Welcome to the Profit of Education website. Continuing the conversation begun in the book Profit of Education, we discuss the latest economic evidence on education reform.

The achievement gap in education: Racial segregation versus segregation by poverty: What would Dr. King say?

My most recent post on the BROWN CENTER CHALKBOARD at the Brookings Institution.

“In elementary schools, Negroes lag one to three years behind whites, and their segregated schools received substantially less money per student than do the white schools.” – Dr. Martin Luther King, 1967

“[W]hite students score an average of 1.5 to 2 grade levels higher than black students in the average district.” – Reardon, et al., 2019

Dr. Martin Luther King is celebrated today as a civil rights leader and one of the great orators of the 20th century. He was also a keen commentator on social issues—one who believed in bringing scientific research to bear on such issues. King linked questions of race and class as well. In this light, I want to discuss new research about the impact of segregation on today’s education achievement gaps. Surprisingly, this evidence strongly suggests segregation’s effect on current racial gaps are largely due to segregation by poverty, rather than segregation by race. There is some complexity to this finding that requires some explanation, but first, some reminders of history.

King fought first to eliminate de jure racial school segregation, which at the time was largely a Southern phenomenon. Before and during the civil rights movement, the “equal” part of “separate but equal” was essentially a fraud. As an example, a study by the NAACP that examined Georgia school spending in the 1920s reported per pupil spending of $4.59 for Black students as compared to $36.29 for white students. In the Northern states, racial disparities in school funding were also common. In the 1960s, King wrote:

“Statistical evidence revealed in 1964 that Chicago spent an average of $366 a year per pupil in predominantly white schools and from $450 to $900 a year per pupil for suburban white neighborhoods, but the Negro neighborhoods received only $266 per year per pupil.”

Times have changed. Times have improved. De jure segregation is gone, although schools remain de facto segregated to a large extent. According to a new study by Reardon, Weathers, Fahle, Jang, and Kalogrides on segregation’s effects on racial achievement gaps, segregation reached its peak in 1968, declined through about 1980, and has remained more or less stable since then. In other words, it has been four decades since progress toward more integrated schools flat-lined. There has also been much progress toward equalized spending. This finding comes from a current study on racial spending differences: “Black … total per pupil expenditures exceed White total per pupil expenditures by $229.53.”

However, care is required in interpreting “equal spending” for two reasons. First, we mostly know about spending at the district level and not how the spending is divided up among schools within a district. Second, as Reardon, et al., write, “[H]igh-poverty schools typically have greater financial needs, given their larger shares of students identified as needing special education services and their often larger shares of English Learner students.”

So, as indicated in the opening quotes, while racial achievement gaps have been reduced since the days of King’s campaigns, the remaining gaps are still large. There are many reasons for persistence in the achievement gap, including that the legacy of separate and unequal may cause segregation in the past to have continued effects today. Looking directly at today’s situation—how important is today’s segregation per se? Black students mostly go to school with other Black students. Black students also mostly go to school with low-income students. Do either of these forms of segregation contribute to the racial achievement gap? The new work by Reardon, et al., “Is Separate Still Unequal? New Evidence on School Segregation and Racial Academic Achievement Gaps,” suggests that it is primarily poverty segregation rather than race segregation that accounts for segregation’s effect on the achievement gap.

The authors use a massive dataset that covers achievement in grades 3-8 in about half the school districts in the United States; notably, these districts include 96% of all Black public school students. The outcome variable studied is the Black/white achievement gap on test scores in math and English language achievement. (The results discussed below are for third graders.) The central question studied is the extent to which the achievement gap is explained by four differences in the average experience of Black versus white children measured by exposure to: minority schoolmates; poor schoolmates; minority neighbors; and poor neighbors.

The authors control for the usual measures that help explain the achievement gaps, and then focus on segregation per se. So the authors are telling us how much segregation matters over and above other differences that explain achievement.

The authors’ first result is that differences in exposure to minority schoolmates appears to matter a lot—if taken alone. According to my calculations using the authors’ reported outcomes, a one standard deviation increase in this measure accounts for about 9% of the total achievement gap. If exposure to minority neighbors is accounted for, the effect is even a bit larger, with the effect of schoolmates being about twice as large as the effect of neighbors. In other words, the apparent effects of racial segregation are pretty much what you’d expect.

But the authors point out that schools are also highly segregated by income level, specifically by the fraction of students living in poverty. And measures of racial segregation and “poverty segregation” are highly correlated. (Depending on the exact measures used, the reported correlation coefficients are between 0.82 and 0.93.) Is the apparent effect of race really an effect of poverty? According to the authors’ work, yes.

The authors repeat their analysis including both exposure to minority schoolmates and exposure to poor schoolmates. The effect of exposure to minority schoolmates completely disappears, while a one standard deviation increase in the Black/white gap in exposure to poor schoolmates explains about 10% of the total achievement gap. The evidence, then, is that differences in exposure to poor schoolmates rather than differences in exposure to minority schoolmates are what matters.

Finally, a horse race is run in which all four factors are included. Having minority schoolmates still doesn’t matter and having poor schoolmates still does matter. There is a small effect of exposure to minority neighbors and no effect of exposure to poor neighbors.

In summary, the authors find that the role of segregation in explaining today’s racial achievement gap comes from the fact that Black students are much more likely to go to schools with impoverished classmates, and not from the fact that they go to largely Black schools.

To be clear, this does not mean that poverty is the only thing that matters, and that race is now inconsequential. Notice that about 90% of the racial achievement gap is still unexplained in the authors’ model, meaning there’s still plenty of room for race to be impacting gaps through other policy differences both inside and outside of the classroom. Even more important, integrating by income can significantly reduce achievement gaps—and this will also mean integration by race, due to the strong relationship between the two.

If King were still alive today, would he agree that the consequences of segregation, in part, have more to do with poverty than with race per se? We can’t know, of course, where King would have come down on the question of race versus income segregation. But there is no doubt that he believed the problems were inextricably linked. (In fact, King linked solving the problems of racism to societal changes that uplifted all people.) Take this quote from a talk King gave in 1967:

“Let us turn first to the evil of racism. There can be no gainsaying of the fact that racism is still alive all over America. … The second evil that I want to deal with is the evil of poverty. Like a monstrous octopus it spreads its nagging prehensile tentacles into cities and hamlets and villages all over our nation. Some 40 million of our brothers and sisters are poverty stricken, unable to gain the basic necessities of life. And so often we allow them to become invisible because our society’s so affluent that we don’t see the poor. Some of them are Mexican Americans. Some of them are Indians. Some are Puerto Ricans. Some are Appalachian whites. The vast majority are Negroes in proportion to their size in the population.”


