Popular in Sociology

Culture and Society

The City as a Growth Machine: Toward a Political Economy of Place

New York University Harvey Moltoch

Culture and Society

Carceral Entrapments: Views from the Prison/Street Interface in India

Indian Institute of Technology Delhi Mahuya Bandyopadhya

Culture and Society

Pre-problem families: predictive analytics and the future as the present

University of Southampton Rosalind Edwards

Ethnicity and Racism

Were Muslim Barbarians Really Knocking On the Gates of Ontario?

University of Ottawa Natasha Bakht

Culture and Society

Poor People’s Survival Strategies

The University of Texas at Austin Javier Auyero

Ethnicity and Racism

Sociology of Whiteness

Arizona State University Monica McDermott

Ethnicity and Racism

Narratives of Harm and the case for detention abolition

Queen's University Sharry Aiken

Generating Workspace

Export as PDF

The speaker discusses their long-standing interest in educational inequality and shares insights from their academic journey over the past 25 years. Starting as an undergraduate working in classrooms with disadvantaged students, they later pursued graduate studies and researched the impacts of college attendance on various life outcomes. Their thesis revealed that students with lower odds of attending selective colleges reap substantial benefits from completing college, prompting a continued exploration in this area during their postdoctoral studies.

At UCLA, the speaker observed that first-generation college students and students of color often remarked on the transformative effects of college on their lives. This video video motivated them to examine civic engagement and the broader impacts of college education, compiling their findings into a book that highlights the long-term benefits of degree completion, particularly for marginalized groups.

The research utilized extensive data from national surveys, particularly focusing on how college completion influences socioeconomic status, job stability, civic engagement, and family formation. Statistical methods, including machine learning and matching analyses, were employed to assess outcomes for college graduates versus non-graduates.

The findings indicate that college graduation significantly benefits disadvantaged individuals, helping to reduce poverty, unemployment, and dependency on social assistance, while enhancing earnings and civic participation. The speaker emphasizes the importance of framing education not just in terms of individual gain but as a collective societal good, advocating for public investment in higher education to promote equal opportunities and strengthen communities.

In conclusion, the speaker argues that a broader understanding of educational implications is critical, especially in combatting flawed narratives surrounding the diminishing value of college degrees. They contend that increased access to higher education delivers long-lasting benefits, particularly for those who historically have faced barriers to completion.

Language:

Export as PDF

X-TIMESTAMP-MAP:LOCAL=00:00:00.000,MPEGTS=900000

00:00.345 --> 00:04.765
My interests in educational inequality really go back

00:04.765 --> 00:08.165
to my undergraduate days, some 25 years ago when

00:08.685 --> 00:10.565
I was working in classrooms

00:10.945 --> 00:12.885
and studied a d tracking program

00:13.075 --> 00:15.165
that supported disadvantaged students

00:15.165 --> 00:17.005
to achieve in high school and attend college.

00:17.945 --> 00:21.125
And then, as I pursued graduate studies at the University

00:21.125 --> 00:23.805
of Wisconsin, I continued these interests in educational

00:23.805 --> 00:27.405
inequality and also developed interests in causal inference.

00:28.105 --> 00:31.245
And when I was writing my thesis in graduate school,

00:31.325 --> 00:34.285
I uncovered this interesting pattern about how students

00:34.395 --> 00:37.085
with a low likelihood to attend, in this case,

00:37.445 --> 00:40.245
selective colleges have these really large returns

00:40.465 --> 00:43.965
and even the largest returns to attending than those

00:44.075 --> 00:47.085
with a higher likelihood, the more advantaged students.

00:47.265 --> 00:50.805
So I kept pursuing that, that field of study

00:50.905 --> 00:53.125
as a postdoc at the University of Michigan,

00:53.205 --> 00:54.925
I started working with a colleague Y

00:55.305 --> 00:57.125
and we looked at the effects

00:57.125 --> 00:58.885
of college completion on earnings,

00:58.905 --> 01:01.805
and again, saw this pretty striking pattern, a pattern.

