from Hacker News

Peter Thiel’s Graph of the Year

by marchustvedt on 12/31/13, 8:27 AM with 91 comments

  • by Someone on 12/31/13, 9:01 AM

    That graph is a bit of a lie. Student debt is summed over all students, so it grows if the student population grows (which it has; 2013 saw a decline, but that will not factor much in this data yet). It also integrates over many years. That makes the total debt look more threatening.

    It would be more reasonable to show a per-year, per student debt (people graduating in year X had an average debt of $Y; people leaving without graduating had an average debt of $Z)

    Such a graph probably still looks bad (median income has dropped, after all), but it will be more informative.

  • by ForHackernews on 12/31/13, 9:53 AM

  • by w1ntermute on 12/31/13, 9:53 AM

    I think these trends are largely irrelevant to the Thiel Fellows. After all, most of them come from elite universities and could easily find a high paying job after graduating. If Thiel wants to put his money where his mouth is, he needs to find a way to solve the problem of the bottom 50% not finding a job. And I don't think that spawning more startups will help - just look around the startups you know - how many of the people there would have been unemployed if they hadn't joined a startup?

  • by ntoshev on 12/31/13, 9:00 AM

    That chart compares median income with total student debt. Also the scale for median income is not zero-based. The point is still valid, but this widespread use of misleading charts is awful.

  • by pmarca on 12/31/13, 10:19 AM

    Compare and contrast:

    http://www.bls.gov/emp/ep_chart_001.gif

    Summary for the graphically challenged: unemployment in the US is low (<5%) for people with bachelor's degrees and up, and high (>5%) for people with less than a bachelor's degree. Unemployment rises directly in proportion to how little education one has.

  • by jmathai on 12/31/13, 9:12 AM

    The cost which bothers me the most about the current university landscape is the amount of potential we as a society waste.

    So much of our population goes straight from high school to a 4 year college program. That's while they're between the ages of 18 and 22.

    Unless you're studying to be a doctor, lawyer, nurse, engineer, etc. then that's 4 years of lost productivity.

    I think most professions could be trained for in under a year + on the job training.

    Even if the argument is that college is largely social growth for the student it's insanely expensive if we need 4 years to do it.

  • by pwang on 12/31/13, 10:37 PM

    This is a bad visualization because it is hinting at a visual congruence that isn't there. When plotting things with two Y axes, you have to be really careful and honest about what you're trying to look at.

    When you use a line graph, visually, you're displaying a trend (in the most naive way possible). When you have two Y axes, then you need to ask, "What is a valid way to compare two trends?" Remember, a trend is intrinsically about slope. If you arbitrarily adjust the min/max axis value, then you can arbitrarily adjust the appearance of the trend.

    Here is a fixed version that shows the actual numbers on full, 0-based scales: http://i.imgur.com/xiS2OZq.png

    Hopefully this shows you how deceitful Thiel's plot is, even though the axes are "clearly labelled". Iconic perception trumps reading numbers.

    Here is a year-over-year plot: http://i.imgur.com/diDVNeE.png

    An alternative, more meaningful visualization, would be to show % change from a baseline, in this case the 2003 value. That would be far more honest and informative. And here it is, as a bar plot, which is more effective for comparing values along a common scale. (Since we have computed percent change, we have a common scale.)

    http://i.imgur.com/TQd2u22.png

    *Note: I just read the numbers off of Thiel's plot and don't have the source data, so this plots are approximations. I can share a simple IPython notebook of this if anyone really cares.

  • by 001sky on 12/31/13, 9:05 AM

    uni is a binary [0,1] with several tiered exceptions. value to the degree is primarily as a sorting / funnel. as a result, its not really going to be linear. The problem with the current pricing is that what was once a "hack" (using the degree as a legal iq proxy) is now so far out of wack with the information content of the degree itself. After all, the distribution of iq doesn;t really change. so nobody is in aggregate all that better off by paying more for an iq test. there is sureley some productivity gains by "better" educaton over time (and these accrue to both students and to economies), but again...the rents are extracted from this production by people in structural positions of power (ie, not "productivty" in the normal sense; these are people who have the ability to get paid for not destroying value), long before they accrue to the more productive economic agents.

    This disconnect between marginal productivity and actual wages for marginal work is evident in other data, eg

    http://www.nytimes.com/2013/12/30/opinion/america-in-2013-as...

  • by guard-of-terra on 12/31/13, 9:45 AM

    This is a failure of middle class.

    Why people go into higher ed? Because that's how you keep in the middle class. Why do they want to be in it? Because it's either this or poverty. Why do debt rise? Because they are desperate. Why are they desperate? Because there is less and less jobs in middle class and even worse and worse condition outside. When the music stops most of young people won't have a chair to sit on. At least in the USA.

  • by JunkDNA on 12/31/13, 2:55 PM

    I learned a long time ago from Edward Tufte to be wary of the two axis graph, especially when different scales are used. One can manufacture just about any point using one of these and still be telling the truth.

  • by dpierce9 on 12/31/13, 2:14 PM

    Not to pile on, but even if you disaggregated student loans on a per-individual basis, this graph would still be misleading because it does not include a line for the median wage of non-bachelor holders over the same period. In general, to know whether something is a good investment you must at least consider the next best alternative.