Note: Most quotes attributed to Dr. Martin Luther King Jr. throughout this post are from King’s book, “Where Do We Go From Here: Chaos or Community,” originally published in 1967 by Harper & Row.

Share
Posted in Uncategorized | Leave a comment

Leave a Reply to Chad Aldeman (@ChadAldeman) Cancel reply

Your email address will not be published. Required fields are marked *

Advice matters: Faculty advisers and college student success

My most recent post on the BROWN CENTER CHALKBOARD at the Brookings Institution.

Academics and policymakers are actively looking at creative ways to help college students succeed. While increasing college access and enrollment is an important first stage, too many students matriculate but fail to thrive. Bluntly, too many never graduate.

Like most faculty, I have many informal “advising conversations” with undergraduates. I like to think these conversations do some good. I’d never thought much about whether formal advising systems do much and whether they can contribute to overall student success. But there is now some solid evidence that good advising makes a real difference.

Authors Serena Canaan, Antoine Deeb, and Pierre Mouganie offer a very nice piece of work in “Advisor Value-Added and Student Outcomes: Evidence from Randomly Assigned College Advisors.” (In the interest of full disclosure, I like to brag about have a slight conflict of interest regarding two of the authors: Canaan is a UCSB alumna and I’m currently a member of Deeb’s dissertation committee.) The authors use data from a natural experiment at one university to show that good advising can really matter.

First, a little background about college advising offers some context. About half of U.S. colleges have freshman advised by full-time faculty, accounting for more than three-quarters of all new undergraduates. Interestingly, the use of faculty as advisers varies among type of institutions. Most baccalaureate-granting institutions use faculty, but research universities tend to have a professional advising staff. (Faculty serve as official advisers at only about a fifth of research universities.)

Before looking at the papers’ details, here are some of the headline results (see Figure 1 below). A one standard deviation increase in the estimated value-added of the freshman-year adviser:

  • Increases freshman GPA by 5.7 percent of a standard deviation.
  • Increases 4-year graduation rates by 2.5 percentage points.
  • Increases the probability of high-ability students enrolling in and graduating from a STEM major by 4 percentage points.
Figure 1. Effect of Higher Adviser Value-Added on College Student Outcomes

These effects of having a good adviser are meaningful; not huge, but remarkable considering the relatively few interactions that most advisers have with undergraduates.

What makes these estimates so convincing is that the data comes from a natural experiment that eliminates a common weakness seen in many value-added studies. The data is sourced from a university in which students are randomly assigned to advisers; thus, the estimated relationship here is causal, which is a high bar for most other studies to clear. (The administrative assignment process is intentionally random, and the authors ran statistical checks that show the randomization worked.) In most studies of student outcomes, we worry that we are missing some unobservable factor that is correlated with what we’re studying; a teacher having a class in which more parents than usual buy outside tutoring for their child would be an example that could compromise estimates. Random assignment eliminates such concerns.

The university involved in the study was the American University of Beirut (AUB), which raises a question of what scientists call “external validity;” in other words, do we think the results are relevant to the typical American college? The answer is “yes.” A little background about AUB explains why. AUB was founded by American missionaries in 1866. Classes are taught in English and AUB degrees are registered with the New York State Board of Regents. Enrollment is around 7,000 students and admissions depend mostly on SATs and grades—average SAT scores look a lot like average American SAT scores. As the authors put it, “AUB is comparable to an average private nonprofit 4-year college in the United States.”

The results cited above demonstrate that, on average, good advising is valuable. The researchers then delve deeper and find is that the results are heterogeneous—having a good adviser is more important for some students than for others and who benefits most is situational. Improvements in freshman-year GPA are larger for high-ability students than for low-ability students, though both groups gain. There is not a noticeable difference by student gender.

Also, top advisers are more likely than others to guide their students into selective majors (STEM and business). This is especially true for high-ability students, defined as students entering with SAT scores in the top quartile. It appears that these results reflect pointing top students to STEM majors and a smaller effect for students being pointed to business, where the increase in entering a business major is also true for students with lower SAT scores.

The authors investigate the mechanism for the various improvements. What they tease out from the data suggests that freshman-year advisers are particularly good at helping students get good freshman-year grades. (By the way, good advisers don’t steer students into easy courses—the authors checked.) These higher grades likely increase student confidence, leading to better performance later on, and directly increase the chances of making it into a selective major.

Improving college student achievement and graduation rates has become a major concern. In the U.S., only about a third of college students complete their four-year degree within four years at public institutions, and about half do so at private nonprofits. The results from this study suggest that more or better faculty advising might be a promising avenue to improve these numbers, although it’s obviously important to ask where the needed resources will come from. It’s worth noting in this regard that AUB had a student/faculty ratio of 11, while the U.S. ratios are about 15.5 at public institutions and 10.2 at private nonprofits. Arranging more and better faculty advising would be beneficial, as this study shows, but would be especially challenging at large public universities. In light of the numbers above showing that large universities rely on staff rather than faculty for advising, it’s worth remembering that the authors of this study don’t make comparisons between faculty and staff as advisers. So, while we’ve learned that having a good faculty adviser makes a big difference, more research will be needed to know if having a good staff adviser would have the same effect.

Share
Posted in Uncategorized | Leave a comment

Leave a Reply to Chad Aldeman (@ChadAldeman) Cancel reply

Your email address will not be published. Required fields are marked *

Community college ‘free-for-all’: Why making tuition free could be complicated

My most recent post on the BROWN CENTER CHALKBOARD at the Brookings Institution.

Presidential candidates are pushing for “free” community college for all; Douglas Harris described some of the early details here on the Chalkboard earlier this year. Greater access to community college is a clearly overdue policy for reasons we also talked about here in July, and it’s worth thinking through some of the details. (As an aside, some candidates propose four years of free college, others that college should be “debt-free.” We’ll leave discussion of those proposals for another day.) Currently, we have no national policy on community college attendance: States have very much gone their own ways. Policy diversity has some definite advantages. But things are so different across states that making community college free nationwide is going to be tough—implementation details are going to require some thought.

