A Skeptical Look at Education
Details
This meetup discussion will focus on the economist Arnold Kling's "null hypothesis" for education - i.e. that no educational intervention makes a long-term, scalable, replicable difference:
http://www.arnoldkling.com/blog/the-null-hypothesis-in-education-is-hard-to-disprove/
At first glance, this seems absurd -- people who are taught a skill are obviously better at it than those who are not. However, this makes more sense when you think of Kling's hypothesis as if it contained several unstated assumptions:
(1) A child's educational outcomes are largely a result of underlying factors that schools can't do much to improve, such as the child's IQ & personality traits and the parents' socio-economic status.
(2) The control group is not feral children or children who've suffered parental neglect, but rather "unschooled" children who are cared for by parents but not formally enrolled in a school and who are able to seek out various sources of knowledge (often via the internet) & learn independently.
(3) Once you get beyond basic literacy & arithmetic, forcing students to learn things they're not interested in may produce short-term knowledge gains but this will dissipate if the skills are not used regularly.
(4) While voluntary self-study may not be optimal, it still tends to be at least as good as the optimal teaching methods once they're scaled up to be taught at public schools nation-wide since this almost inevitably results in diluting them.
(5) Much of the benefit of schooling -- especially college -- is in the social prestige attached to a degree rather than the actual knowledge learned.
Essentially, Kling's null hypothesis is similar to research you may have seen in Steven Levitt & Stephen Dubner's bestselling book Freakonomics that indicated parenting styles don't seem to have much of a lasting effect on children and that school choice (in the Chicago public school lottery) showed little to no effect on children's educational outcomes:
http://www.gradesaver.com/freakonomics/study-guide/summary-chapter-5
We'll start our discussion by looking at whether or not education can raise IQ or beneficial character traits like self-control or "grit". Next, we'll look at research that compares the test results & life outcomes of homeschooled children versus those that attended public school. Third, we'll compare U.S. educational spending to other wealthy, developed countries and try to assess whether or not funding is related to American students doing relatively poorly in terms of testing. Fourth, we'll look at some surveys of research to see what type of educational interventions show significant results. Lastly, we'll try to assess whether or not a college education yields beneficial results aside from "signaling" to employer's that a person with a degree has a higher level of intelligence & perserverance.
Our discussion will be based around the 5 topics listed below. I certainly don't expect discussion participants to read every article or watch every video linked below. Please just try to look at one from each section. I'll sketch some notes under each article & video to give you a general sense of the main points.
Our 2-hour discussion will be divided up so that we spend roughly 25 minutes discussing each section, except I'll just budget about 20 minutes for the 3rd section (since "fade out" is probably less controversial).
I. CAN EDUCATION SUBSTANTIALLY RAISE I.Q. OR TEACH BENEFICIAL PERSONALITY TRAITS LIKE "GRIT"? HOW MUCH EFFECT DO I.Q. & GRIT HAVE ON LIFE OUTCOMES?
- Neerav Kingsland, "Can We Raise IQ Through Schooling?" (short blog post)
https://relinquishment.org/2016/06/23/can-we-raise-iq-through-schooling-fluid-intelligence/
Kingsland summarizes the recent research on IQ & education, and says that IQ is based on a combination of crystallized and fluid intelligence, and we have evidence that schooling can increase crystalized knowledge but much less evidence that schooling can increase fluid knowledge. That said, a new study from Sweden claims that an increase in years of schooling raised IQs for low-income / farmer families. (However, the extra year of schooling reduced emotional control across most classes. The study's authors posit that this might be because education instills less emotional control than actually working, and that the rolling out of an additional grade might have stressed the educational system and negatively impacted school culture.)
- Kristin Ozelli, "Should Grit Be Taught and Tested in School?" (short article)
https://www.scientificamerican.com/article/should-grit-be-taught-and-tested-in-school/
Ozelli interviews psychologist David Yeager on "social & emotional learning" (SEL) and Angela Duckworth on "grit" (apparently a facet of Conscientiousness) and asks whether these traits can & should be taught in schools and measured through standardized tests. They both urge caution because tests that measure noncognitive traits were designed for use in research settings and not as part of a high-stakes measure of student growth & school performance. These traits can absorbed by children from their social environment, but intentional efforts are far less successful than unintentional ones. The results of Duckworth's program to teach grit were modest: the program raised grades over just one marking period and only among low achievers. They also note that noncognitive learning progress can be hard to assess because SEL programs tend to produce more self-critical students, who will rate themselves lower on SEL questionnaires (similar to "Valley of Despair" on the popular version of the Dunning-Kruger curve).