01:01.905 --> 01:05.725
We started calling negative selection that students unlikely

01:05.785 --> 01:08.565
to complete college reap these really large rewards.

01:08.945 --> 01:10.765
And then in my early years at UCLA,

01:10.845 --> 01:15.325
I explored other outcomes, family formation pattern, civic

01:15.845 --> 01:20.805
participation, and saw some similar, patterns

01:20.825 --> 01:22.285
of, for these really large returns.

01:22.465 --> 01:25.085
And I learned a great deal from my UCLA students.

01:25.345 --> 01:28.325
So students generally expressed many benefits

01:28.325 --> 01:30.605
of completing college, but it was my first generation

01:30.885 --> 01:33.285
students, students of color, students

01:33.365 --> 01:35.605
who really overcame the odds against 'em

01:35.605 --> 01:38.045
to attain these high levels of education,

01:38.105 --> 01:39.565
who invariably kind of told me

01:39.565 --> 01:41.845
that college changed the course of their life.

01:42.465 --> 01:45.085
And many of them also told me they wanted to go back

01:45.085 --> 01:46.805
to their communities and their youth.

01:46.945 --> 01:49.805
And that motivated some work on civic participation.

01:50.545 --> 01:53.085
So the book was this effort to kind of compile

01:53.145 --> 01:57.085
and advance, the work that I'd been doing on variation

01:57.085 --> 02:01.325
and effects of college for, over 20 years now.

02:01.825 --> 02:04.565
And, and I wanted to show the ways that college is worth it,

02:04.645 --> 02:07.325
particularly for those students on the margin

02:07.425 --> 02:09.525
who may be choosing whether or not to attend

02:09.745 --> 02:12.885
and who face, you know, a lot of hurdles along the way.

02:13.425 --> 02:16.725
So my goal was to assess how long-term benefits

02:16.785 --> 02:20.165
of four year degree completion varies across the population

02:20.785 --> 02:24.165
and to look at a large range of outcomes, socioeconomic,

02:24.185 --> 02:27.925
family level, social assistance, civic outcome measures.

02:28.545 --> 02:32.165
So I draw on nationally representative longitudinal survey

02:32.235 --> 02:34.445
data from the Bureau of Labor Statistics.

02:34.475 --> 02:37.405
This is the National Longitudinal Surveys Program.

02:38.465 --> 02:40.885
Um, and I draw on two cohorts from

02:41.505 --> 02:43.365
the National Longitudinal Surveys

02:43.365 --> 02:44.405
and then look at the effects

02:44.405 --> 02:47.765
of college completion on these long-term life outcomes.

02:48.185 --> 02:50.925
Um, and so I conduct statistical analysis

02:50.985 --> 02:54.525
to assess the differential effects of completing degrees

02:54.705 --> 02:58.365
and comparing those who completed a four year college degree

02:58.385 --> 03:01.925
to those who didn't, and, and,

03:02.025 --> 03:03.645
and look at these various outcomes.

03:03.905 --> 03:07.045
And I use a variety of statistical approaches,

03:07.685 --> 03:11.045
matching machine learning models to, to do this.

03:11.785 --> 03:13.245
Um, and over the years I've worked

03:13.245 --> 03:15.805
with fantastic collaborators, both faculty

03:16.025 --> 03:18.845
and a lot of my graduate students to kind

03:18.845 --> 03:21.725
of inform these analyses and make the book what it was.

03:22.135 --> 03:23.765
First, as I was saying about the,

03:24.025 --> 03:25.885
the National Longitudinal Surveys program

03:26.065 --> 03:28.285
that's been ongoing for more than 50 years.

03:28.425 --> 03:31.325
So we have these great longitudinal data

03:31.435 --> 03:35.085
that looks at these two cohorts born in their early 1960s

03:35.305 --> 03:37.005
and early 1980s.