Here’s the first issue: Some states have chosen to go all in on community colleges, while other states do very little. Will this be a political problem? Will politicians from states with low community college attendance be reluctant to support subsidizing students in other states? Hopefully not, as states with low community college numbers today are also the states with the greatest likelihood of growing their attendance. What’s more, community college attendance doesn’t follow a simple red state/blue state pattern.

Right now (well, per the latest data from the Digest of Education Statistics in 2017), 22% of community college students are in California—which has 12% of the nation’s population. Another way to say this is that 33% of college-aged folk in California attend community college, as compared to a national average of 20%. Here’s a chart showing that same two-year school enrollment as a fraction of the college-aged population across states.

Figure 1 - Fraction of college aged population enrolled in public 2-year colleges

The other states that look something like California in this regard are pretty small in terms of population. (New Mexico and Wyoming actually have a higher ratio than California.) Texas, though, is a large state that is also well above the national average. Arizona, Iowa, and Virginia are also high. The state that stands out on the low end is Florida.

A complicating wrinkle in thinking about free community college is that the boundaries between community college and four-year schools are sometimes blurred. In some states, some bachelor’s degrees are offered by community colleges. (This may account for the low two-year enrollment reported for Florida.) Does that mean free tuition would not be covered in such schools? Or does it mean that a bachelor’s degree is covered if offered by a community college, but not if the same degree is offered in a “four-year school”? In some states, this is a big issue. In other states, the issue doesn’t arise. But it’s one more complication that will require careful thought and quite likely careful political negotiation.

A second issue is that states charge very, very different tuition levels. Compared to the status quo, very different subsidy levels will be needed across states to achieve zero-tuition nationwide. The national average annual tuition at public two-year colleges is $3,200. California is way below that. Texas is relatively low as well.

Figure 2 - In-state tuition and required fees at public 2-year colleges

Here, too, the breakdown is not especially red versus blue, which probably helps with the politics. Nonetheless, some thought will be required to figure out how to cover tuition both in New Hampshire, where the current price tag is $7,300, and in California, which charges $1,300.

Some of the proposals for free tuition include a requirement for cost-sharing by states. States currently differ significantly in how much they spend per student, raising a third issue. I’ve calculated total expenditures on public two-year colleges and subtracted off tuition. The national annual average is $6,100. (Note: While I’m confident that states vary wildly in what they spend, don’t put too much weight on the numbers for a particular state. Also, the latest data is three years old, though that probably shouldn’t matter much.)

Figure 3 - Expenditures in public 2-year colleges less tuition

Mississippi spends almost $11,000 above tuition costs, while Virginia spends only $1,500. The situation is made more complicated by the fact that costs of real estate, construction, salaries, etc., vary so much across states. Simple formulas about cost-sharing may be difficult to reconcile with varying levels of existing contributions. (By the way, picking up the cost of tuition will increase total spending on community colleges by about a third, very roughly. Of course, free tuition will increase demand for community college—that’s kind of the idea, after all—which will further raise the required level of funding.)

Looking ahead, some thought should be given to how states will respond strategically to various proposals. For example, Joe Biden’s plan calls for “the federal government covering 75% of the cost and states contributing the remaining obligation.” (Fellow candidates Cory Booker, Kamala Harris, and Amy Klobuchar also call for free community college, while Bernie Sanders and Elizabeth Warren would extend free tuition to four-year public colleges as well.) Right now, California is a low-tuition, high-expenditure, high-participation state. Should California raise its $1,300 tuition to the $7,300 in Vermont— thus increasing tuition revenue—and then contribute a fourth of that ($1,825) to keep tuition free while the federal government contribute three-fourths ($5,475)? Students won’t care, since one government or another is picking up the tab. The extra revenue from Washington, D.C., would free up a lot of money that California is now using to subsidize its very large community college system.

The admirable goal of a federal program is to make community college available to Americans wherever they live. If that happens, we are likely to see community college within reach of a much greater number of students across the country. That will itself diminish the differences that exist today, but not on day one. Figuring out the details is going to require some good technocrats. And on the political side, there should be considerable appeal across both red and blue states. Perhaps this might be a good venue for bipartisan cooperation.

None of this is an argument against the federal government finding a way to make community college tuition-free. It is an argument that figuring out the details will take some work.

Share
Posted in Uncategorized | Leave a comment

Leave a Reply to Chad Aldeman (@ChadAldeman) Cancel reply

Your email address will not be published. Required fields are marked *

Free college for all Americans? Yes, but not too much.

My most recent post on the BROWN CENTER CHALKBOARD at the Brookings Institution.

Promising free college has obvious political attraction for presidential candidates. From my perspective, free college is the right idea, but some of the promises go too far; in fact, exactly twice too far. There is a very strong argument for promising two years “free” at public colleges; the argument for a four-year free ride—not so much.

In a nutshell, my argument is that America has long supported free K-12 education and that two years of college is pretty much what K-12 used to be, with most Americans now obtaining at least some college education. For most people, getting some college is the key to the middle-class—creating a case for public support during the first two years after high school.

Douglas Harris has a great Chalkboard post that will bring you up to date on what candidates have said, and will give you many of the pros and cons for various higher education proposals. Harris also talks about popular support for different proposals in another Chalkboard post. Here, I’m going to stick to the narrower question of why paying for two years of public college is the right goal.

In the picture that follows I present educational attainment numbers for the American population from 1950 through the most current data. I’ve divided the population into those with a high school education or less, colored in red, those with some college but less than four years, blue, and those with four or more years of college, colored in green. (The category “some college” includes vocational training such as certificate programs at community colleges as well as more purely academic courses.

The vertical axis shows the fraction of the population with a given educational attainment—categories are stacked at each point in time to sum to 100%.