    Even this suggestion wouldn't be that great. To make a simple apples to apples comparison, we would need the graph of salaries of representative samples of comparable bachelors and non-bachelor holders. I don't really know what 'comparable' means in this context but it might be something like comparing the salaries of kids who graduated from college with kids who didn't go to college but could have (and maybe received comparable on the job training).

    One might also think that college affects, for instance, your likelihood of getting a job but we wouldn't want to get too complicated.

  • by dsundstrom on 12/31/13, 9:21 AM

    The graph is misleading.

    The implication is that a bachelor's degree is costing more and generates less income, but what the graph shows is "total student loan debt" not the "cost of a bachelor's degree".

    Technically this graph could be a good thing if we could graduating significantly more people who make more than someone without a degree (but 10% less someone who earned a degree 5 years ago), since more people would be better off.

    Other things to note: The time period is arbitrary and mostly shows the impact of the recession over the past 5 years. The graph shows total student load debt which includes everything from vocational programs to PHD students in public and private institutions. The graph has two independent scales so the slopes are arbitrary.

  • by ontheinternets on 12/31/13, 2:42 PM

    As long as the lifetime earnings are substantially higher, you should be willing to put up the NPV of your annuity (income). As long as you make even one dollar more than you would, you are better off.

    If you compare the entire education investment to the return during year one in the workforce, it isn't going to look good.

    If you compare the entire education investment to the impact it has on that person's lifetime, e.g. lower unemployment, easier time getting into grad school if the economy goes sour, easier time retraining, and just general high earnings, then the investment makes a bit more sense. I assume if this were a business, Thiel would wonder why education is underpriced.

  • by lzhou on 12/31/13, 9:42 AM

    Aside from the total debt vs median wage issue - the chart also doesn't mention that THE recession that happened. What happens in a recession? 1) parental wages and/or job security are lower [ie, more debt needs to be taken on] 2)median wages coming out are lower. That may be partly the cause of the relationship.

    Median wages dropped (starting in 2007) because of a recession, and we are still in the adjustment process of that (ex. on the wage side). We notice that post popping of the 2001-2002 bubble, the median wages actually increased into 2007.

    I don't have all the facts, but the chart is misleading.

  • by wch on 12/31/13, 3:53 PM

    In addition to the _total_ debt vs. _median_ income issue that others have mentioned, this is a classic case of how 2 y-axes on a graph can mislead. Or it can be used to tell any story you want. Thiel's graph "shows" that debt is taking off and way exceeding income. Here's a different version of the same data which "shows" that debt and income are converging: http://i.imgur.com/OBCyfVb.png

  • by bonemachine on 12/31/13, 7:35 PM

    Lies, damned lies.. and charts with not only non-zero, but mismatched baselines!

  • by mathattack on 12/31/13, 2:40 PM

    I like the story of the graph a lot more than the graph itself.

    The two things I would like to see are:

    - Per capita debt per person who attended college. (Or perhaps who graduated college) This would answer an implied question of "What if we're just getting more people going to college?"

    - The salary legend should start at 0. This would put the relative movement of salary in a more accurate context.

    I don't think fixing these changes the story of "The long term cost of college is going up, while the short term benefits are going down." but when I see tricks out of "How to lie with Statistics"[1] my BS detector goes up.

    [1] http://www.amazon.com/How-Lie-Statistics-Darrell-Huff/dp/039...

  • by gejjaxxita on 12/31/13, 12:48 PM

    I don't really agree with the people saying the income axis should be zero-based. The income drop is not tiny, it's 8%, and a zero-based axis wouldn't show that.

  • by kayoone on 12/31/13, 10:38 AM

    While the underlying message might be true, this chart is constructed to look fatal.

    1) Income scale is not zero based and makes a small change seem enormous

    2) Student debt should be per person per year, not a total sum which is influenced by the number of students and numerous other factors.

  • by safeerm on 1/2/14, 6:02 AM

  • by ecocentrik on 12/31/13, 10:04 PM

    This trend of overlapping unrelated data sets into a single 2-axis graph is completely disingenuous. At least this graph doesn't mix logarithmic and linear scaling for values on the y axis.

  • by JeffL on 12/31/13, 9:32 AM

    It might be interesting to see the same chart, but just for the subset of people getting STEM degrees. (Science, tech, engineering, and math.)

  • by michaelochurch on 12/31/13, 3:28 PM

    People are attacking this graph for the difference in scales, but the US population didn't increase by that much in 9 years (2003 to 2012): only about 10%. It's a shame that people are missing the forest for the trees here. If there were a 10-15% rise in student debt, that would be reasonable in the context of US population growth. But it's 200%.

    Yes, the axes have been chosen in such a way to exaggerate the (still valid) point. This isn't any different from a visual model of the solar system that downplays the (vast) differences between the planets, relative to their sizes. Is it "misleading"? Only if it's presented and interpreted as a scale model. In this graph, the axes are clearly labelled, informing the reader of what's going on.

    Thiel is right that there's a problem. The Satanic Trinity (housing, healthcare, tuition) that is killing the middle class is fundamentally extortive at this point, and extortionists are known for rapidly ratcheting up their demands. What's happening is that, as the middle-class breaks down, people are panic-buying the credentials and connections that might keep them alive in this game of musical chairs.

    I don't support Thiel's solution (as an aside, he's an investor in Knewton, one of the most pro-establishment ed-techs out there) but he sees the problem with a clarity that few do.