03:37.505 --> 03:39.765
And then I'm able to follow them

03:39.765 --> 03:42.005
because this has been ongoing for so long

03:42.005 --> 03:44.045
to follow these individuals over time

03:44.105 --> 03:45.725
and see how their lives unfold

03:46.215 --> 03:47.445
after they,

03:48.225 --> 03:50.285
either complete college or don't complete college.

03:51.185 --> 03:54.885
Uh, and then, so to assess the effects of completion,

03:55.065 --> 03:57.845
we wanna draw on these counterfactual comparisons

03:58.105 --> 03:59.605
of different outcomes.

03:59.705 --> 04:03.045
So the outcome when an individual completed college

04:03.145 --> 04:05.725
and the out kind of the, the outcome, when

04:05.725 --> 04:08.525
that same individual ideally did not complete college,

04:08.555 --> 04:11.125
that would be our, our best way to assess the effects.

04:11.225 --> 04:13.765
And in theory, we could observe, you know,

04:13.765 --> 04:16.405
the outcome had someone gone and not gone to college.

04:16.545 --> 04:19.725
But in reality, of course, only one state is observed.

04:19.865 --> 04:23.125
So we, we turned to these statistical approaches to

04:23.805 --> 04:26.605
estimate the unobserved counterfactual pathway.

04:27.195 --> 04:29.925
What would've happened had that person not,

04:30.235 --> 04:31.645
completed a college degree.

04:31.825 --> 04:34.085
So that's where I turn to these matching analyses

04:34.225 --> 04:37.205
and machine learning approaches to estimate effects.

04:37.225 --> 04:40.405
And then look how those effects vary across different

04:40.405 --> 04:43.405
subgroups of students and matching is pretty intuitive.

04:43.545 --> 04:45.645
We, we take someone who graduated from college

04:45.905 --> 04:48.405
and compare their life outcomes to someone

04:48.425 --> 04:50.725
who had the same estimated likelihood

04:51.265 --> 04:52.805
of graduating from college,

04:52.985 --> 04:56.045
but didn't actually complete a degree for whatever reasons,

04:56.865 --> 04:59.005
and, and then estimate the effects based

04:59.005 --> 05:00.165
on all these match pairs.

05:00.165 --> 05:02.005
So we take thousands of match pairs

05:02.065 --> 05:06.325
and look at the average outcomes, across 'em, then

05:06.325 --> 05:08.925
to make the, the book more,

05:09.225 --> 05:11.365
and even more intuitive and appealing.

05:11.385 --> 05:13.485
And, and with all these quantitative analyses,

05:13.605 --> 05:16.965
I constructed 12 matched vignettes to describe some

05:16.965 --> 05:18.405
of the narratives of both the college

05:18.505 --> 05:19.885
and non-college graduates.

05:19.905 --> 05:23.205
So this gave, you know, some names

05:23.345 --> 05:26.565
to these people from the National Longitudinal Survey data

05:27.025 --> 05:29.485
and how their lives looked similar before college

05:29.585 --> 05:31.885
and how they diverged after college.

05:32.065 --> 05:33.805
And you know, what it, what it kind of looked like.

05:34.855 --> 05:39.075
So the problem, that that's ideally

05:39.095 --> 05:43.715
how we wanna do things in, in public and in the press.

05:44.015 --> 05:46.155
And in a lot of discourse, people kind

05:46.395 --> 05:48.635
of routinely make statements without reference

05:48.635 --> 05:49.875
to any counterfactual.

05:50.415 --> 05:53.795
So in the book, I kind of consider a lot of forms of

05:54.145 --> 05:56.595
what I would call a kind of faulty reasoning related

05:56.615 --> 05:58.765
to the value of a college degree.

05:58.945 --> 06:01.365
So for example, you get these articles like,

06:01.385 --> 06:03.045
too many students are going to college

06:03.265 --> 06:04.685
or are too many students going,

06:04.785 --> 06:07.165
and that college may not be the best option for,

06:07.825 --> 06:09.165
for all students.