You can see that in the early post-war years, most Americans had no more than a high school education. Today the majority has picked up at least some college. I’ve also drawn in a line at the 50% point on the theory that the middle is “middle class.” Up until about 1990, the median American had a high school education or less. That’s fallen to about a third, and the median American now has some college. So “some college” has replaced high school or less as the standard educational attainment. If it made sense in the past to provide free public education through high school, then doing the equivalent today means paying for some level of college. Note that the green (4+ years of college) area is still way, way above the 50% line. Only about 30% of the population currently attains that much education. So such an appeal to the past does not provide an argument for four years of free college.

ed attainment over time

There are many motivations for increasing education, but probably the strongest is that it leads to better jobs and higher incomes. The next figure links educational attainment to income, plotting median income in each educational group against the overall national income distribution. In “the old days,” the median person in both the high school or less and the some college categories earned near the middle of the national income distribution. That’s still true for “some college.” The middle person with some college education gets to be right in the middle of the national income distribution. But getting to the middle is now much more difficult for those in the “high school or less” category. The middle person in the bottom group reaches just above the bottom third of the national income distribution. So in terms of income, “some college” has replaced “high school or less” as the middle-class norm.

median income

Some presidential candidates propose free community college, others advocate that all public college should be free. One reaction to the latter position is that we shouldn’t be subsidizing college graduates as they typically end up with higher incomes than the rest of the population. Look at the green line in the figure above. Not only does the average person with four or more years of college have a higher income than most people—they have a much higher income. You can argue that the 70th percentile of income is still middle class. (In the United States, it seems that everyone below the top 0.1% thinks of themselves as being in the middle class.) But the fact is that the income in the top educational attainment group is a lot higher.

The argument that “some college” is the new “high school” and should be similarly free makes sense to me, but this logic doesn’t extend to a free ride for four years.


Data from census and American Community Survey from IPUMS USA, University of Minnesota, www.ipums.org.

Share
Posted in Uncategorized | Leave a comment

Leave a Reply to Chad Aldeman (@ChadAldeman) Cancel reply

Your email address will not be published. Required fields are marked *

Los Angeles Times Op-Ed

Me in the LA Times on raising teacher pay.

https://www.latimes.com/opinion/op-ed/la-oe-startz-teacher-raises-20190417-story.html

Share
Posted in Uncategorized | Leave a comment

Leave a Reply to Chad Aldeman (@ChadAldeman) Cancel reply

Your email address will not be published. Required fields are marked *

As more women graduate from college, the teaching profession becomes more female

My most recent post on the BROWN CENTER CHALKBOARD at the Brookings Institution.

One of the great accomplishments of the late 20th century was to bring women onto a more equal footing in the labor market. Salaries became more equal. Employers opened up jobs for women. Educational opportunities became more gender-equal. And for college-educated women, all of this meant that careers outside teaching and nursing became possible.

One might think that as more career paths opened—even if not opened all the way—increased alternatives would have meant relatively fewer women in teaching. In fact, that is not what happened. As we celebrate Women’s History Month, we look at how one of the very few historically female-dominated professions has become even more so. Here, I briefly describe what did happen and offer my explanation as to why.

Unlike with most jobs, teacher salaries have long been based on a schedule that minimizes opportunities for discrimination. Little matters for salary other than years of experience and level of education. (Oh yeah, and whether you teach in a district with high or low salaries.) So women and men get almost equal pay. In the most recent available data, base pay for female teachers was 96 percent of the pay received by male teachers. Even if you include extra school pay above the base level (for extracurricular activities, coaching, and a few other things), women still earn 93 percent of what men earn. High school teachers get paid somewhat more than elementary school teachers, and men are more likely than women to teach at the secondary level. But women teachers have slightly more education than men do.

While the gender pay gap has not yet been fully eliminated, it is smaller for teachers than in most of the economy. As a rough comparison, in “management, professional, and related occupations,” the Bureau of Labor Statistics reports that women earn 73 percent of what men earn. While a large majority of teachers are women, a majority of these management and professional workers are also women. The difference in the wage gap is much smaller in teaching, making teaching especially attractive to women.

What about in the not-so good old days? It’s hard to get exactly comparable data, but reports from the 1970 census suggest the male-to-female earnings ratio for year-round workers with four years of college was 55 percent. (Something else from the old days: the reports are all separated by race. The 55 percentage is for whites only.) In contrast, teaching salaries have been approximately equal by gender, although not perfectly equal. The earliest year for which I was able to track base teaching salaries was 1987, when the salary gender ratio was 0.89.

In other words, women’s and men’s teaching salaries have long been almost equal. Outside teaching, earnings are far less equal—but they’re a lot closer together than they were in past decades. So what would you expect to happen to the relative number of women and men in teaching over this period? It used to be that few opportunities outside of teaching were available and women had equal-ish pay, which incentivized talented women to go into the classroom. Now with a more open world, the incentives for women to teach are lower, so relatively fewer will become teachers and the gender balance in teaching should become somewhat more equal.

At least that’s what I thought. It turns out that I was wrong and, in fact, the opposite has happened. Here’s a picture of the proportion of female teachers over time.

teachergender
Source: The Current Population Survey via IPUMS USA, University of Minnesota, www.ipums.org. Additionally: The National Center for Education Statistics’ Schools and Staffing Survey via The Department of Education.

The green line tells the real story. The share of teachers who are women rose, not fell, over the past three-plus decades. The proportion rose quite a bit through some time in the 1990s and then has edged up a bit more since.

Now some more explanation about the graph. The solid lines are taken from the Current Population Survey (CPS)—a representative survey of the U.S. population—and look only at full-time teachers. The dots are taken from the Schools and Staffing Survey (SASS), which is a survey of teachers only and does not reach back quite so far. The SASS trend is also upward, but somewhat less dramatically so.

We can split out a few more details using the CPS data. The blue line at the top shows the female share for elementary school teachers: unsurprisingly, heavily female. The percentage moved up some through the 1980s, then back down a little. Overall, it has been pretty flat over time. Elementary school teachers are overwhelmingly women, both historically and today.

The orange line at the bottom tells us the real change in the female share is in secondary school. Teaching secondary school used to be a majority male occupation, and today a solid majority of secondary school teachers are women. Notably, the majority of math and science teachers in grades nine through 12 are women. Probably a small piece of the story as to why women’s share in teaching has risen rather than fallen is that secondary school teaching jobs have become more open to women.

Here’s what I think is going on: While it’s true that the relative attractiveness of teaching versus other careers has declined for women versus men, there has been an enormous increase in the number of women with the basic required education to be a teacher—a bachelor’s degree. The picture below shows what happened over the last 40 years; the ratio of bachelor’s degrees going to women relative to men has risen by 60 percent. In other words, the greatly increased number of women in the qualified labor pool for teaching overwhelms any effects of changed incentives of teaching relative to other professions. (Ingersoll et. al. provide a very readable report on these and other trends in the teacher workforce; also see the nice summary in the Atlantic.) From an economics point of view, it is interesting that the quantity effect dominates the incentive effect.

bawomen2men
Source: The National Center for Education Statistics’ Schools and Staffing Survey via The Department of Education.

