often more focused on getting people training for jobs.
an old instructor
I have been a college instructor in two state systems over the last 20 years. Here are some observations from that time:
— Cheating is epidemic and students do it without remorse. Their only regret is getting caught. That’s the main way to get through without really trying.
— If I had actually applied what I thought were anything like remotely collegiate standards the core-level courses that I taught, I would have assigned failing grades to 70% or more of my students. My colleagues felt generally the same way. In reality, of course, we could not really fail entire sections — even when it was clear that fully half our students didn’t even bother to buy the books for the course. Many of these students were the same ones who skipped entire weeks of lecture. Those of us without tenure (and increasing numbers of faculty on all campuses are adjuncts who will never have it) had our continued employment depend on positive student evaluations. There’s no more effective way to get good evals than to be an easy grader. But regardless of one’s ratings, administrations took a dim view of high failure rates; low retention means diminished revenues. So we were effectively bound to pass at least 60% of our students.
Which means that our essayist is largely in the right, even if he expresses his frustrations poorly.
Barry
@Samara Morgan: (re Megan)
“its interesting that she just has an MBA. isnt she the economics editor, blogger, pundit…w/e she is?”
She’s blithe, ideologically committed writer; that’s all that The Atlantic wants, for the most part. I have no idea how James Fallows keeps his job, and how they ever hired Ta-Nesi Coates – and I mean that as a compliment to them.
Samara Morgan
@an old instructor:
well no.
read the de Boer post i linked in #142.
freddie is unhappy because the Whining of the Angry Young [lib arts] Millenials about how they cant get jobs is contradicting one of his (and to be honest, most glibertarians) core themes– that a 4 year degree in the humanities/lib arts will get you more success and higher wages.
McJulie
@Samara Morgan:
And what’s your point? That the only degrees anyone should be getting are in engineering/math/science?
Also, I’m not entirely sure I understand how getting an English lit degree fits into the libertarian mindset. In my experience, it’s the engineers who are the libertarians and the humanities grads who are the liberals.
Samara Morgan
@McJulie: my point is freddie is WRONG.
grade inflation only occurs in humanities/libarts, and it has null effect on job acquisition.
libertarians are anti-empirical first culture intellectuals whose reasoning procedes from first principles, like freddies. Their reasoning is based on Dead White Guy Phailosophy, a dated and anachronistic Failed Paradigm.
your anecdotal data is neither useful or statistically significant.
why yes, if you want to be assured of getting a job.
Freddie says that four year lib arts degree will increase SES, but im not so sure.
the stats he uses to support that are pretty crappy.
and of course we have the empirical example of buy-me-a-steak whining guy.
no job for him.
:)
Samara Morgan
@McJulie: That the only degrees anyone should be getting are in engineering/math/science?
do you think ANY of freddies Angry Young Millenials has a computer science or engineering degree?
i sure dont.
:)
TheF79
@Barry:
As an economist who publishes regularly using statistical methods, there is nothing McArdle does that is remotely close to economics. That someone with an English undergrad and an MBA holds the title of “Business and Economics Editor” at a major publication is nothing short of astonishing. Her understanding of economics (as opposed to business) is like the Platte River, a mile wide, an inch deep, and very often full of shit.
dcdl
@Capri: Your comments were exactly what I was thinking.
Also, all I can say is Liberal Arts degree and you expect a job without pursuing an advanced degree or networking. What a freaking moron. He sounds like he has no life skills and maybe there are other reasons he’s not getting a ‘good’ paying job.
The whole grade spread, it might not be right or whatever, but that is life better get used to disappointment.
Samara Morgan
c’mon guys…its the Atlantic for cripes sakes.
in the new austerity economy no one wants to pay for bloviators that worship the freed market and Dead White Guy Phailosophy.
i guess these people usta go into advertising or daytrading….or academe.
but those days are gone and they aint coming back.
one McMegan is an excess of McMegans.