06:09.715 --> 06:12.765
Usually the basis for concluding that a lot of people in,

06:13.025 --> 06:16.125
in a lot of these cases, let's say they, they refer

06:16.125 --> 06:19.205
to people who didn't score high on some threshold on a test

06:19.305 --> 06:22.325
or something that they shouldn't pursue a college degree, is

06:22.325 --> 06:25.205
how they compare to students with a high test score.

06:25.475 --> 06:28.445
They don't draw on counterfactual comparisons of

06:28.445 --> 06:30.325
what these students would be doing had they not gone

06:30.625 --> 06:34.405
to college and what the value is, you know, by,

06:34.625 --> 06:36.525
by making those types of comparisons.

06:37.065 --> 06:39.685
And they usually don't even look at the outcomes

06:39.855 --> 06:41.165
associated with the degree.

06:41.175 --> 06:43.725
Maybe it's the experience within college itself

06:43.865 --> 06:47.405
or, you know, a year out of college, so they're doing this

06:47.405 --> 06:50.525
or that, and not the long term life trajectories

06:50.625 --> 06:53.045
and across a range of different domains.

06:53.975 --> 06:55.235
Others have argued that

06:55.815 --> 06:58.955
low income students are mismatched on college campuses

06:59.915 --> 07:01.075
relative to high income students.

07:01.165 --> 07:04.635
Again, not drawing on counterfactuals affirmative action

07:05.035 --> 07:07.915
policies have, in some narratives, hurt, claim

07:07.915 --> 07:11.195
that they hurt students by placing them in situations

07:11.355 --> 07:12.395
where they can't compete.

07:12.455 --> 07:15.035
And, and all these kinds of debates, the,

07:15.135 --> 07:17.675
the same comparisons are habitually invoked that

07:17.675 --> 07:20.755
between the achievement of different college students

07:20.905 --> 07:22.075
with different backgrounds

07:22.295 --> 07:25.035
or races, it's not between, you know, these,

07:25.165 --> 07:27.555
these counterfactuals and what we'd see long term

07:27.615 --> 07:29.995
and how these, these lives unfold.

07:30.695 --> 07:33.155
Um, and then a lot of debates kind of point

07:33.155 --> 07:34.555
to the students are struggling

07:34.555 --> 07:36.475
to find stable jobs or they're in debt.

07:37.175 --> 07:42.075
Um, and even contend that it's, it's, it's obvious that,

07:42.135 --> 07:44.035
you know, colleges don't guarantee success,

07:44.035 --> 07:45.555
have seen quotes like that.

07:46.705 --> 07:49.245
Uh, and there's often this, this implication that students

07:49.245 --> 07:51.445
who can't afford to pay tuition

07:51.465 --> 07:54.525
and who are taking on debt shouldn't be attending college.

07:54.585 --> 07:56.725
And there's, again, there's no mention of kind

07:56.725 --> 07:59.485
of long-term benefits to college estimated

07:59.485 --> 08:01.245
by counterfactual trajectories.

08:01.245 --> 08:05.485
And so that's just a, a very common, you know, reasoning

08:05.485 --> 08:08.485
that's invoked in, in popular discourse that

08:08.635 --> 08:11.765
that doesn't really get at the value of degrees.

08:12.385 --> 08:15.245
Um, and the, the, the final one that kind

08:15.245 --> 08:17.645
of comes up a lot is that college degree returns are

08:17.645 --> 08:19.885
decreasing across generations

08:20.025 --> 08:22.925
or that, you know, students are filling jobs that

08:23.685 --> 08:25.445
previously didn't require a degree,

08:25.985 --> 08:28.445
or they're working at Starbucks after, you know, a year

08:28.445 --> 08:29.485
after school or something.

08:30.435 --> 08:33.445
Same kind of problem, not addressing counterfactual.

08:33.445 --> 08:35.005
And we know that returns to college are

08:35.005 --> 08:36.845
increasing, not decreasing.