We should also ask whether it matters for the kids if their teachers are women or men. Hopefully, you already have a good idea of the answer, because Michael Hansen and Diana Quintero reviewed the evidence last July right here on the Chalkboard. Here’s the nickel summary:

“It is not clear that the plethora of females in the teacher workforce is worrisome in most circumstances—more female teachers may even be preferred in math and science classrooms.”

But I’d place one big caveat on the conclusion that teacher gender doesn’t matter. Most of what we know about the effects of teacher gender are about what happens in K-12. As you’ve seen in the chart above, there has been a massive change in college attainment of men versus women—and though everyone looks at four-year degrees, the change in associate degrees is just as large.

Interestingly, the large increase in the fraction of college degrees going to women is a worldwide phenomenon. And perhaps not coincidentally, women as a fraction of the teaching force has increased in other developing countries as well.

Fraction of women in teaching force in developed countries

Primary

Lower Secondary

Upper Secondary general

Upper Secondary vocational
OECD 2016836962
56
OECD 1996755750
42

The table shows the increase in the fraction of teachers who are women in the OECD. (Note that the comparison is not perfect, as the set of countries included changed a bit over these two decades.) What we see is that the change in the teacher gender ratio—including in secondary school—is not just a U.S. phenomenon.

The “coincidence” of a shift to relatively more women going to college during a period in which students also had relatively fewer male teachers is interesting. One possibility is that high school boys are in need of male role models to keep them on track to college. Another possibility is that it’s the decline in discriminatory barriers that has led to both more women teachers and to more women going to college. Evidence on how much either explanation contributes to the changes we’ve seen will have to await further research.

Share
Posted in Uncategorized | Leave a comment

Leave a Reply to Chad Aldeman (@ChadAldeman) Cancel reply

Your email address will not be published. Required fields are marked *

Equal opportunity in American education In memory of Martin Luther King Jr.

My most recent post on the BROWN CENTER CHALKBOARD at the Brookings Institution.

Today would have been Dr. Martin Luther King’s 90th birthday had he not been taken from us in 1968. Whatever position one takes on how to deal with America’s racial issues, most people of goodwill ascribe to the idea of equality of opportunity in education. I want to look today at a number of indicators of educational opportunity.

I emphasize educational opportunity, as distinct from educational outcome. Outcomes depend on opportunities offered by schools, on backgrounds that students bring through the school room door with them, and on choices made by those students. I will begin by reviewing a bit of what is known about differences in outcomes, in part because some differences by race are shockingly large. But let me be frank about why I will largely focus on equality of opportunity: There remain a number of people who want to blame the disadvantaged for their situation, and who do so increasingly openly. Consequently, I think it is important to identify circumstances where disadvantage cannot be attributed to the disadvantaged.

I wish to note three background items before we begin. First, the Equality of Opportunity project, led by a team of scholars at Harvard University, offers a wealth of data and analysis on inequality along many dimensions. Second, ProPublica has a nice piece on disparities in outcomes written by Lena V. Groeger, Annie Waldman, and David Eads. Third, I am going to limit my analysis to black/white differentials rather than looking at all disadvantaged groups. Partially this is for simplicity and partially as a reminder of the struggles led by Dr. King. (Though I suspect Dr. King would not agree with my narrow focus.)

LARGE GAPS IN OUTCOMES PERSIST BY RACE

Let’s begin with two quotes about findings from the Equality of Opportunity project, as discussed by Raj Chetty and Nathaniel Hendren. First, “The black-white gap is not immutable: black boys who move to better neighborhoods as children have significantly better outcomes.” What’s more is that the researchers are explicit that “environmental conditions during childhood have causal effects on racial disparities, demonstrating that the black-white income gap is not immutable.” The focus here on the Brown Center Chalkboard is education, but as we move on please remember that the educational system is not the only place there are inequities. As Chetty and Hedren write, “In 99% of neighborhoods in the United States, black boys earn less in adulthood than white boys who grow up in families with comparable income.”

While schools are not all that matter, let’s talk about some of the ways that schools do matter. Here is a graphic from ProPublica showing the achievement gap between black and white students across the nation.

ProPublica Miseducation
Source: ProPublica’s “Miseducation” project.

The big visual differences are in the South (still!), which is where Dr. King mostly fought. But the disparities are not just in the South. In LA Unified, the second largest school district in the nation, ProPublica reports, “Black students are, on average, academically 3.1 grades behind White students.” In Chicago, the gap is three grades. (ProPublica lets you check out your own local schools as well.)

Achievement gaps depend on opportunities inside school walls, and also on which groups end up being advantaged by those opportunities. ProPublica looked explicitly at racial gaps in participation in Advanced Placement and gifted programs. If a student isn’t enrolled in one of these programs, she or he can’t very well benefit from the program. Here’s the picture from ProPublica.

ProPublica's "Miseducation" project.
Source: ProPublica’s “Miseducation” project.

Again the South stands out, but, again, it’s not just the South. In NYC—the nation’s largest school district—white students are two-thirds more likely to be enrolled in AP courses than are black students.

OPPORTUNITY GAPS UNDERNEATH THE OUTCOME GAPS

Still, one could argue that the differences in enrollment in these advanced opportunities reflects something that students bring into school rather than the opportunities that the schools make available. To nail the point that inequality of opportunity is an issue, I want to look at issues of availability. To be pointed, if a student attends a school that doesn’t even offer an AP course, it’s pretty clear that the gap issue is one of opportunity rather than student choice.

A central issue, of course, is that black and white students don’t attend the same schools. To a great extent, schools remain de facto separate. (For discussion, see Chalkboard posts hereherehere, and here.) The most recent Department of Education Office of Civil Rights data collection reports information from almost 25,000 high schools (defined for our purposes as schools including a 12th grade). About 15 percent of students are black, but 40 percent of black students are in majority-black schools, and only 24 percent of black students attend majority-white schools. There was a time when many thought “separate but equal” was viable. Are opportunities in our still largely separate schools now equal?