- David Z. Hambrick & Christopher Chabris, "Yes, IQ Really Matters" (medium-length article)
Contrary to critics of the SAT who argue that it measures nothing but the ability to score high on a standardized test, Hambrick & Chabris point out that numerous studies have shown that SAT scores predicted college GPA in both the 1st and 4th years, as well as graduation rate. A student who scored in the 95th percentile on the SAT or ACT was about 60% more likely to graduate than a student who scored in the 50th percentile. SAT scores even predict annual income in the years afer college. (Critics also claim the SAT score is biased towards wealthy students, but the correlation is weak -- only 0.25 -- and expensive prep courses typicall only add about 30 points to a student's score.)
They point out that the SAT & other standardized tests measure something real & stable, i.e. a student's IQ. may in fact be the only such opportunity for students who graduate from public high schools that are regarded by college admissions offices as academically weak. Evidence from nearly a century of empirical studies suggest that IQ is the single best predictor of job training success and accounts for differences in job performance even in workers with more than a decade of experience. IQ predicts not only income & employment, but also health & longevity, and the latter effects make sense because studies show higher-IQ individuals engage in more positive health behaviors, such as deciding to quit smoking. They note that personality traits like Conscientiousness also correlate with positive life outcomes, but meta-analyses indicate that these factors do not have a larger effect than IQ does on measures of academic & job performance.
- Dan Vergano, "I.Q. scores don't predict success as much as motivation" (short article)
Vergano details research by Angela Duckworth that concedes that one of the most robust social science findings of the 20th century is that IQ predicts a broad range of life outcomes, but individual differences in low-stakes test motivation are much greater than currently assumed in the social science literature. Their research found that test motivation can be increased by paying students for good test results. After adjusting for the influence of test motivation (presumably connected to Conscientiousness), they found that IQ was still significantly stronger than was test motivation for academic achievement, but the predictive validity of IQ scores for non-academic life outcomes was significantly diminished.
II. HOW DO UNSCHOOLED CHILDREN COMPARE TO THOSE TAUGHT IN PUBLIC SCHOOL AS ADULTS?
- Luba Vangelova, "How do Unschoolers Turn Out?" (medium-length article on Peter Gray & Gina Riley's research)
https://ww2.kqed.org/mindshift/2014/09/02/how-do-unschoolers-turn-out/
Gray & Riley surveyed 75 adults who were unschooled children. It's a small sample size and limited in value because the respondents self-selected and thus were probably more likely to have had a positive experience, but Gray & Riley argue that it at least provides evidence that it's possible to be unschooled as a child but go on to be successful in life. All but three of the 75 respondents felt the advantages of unschooling clearly outweighed the disadvantages. Very few had any serious complaints against unschooling, and more than a third of the respondents said they could think of no disadvantages at all, although 21% cited social isolation as a problem. Only 11% said they felt behind in one or more academic areas (most commonly math), which they overcame by applying themselves when the need arose, and only two felt their learning gaps hindered them from succeeding in life. Only three people were very dissatisfied overall, and in all three cases, the respondents said their mothers were in poor mental health and the fathers were uninvolved, and 2 of the 3 mentioned having been raised in a fundamentalist religious home. Overall, 83% of the respondents had gone on to pursue some form of higher education, and almost half of those had either completed a bachelor’s degree or higher, or were currently enrolled in such a program.
All survey respondents were also asked about their job status, and 63 answered a follow-up survey about this. More than three-quarters said they were financially self-sufficient; the rest were either students, stay-at-home parents, or under the age of 21 and launching businesses while living at home. An unusually high percentage (about 50%) of the survey respondents went on to careers in the creative arts, and high number of respondents (half of the men and about 20% of the women) went on to science, technology, engineering or math (STEM) careers. Just more than half of the respondents were entrepreneurs, and this overlapped considerably with the creative professionals.