08:36.905 --> 08:39.325
And, you know, many jobs have changed requirements

08:39.325 --> 08:41.645
because of increasingly complex tasks

08:41.815 --> 08:43.405
associated with these positions.

08:43.465 --> 08:46.945
So even if returns had been decreasing,

08:46.945 --> 08:50.105
it doesn't invalidate the return, to the degree.

08:50.445 --> 08:52.185
And because if we consider

08:52.215 --> 08:54.425
what people would be doing in the absence of a degree

08:55.065 --> 08:56.765
who don't hold a college degree, their,

08:56.935 --> 08:59.325
their prospects are increasingly limited.

09:00.265 --> 09:04.525
Um, so in each case, these types of comparisons kind

09:04.525 --> 09:07.245
of obscure the logic of causal effects and,

09:07.745 --> 09:10.125
and don't always describe, you know, the long term outcomes

09:10.265 --> 09:14.085
and look at the value of college with a counterfactual lens.

09:14.945 --> 09:17.565
Um, and so, so what are the effects of,

09:17.665 --> 09:19.645
of college across, students?

09:19.745 --> 09:22.205
So throughout the book, I ask not only

09:22.225 --> 09:24.485
how a four year degree increases wages

09:24.485 --> 09:25.965
and income over the career,

09:26.065 --> 09:30.165
but how it also reduces low wage careers, job instability,

09:31.025 --> 09:33.965
low skilled work poverty,

09:34.385 --> 09:36.125
how it reduces social assistance

09:36.305 --> 09:38.005
and increases civic engagement.

09:39.025 --> 09:43.125
And, and evaluating this dis differential response

09:43.185 --> 09:46.045
to four year completion, find that the benefits

09:46.045 --> 09:49.005
for unlikely college graduates are, are very large.

09:49.145 --> 09:52.405
So its college reaches a more diverse population.

09:52.425 --> 09:54.445
It kind of entails more diverse benefits.

09:55.145 --> 09:57.045
And so for disadvantaged use,

09:57.365 --> 10:01.005
I observe these increases in wages, household income, civic

10:01.025 --> 10:02.645
and political participation,

10:03.345 --> 10:05.485
but also find that completing college circumvent

10:05.645 --> 10:06.725
a lot of disadvantage.

10:07.165 --> 10:09.925
Unemployment, low wage work, job instability,

10:10.625 --> 10:12.485
family poverty, social assistance,

10:12.745 --> 10:15.765
and the patterns are very similar across the two cohorts.

10:15.825 --> 10:20.125
So we're seeing the same kind of, set of, set of results

10:20.225 --> 10:22.085
for the more recent cohort who were,

10:22.195 --> 10:23.645
were at the time in their thirties.

10:24.475 --> 10:26.805
It's not that college equalizes everything.

10:27.225 --> 10:31.705
Um, so I, I find that it equalizes a large range

10:31.705 --> 10:34.345
of outcomes, especially those disadvantaged

10:35.055 --> 10:36.505
ways in which it disa, you know,

10:36.535 --> 10:37.905
circumvents that disadvantage.

10:37.995 --> 10:41.065
Those look very equal across college graduates.

10:41.065 --> 10:43.105
But for things like who marries whom

10:43.125 --> 10:46.665
and household income, I see a lot of large benefits

10:46.785 --> 10:48.425
for the more advantaged students as well.

10:48.885 --> 10:50.305
And some of those are even larger.

10:50.725 --> 10:55.265
So college doesn't, you know, completely eradicate,

10:55.265 --> 10:57.625
social background as we wouldn't expect it to.

10:58.165 --> 11:01.105
But we see these, these really large returns for the

11:01.615 --> 11:04.985
more disadvantaged youth would always say everyone benefits.

11:05.425 --> 11:07.385
There's, there's kind of broad benefits

11:07.445 --> 11:09.945
and there's notably large benefits

11:10.205 --> 11:12.305
for advantage college graduates.

11:12.305 --> 11:15.505
Those are saying for those for kind of family formation,

11:15.505 --> 11:17.185
household income and wages.