Only 36 percent of high schools where the majority of students are black offer a calculus course. In contrast, 60 percent of majority white schools offer calculus. It’s hard to see how this makes for equal opportunity.

As a practical matter, to take calculus in high school—assuming it’s available in your high school—it helps to have gotten a head start on earlier math courses. The U.S. Department of Education’s Office of Civil Rights asks schools whether Algebra 1—normally a ninth grade course is available to seventh or eighth graders. This head-start offering is available in 42 percent of majority-black schools as compared to 66 percent of majority-white schools. An alternative comparison sounds not quite as bad (but still not good): 77 percent of white students are in schools offering early algebra while the corresponding number gets up to 64 percent for black students.

Perhaps I am hung up on the importance of math courses. After all, I am an economist, and in economics and other STEM fields, getting an early start on calculus before going to college is awfully valuable. But math isn’t everything. Let’s look at a couple of more general indicators.

The vast majority of teachers in the United States are certified. But schools that are disproportionately black have a noticeably disproportionate share of teachers who are not certified. Sometimes that’s intentional. Some uncertified teachers come through Teach for America or a school district academy and perform well. Of course, schools that receive Teach for America teachers are typically pretty disadvantaged in a variety of ways. So whether you think that uncertified teachers are less qualified or you just think that the presence of uncertified teachers is a signal that the school environment is more challenging than where you’d like to send your kids … well, neither is good. Here’s the figure showing the relation between the fraction of students in the school who are black and the fraction of teachers who are not certified. The relation between student race and teacher certification is quite clear.

It is a well-established fact that brand new teachers do not teach as well as teachers with more experience (on average, that is). Being a classroom teacher is just a very, very hard job that requires practice. Educational opportunities are therefore better for students assigned teachers who are not just starting off. Who is more likely to get a rookie teacher?

Here’s a figure showing the fraction of teachers in a school in their first year against the fraction of the students in the school who are black. Students in schools that are under 10 percent black have about 6 percent first-year teachers. In schools that are 90 percent or more black, that fraction doubles to almost 12 percent.

EQUAL OPPORTUNITY STILL ELUDES US

We all know that educational outcomes—test scores, high school graduation, college attendance, school discipline, and so on—differ by race. Some part of those differences reflect what students bring to school. But the data above also shows that the opportunities offered black and white students have not yet become equal. Dr. King’s dream of equality of opportunity sadly remains, in the words of Langston Hughes, “a dream deferred.”

Share
Posted in Uncategorized | Leave a comment

Leave a Reply to Chad Aldeman (@ChadAldeman) Cancel reply

Your email address will not be published. Required fields are marked *

Humanities PhDs: After Graduation

Back in October I asked why universities continue to produce so many humanities PhDs in the face of the abysmal humanities job market. The flip side of that question is to ask why young scholars are willing to invest so much of their lives in earning a humanities PhD. I hope the answer is that it’s their love of the subject, because it surely isn’t because of the economic return.

Let’s begin with the economics of getting that PhD. The NSF surveys new PhD recipients and asks them how much debt they acquired in graduate school. The figure below shows that humanities (and art) PhDs fare worse than STEM students. Just over half of humanities students escape with a degree without having acquired more debt in graduate school. In physical sciences, math and computer science, and in engineering about 80 percent avoid graduate student debt. Probably more important, more than a quarter of humanities PhDs end up with over $30,000 in new debt. For life sciences the percentage is about half that and in other STEM fields the fraction with large debts is a quarter of the rate for humanities and arts.

[DS1] 

There is also the “opportunity cost” of being in graduate school. Graduate school stipends are a lot lower than pay after earning the degree. Humanities degrees take longer to get than do STEM degrees. On average, one to two years longer. That’s a lot more foregone earnings for humanities students.

How about employment on graduation? Unsurprisingly, humanities PhDs are less likely than STEM graduates to have post-graduation plans lined up. But the difference is not all that large. In the humanities and arts, 43 percent of students reported to the NSF that they had no definite commitment. In math and computer science that fraction was only 27 percent, and in engineering the same number was 36 percent. The NSF also asked doctoral recipients who had definite plans whether the next step was a job or postdoctoral study. Of the humanities grads who had plans, the next step was overwhelmingly employment—only 21 percent were headed to post-docs. In the physical and life sciences, 3 out of 5 students went the post-doc route.

The real difference after graduation? Among humanities and art students who do have jobs lined up, the vast majority are going to work for colleges and universities. In contrast, STEM doctorates very often go to work for business or industry. Working for a university is not a bad outcome (I mention in case my employer is reading this), but industry certainly pays better. Here’s a figure with the breakdown from NSF data.

[DS2] 

For those new PhDs with an employment commitment after graduation, the NSF surveys expected annual salaries. In life sciences, the median is $80,000. (I report male salaries here; salaries for women are 15 percent lower.) Median salary in physical sciences is $86,000. Math and computer science is $113,000. Engineering is $100,000, and in economics the median salary is $100,000.

In humanities and arts, the median expected salary is $52,000. So it’s the salary levels that make the biggest difference between the humanities and STEM fields.

The NSF data is informative about the very beginning of a career. To learn more about economic returns over a lifetime, we can turn to the American Community Survey. (One caveat here: The ACS identifies fields by undergraduate degree, so there is no doubt some minor error in classifying degrees.) Here is a graph of median income in the humanities and STEM at both the BA level and the PhD level.

[DS3] 

The basic message is that a student with a humanities PhD earns about the same salary as a STEM BA. Some of this is because of the different employment venues, but even within colleges STEM PhDs earn more than do those in the humanities. Humanities PhDs working at a college or university have a median income of $72,000. The STEM figure is about 13 percent higher.

I’ve made a graph of the distribution of salaries for PhDs who work at colleges and universities (Wonk note: the picture is from what’s called a “kernel density” estimate.) What you can see is that that humanities PhD salaries in higher education are mostly between fifty and a hundred thousand. In contrast, about a third of STEM PhDs earn over 100K.

[DS4] 

The salary difference is much more noticeable in industry. Here’s the picture:

[

About a third of humanities PhDs employed outside of higher education earn over $100,000. For STEM grads, the number is more like 60 percent.