- Rachel Coleman, "The Homeschool Math Gap: The Data" (medium-length article)
https://www.responsiblehomeschooling.org/the-homeschool-math-gap/
In 2013, researchers Robert Kunzman and Milton Gaither surveyed the evidence on home schooling (not unschooling) and found evidence of a stronger verbal abilities but a deficit in math compared to formally schooled students. They note home schooled students show more intrinsic motivation in reading but less intrinsic motivation in math. They also point out that it's typically easier for the average parent to teach children to read and take them to the library than it is for them to teach a sequential and increasingly challenging math curriculum.
In terms of SAT scores, homeschooled students outscored public school students on both verbal & math sections, but once we correct for background factors (e.g. parental education, gender, ethnicity, and county-level poverty rate) homeschooled students scored better than predicted on the verbal section and slightly worse than predicted on the math section. Note that comparatively few homeschoolers take the SAT -- 2% of all students were homeschooled in 2001, but only 0.5% of students who took the SAT in 2001 were homeschooled.
Studies conducted to date have tended to find that homeschool alumni have higher college GPAs than other students. This is likely at least in part a result of homeschooled students’ lower college attendance rates - only the most prepared go to college. Homeschooled students at Austin College took less math & science classes (0.8 and 1.9) per semester than traditionally schooled students (1.9 and 3.2), and while they achieved a slightly higher average GPA overall they had lower GPAs in math and science courses (2.58 and 2.62) than their conventionally schooled peers (2.72 and 2.65).
III. HOW MUCH OF THE MATERIAL THAT STUDENTS LEARN IS COMMITTED TO LONG-TERM MEMORY? DOES LEARNING "USELESS" INFORMATION STILL SOMEHOW HELP STUDENTS?
- Bryan Caplan, "Fade-out, Teacher Quality, and Summer Learning Loss" (blog post on Jacob, Lefgren & Sims study)
http://econlog.econlib.org/archives/2010/11/fade-out_teache.html
- Sarah D. Sparks, "Studies Explore Reasons for 'Fade-Out' Effect" (medium-length article)
http://www.edweek.org/ew/articles/2016/01/27/studies-explore-reasons-for-fade-out-effect.html
- Bryan Caplan, "Low Transfer of Learning: The Glass Is Half Full" (blog post)
http://econlog.econlib.org/archives/2012/08/low_transfer_of.html
IV. ARE THERE ANY NEW EDUCATIONAL METHODS THAT CONSISTENTLY SHOW SIGNIFICANT RESULTS - PAYING FOR GRADES, TEACHER TRAINING, ONLINE EDUCATION, GAMIFICATION? WHAT ARE THE BARRIERS TO IMPLEMENTING THEM ON A NATIONAL SCALE?
- Steven Levitt, "The Grades Experiment: Freakonomics Movie" (video - 3:00 min.)
https://www.youtube.com/watch?v=8MElbRKprs8
- Arnold Kling, "The Null Hypothesis Compounded" (short blog post on Elizabeth Green's research on teacher training)
http://www.arnoldkling.com/blog/the-null-hypothesis-compounded/
- Arnold Kling, "The Null Hypothesis and Online Education" (short blog post on the Bell & Federman study)
http://www.arnoldkling.com/blog/the-null-hypothesis-and-online-education/
- Ben Betts, "The 2-Sigma Problem" (video - 11:51 min.)
https://www.youtube.com/watch?v=wqLiLH6Sjnw
V. IS A COLLEGE EDUCATION WORTH THE COST ONCE WE FACTOR OUT THE "SIGNALLING" ADVANTAGE OF A COLLEGE DEGREE? IF SO, SHOULD EVERYONE GO TO COLLEGE, OR CAN ONLY SOME PEOPLE REAP THE BENEFITS?
- Tyler Cowen & Alex Tabbarock, "Is Education Signaling or Skill Building?" (video - 9:31 min.)
https://www.youtube.com/watch?v=MvWnyUT7vPk
- Charles Murray, "Down With The Four Year College Degree" (medium-length article)
https://www.cato-unbound.org/2008/10/06/charles-murray/down-four-year-college-degree
A Skeptical Look at Education