Putting it all together, getting a doctorate in the humanities doesn’t seem to be a great dollars-and-sense investment. If it’s a labor of love—and if students heading into these doctoral programs are fully informed—then it can still be a perfectly sensible career choice. I’m not so sure about whether prospective students are “fully informed,” but on that there’s no evidence other than anecdote. What I am sure about is that the rest of us continue to benefit from the young scholars who choose to sacrifice financially to transmit understanding of poetry and history and philosophy to the next generation.


Share
Posted in Uncategorized | Leave a comment

Leave a Reply to Chad Aldeman (@ChadAldeman) Cancel reply

Your email address will not be published. Required fields are marked *

How many humanities Ph.D.s should universities produce?

My most recent post on the BROWN CENTER CHALKBOARD at the Brookings Institution.

If you have been anywhere near a university that trains doctoral students in the humanities, you will know that the market for new Ph.D.s is abysmal.

The plain fact is that we train far more humanities Ph.D.s than there are jobs for. In a recent year, the American History Association reported about 340 tenure-track job openings for historians, plus maybe another hundred jobs outside history departments that might recruit a historian. In the same year, American universities minted about 1,150 new Ph.D. historians. In economics, there were 2,650 new jobs and about 1,250 new Ph.D.s. The contrast is stark and largely typical of the situation in humanities versus STEM fields.

In part, this is not new. The “crisis” has been going on for an academic lifetime—at least since the 1970s. Academics who started their career when the Modern Language Association first reached the “melancholy conclusion” that, “should present trends continue, life … particularly in the humanities, could turn grim indeed,” well, those academics have now mostly retired.

I want to talk about some possible reasons why more humanities Ph.D.s are produced than there are jobs for, especially in light of how long we’ve been in this dismal situation. First though, I want to raise a concern that conditions may be moving from dismal to disaster. Here are four graphs showing the most recent 10 years of data for the number of bachelor’s degrees (in blue) and the number of doctorates (in red) across economics, English, math and statistics, and history. The count of the former says a lot about the demand for the latter.

Figure 1: Comparison of B.A.s and Ph.D.s granted in economics, English, math and statistics, and history, 2005-2015

Comparison of bachelor's degrees and Ph.D.s granted in economics, English, math and statistics, and history, 2005-2016
Source: Digest for Education Statistics. (Note: The right-hand y-axis refers to B.A.s, the left to Ph.D.s.)

Look at economics and math/statistics on the left side of the figure. The number of bachelor’s degrees has grown, and the number of Ph.D.s has pretty much grown in parallel. Now look at English and history on the right. For the first half of the graph, the number of B.A. degrees was flat while the number of Ph.D.s grew rapidly. More recently, the number of bachelor’s granted in a year has plummeted while the number of doctorates has leveled off. One wants to be careful about extrapolating from five years, or even a decade. Nonetheless, things don’t look good. Why do the gaps matter? Because a big part of the demand for Ph.D.s in an area is to provide teaching for students pursuing their bachelor’s.

The economic problem in the humanities market is too much supply chasing too little demand. While one may lament the too-little demand, we’re left with the question of why supply hasn’t adjusted. Why do universities keep training so many humanities Ph.D.s when their market is so weak? It helps to understand that decisions about the number of Ph.D. students to admit are partially at the discretion of individual academic departments and partially decided by the choice of university administrators about how much money to allocate to a given department. It’s worth pointing out that resource allocations change very slowly in universities (but that’s slow on a scale of years, not slow on a scale of decades.) Let me suggest some reasons why universities continue to churn out so many Ph.D.s in the humanities.

Graduate students are cheap labor for teaching undergraduates, grading papers, etc. The suggestion is that it’s cheaper for universities to hire graduate students than to hire faculty.

Maybe, but I am less than enamored of this explanation because of the arrangements of salaries in research universities—which is where Ph.D.s are trained. Teaching assistant (TA) salaries are often set across the board, so they are the same in humanities and STEM disciplines. (More or less, as STEM fields are more likely to provide some supplements to TA pay. The “across-the-board” pattern is generally stronger in universities where the teaching assistants are unionized.) On the other hand, STEM faculty salaries are a lot higher than humanities faculty salaries. So if it’s a “cheap labor” argument, the excess supply should show up in STEM rather than in the humanities because the faculty/TA salary gap is larger there.

Graduate students are necessary for research and scholarship. Universities bring in humanities Ph.D.s to keep humanities faculty productive.

I suspect there is something to the need for graduate students to support research. And that makes it a valid reason, at least it makes it a valid reason for the university.

Graduate students may not be necessary, but humanities faculty really like to teach graduate courses. Humanities departments accept too many Ph.D. students in order to justify teaching graduate courses.

This is the explanation that I mostly believe, and this is also the explanation that I have no hard evidence for whatsoever. Just to be clear, economists also often like to teach graduate courses. I’m not suggesting any difference in attitudes or desires among humanities faculty versus STEM fields. The difference is that after coursework and after dissertations STEM Ph.D.s get jobs.

Faculty in the humanities see keeping a healthy graduate program as part of their mission to protect the role of the humanities in the wider world.

There’s no doubt about the importance of the humanities in helping us all do a better job of thinking through the problems faced in the world outside the ivy walls. I’m less sure whether producing an excess of Ph.D.s helps promote the value of the humanities in general.

Whatever the explanation of how universities benefit from producing more humanities Ph.D.s than the market will support, the question remains: Is it fair to the Ph.D. students? The Modern Language Association recommends “graduate programs routinely send applicants … data on the placement of recipients of PhDs from the department in the last three to five years (names of institutions, numbers of placements, fields of specialization).” Do humanities departments routinely make this sort of information available on the web? (Economics departments typically do.) I did some spot checking. Harvard does. Yale does. Michigan does. (My university, UC Santa Barbara, mostly does.) At three other major public universities I checked—nope. My suspicion is that humanities departments that place well despite the abysmal market are more upfront about placements than are departments with less enviable records.

Nothing I’ve said here will be new to anyone in the humanities. What’s more, the Modern Language Association, the American History Association, and others deserve credit for recognizing the situation and doing what they can about it for decades now. The problem does not lie in particular universities; the problem is that each university’s reasonable behavior adds up nationally to more humanities Ph.D.s than there are jobs for. Everyone wants some other university to shrink production.

One more thing. Humanities Ph.D.s do find jobs. After all, one has to be very smart to get into a Ph.D. program, and very persistent and talented to earn the degree. Unfortunately, while humanities Ph.D.s can find jobs outside academia, the financial reward isn’t much at all considering the 6 to 10 years the Ph.D. takes. The National Science Foundation reports that the average annual salary for new humanities and arts Ph.D.s who do have a job is about $52,000 a year. A comparable salary number for people with a bachelor’s in any field, aged 30 to 35, with full-time employment is about $61,000. It’s a depressing state of the world.

Share
Posted in Uncategorized | Leave a comment

Leave a Reply to Chad Aldeman (@ChadAldeman) Cancel reply

Your email address will not be published. Required fields are marked *

It happens here, too: Lessons for universities on preventing sexual harassment

My most recent post on the BROWN CENTER CHALKBOARD at the Brookings Institution.

While school was out this summer, the National Academy of Sciences published the over 300-page report, “Sexual Harassment of Women: Climate, Culture, and Consequences in Academic Sciences, Engineering, and Medicine.” The report is what you would expect from the National Academy: lots of numbers and a couple hundred references to the scientific literature. Included are a number of positive suggestions for reducing sexual harassment in academia—more on some of these below—but the report has one finding that shocked me.

Worse than the private sector? 58 percent? (Reports of harassment of students by faculty or staff are also high, in the 20-50 percent range.) Despite surface appearances, sexual harassment in academia turns out to be a major concern. While the pervasiveness is a surprise, some of the causes are not. My observation is that, in academia, there are three elements that contribute to such pervasive sexual harassment:I think of universities as pretty decent places to work; while there are certainly individual squabbles—sometimes serious ones—academics are pretty reasonably behaved for the most part. But, shockingly, that’s not what the evidence says when it comes to sexual harassment. According to the National Academy: “[Among employees], the academic workplace … has the second highest rate of sexual harassment at 58 percent (the military has the highest rate at 69 percent) when comparing it with military, private sector, and the government, where a broad definition of sexual harassment is used.”

  • Power and silos
  • Institutional “see-no-evil”
  • Plain, old insensitivity

Something true in both the military and in the academy is that there are many relationships in which the power between two people is extremely unequal. In the private sector, usually the worst someone can do to you is fire you, and even then there are always other jobs. In academia, one person may be in a position to effectively end the other person’s academic career—not just interfere with their current position. (The power structure in the military as compared to civilian society is pretty obvious.) In most organizations, there is at least some avenue of appeal from perceived unfair treatment. In contrast, if I give a student a failing grade, tell a Ph.D. candidate that their dissertation isn’t acceptable, write a referee report judging that a journal ought to reject a submitted paper, or vote against someone’s tenure, the truth is there is little to no recourse.

The problem of power is reinforced in academia because organization is so “siloed.” When I’m advising a Ph.D. student, basically no one else knows what I’m doing and probably no one else on campus could judge whether the mentor/mentee relationship is reasonable, even if they could observe it. The fact that so much interaction is effectively private reinforces the risk of bad behavior where there is a power imbalance. Interestingly, the National Academy study finds the greatest incidence of sexual harassment happens in medicine, which is especially hierarchical: “Research on the medical environment reveals that overall ‘mistreatment’ is commonplace in all levels of the medical hierarchy, especially among medical school students, interns, and residents.”

On the second bullet point, the National Academy study says, “Too often, judicial interpretation of Title IX and Title VII has incentivized institutions to create policies and training on sexual harassment that focus on symbolic compliance with current law and avoiding liability, and not on preventing sexual harassment.”

If you are reading this from a university, when is the last time you heard about a significant punishment to a colleague for egregious sexual harassment? If you were the subject of sexual harassment (I’m hoping you find this a hypothetical question, but in light of the study’s finding I’m no longer so sure this is merely hypothetical for most readers), would you feel it worth the cost and risk to file a complaint? Here, there are two different issues. The first issue is whether the campus takes sexual harassment seriously. The second issue is whether the campus makes public the extent to which there are consequences of bad behavior. If you want to deter misbehavior, be sure that there are consequences and everyone on campus knows it. If you want victims to come forward, they need to know that complaints result in actions.

As a positive example of good reporting, the National Academy study praises Yale’s semiannual “Reports of Complaints of Sexual Misconduct”: “These reports are written to protect anonymity while also providing minimal descriptions and statistical summaries that reveal (1) the complainants and respondents role in the university …and (2) the status of the complaint (if the complainant decided to pursue a formal complaint, if investigation is pending, any action taken by the university after investigation.”

Most harassment stops short of coercion and assault. The National Academy cites a large study done at Penn State of faculty/staff-on-student harassment for female students. About 95 percent of misbehavior falls into the categories of “sexist hostility” and “crude behavior.”So reporting can be done. It’s not that hard. And frankly, making data public is an important step in bringing a university community together when improvements are needed. A university would likely prefer not to read about unsuccessful processes from an external source, such as the recent California state auditor’s report, “The University of California: It Must Take Additional Steps to Address Long-Standing Issues With Its Response to Sexual Harassment Complaints.”

So a little advice to faculty, staff, and teaching assistants who either don’t care how women feel or who—despite all the attention the subject has gotten—still don’t understand how women feel. Don’t make sexual remarks. The fact that women may laugh or go along doesn’t make it okay, because some of them won’t feel they have a choice, and you won’t know that. The issue isn’t whether sexual banter can be fun for everyone—the issue is that you won’t know if you’re making people uncomfortable (unless, of course, they file a complaint against you).

Keep inappropriate comments and “jokes” out of your professional life. Why? Because you know that you might be making someone uncomfortable when that’s not your intention at all. What’s more, you are setting an example. So keep it professional.

One final comment from the report for those who are onboard with improving the situation:

Taken together, the surprisingly sparse—yet robust—set of studies on sexual harassment trainings shows that trainings can improve knowledge of policies and awareness of what is sexual harassment; however, trainings have either no effect or a negative effect on preventing sexual harassment.

 

Given … that actions can be taken to inhibit sexually harassing behavior … and that changing attitudes is difficult, effort seems better spent on developing and using sexual harassment trainings aimed at changing people’s behaviors rather than on their attitudes and beliefs. Ultimately, it is individuals’ actions and behaviors that both harm targets and are illegal, not their thoughts.

Share
Posted in Uncategorized | Leave a comment

Leave a Reply to Chad Aldeman (@ChadAldeman) Cancel reply

Your email address will not be published. Required fields are marked *