11:17.285 --> 11:19.345
You know, we see a lot of benefit there,

11:19.365 --> 11:21.345
but for a range of these other outcomes,

11:22.175 --> 11:24.425
disadvantaged students benefit more.

11:24.425 --> 11:27.585
And there we see the negative selection really coming into

11:28.175 --> 11:30.785
play for all those things I mentioned.

11:31.085 --> 11:35.545
Um, now all the labor market insecurity, poverty,

11:35.645 --> 11:37.185
social assistance, and so forth.

11:37.565 --> 11:39.425
And so the kind of the sum of the findings

11:39.575 --> 11:41.625
that present throughout suggests

11:41.625 --> 11:45.345
that completing college has larger benefits for those,

11:45.555 --> 11:48.185
those outcomes for disadvantaged graduates.

11:48.185 --> 11:49.745
And a lot of this is very consistent

11:49.775 --> 11:52.545
with a large literature showing these,

11:52.755 --> 11:55.885
These large benefits for unlikely graduates.

11:56.525 --> 11:59.805
Universities have really long held an ideal

11:59.835 --> 12:02.365
that higher education should extend

12:02.385 --> 12:04.125
beyond traditionally served students.

12:04.235 --> 12:07.845
That the kind of task of demo democratic society is

12:07.845 --> 12:10.405
to preserve the ideal of liberal education

12:10.585 --> 12:13.165
and to extend it as far as possible to all,

12:13.165 --> 12:14.445
all members of the population.

12:15.025 --> 12:17.805
Yet we're at a place where college completion rates in the

12:17.825 --> 12:19.765
US are, are problematically low,

12:20.265 --> 12:21.725
we have more students attending,

12:21.825 --> 12:24.085
but the completion rates have,

12:24.115 --> 12:25.925
have stagnated to some degree.

12:26.665 --> 12:30.565
And so those who contend that college attendance

12:30.705 --> 12:33.925
and completion have limited benefits, kind of reinforce

12:33.995 --> 12:35.925
that it, that inequality and,

12:35.945 --> 12:39.005
and harm the lives of people who might have gone to college

12:39.145 --> 12:40.565
and experienced these benefits,

12:40.705 --> 12:42.805
but, might have been swayed otherwise.

12:43.625 --> 12:46.805
Um, and especially those disadvantaged youth and,

12:47.065 --> 12:51.245
and those who contend that there are limited benefits,

12:52.065 --> 12:55.165
those, those arguments tend to be flawed

12:55.165 --> 12:58.005
because a college degree is especially benefit, beneficial

12:58.145 --> 13:01.405
for, for these, students, unlikely to complete degrees.

13:01.945 --> 13:03.725
So I think investment in higher education

13:04.925 --> 13:07.845
benefits are collective wellbeing or democratic society.

13:08.195 --> 13:11.125
There's greater long-term earnings, more tax revenue,

13:11.875 --> 13:13.765
less reliance on public assistance,

13:14.305 --> 13:15.965
higher levels of volunteering.

13:16.825 --> 13:19.885
All that indicates that public investment in higher

13:19.915 --> 13:22.525
education for students from more disadvantaged backgrounds.

13:22.525 --> 13:26.445
These yields, these kind of far-reaching societal benefits.

13:26.985 --> 13:30.085
If we frame education in terms of individual rewards,

13:30.435 --> 13:31.805
it's more challenging to kind

13:31.805 --> 13:33.565
of garner widespread public support

13:34.265 --> 13:36.565
and expand to a more diverse population.

13:37.205 --> 13:40.405
I think we should demand a society based on this idea

13:40.405 --> 13:43.005
of a public good, a collective good,

13:43.065 --> 13:46.685
rather than one based on individualism and scarcity

13:46.785 --> 13:50.165
and acknowledge the widespread benefits collectively.

13:50.625 --> 13:52.685
If we invest in more students,

13:52.755 --> 13:54.045
obtaining four year degrees.

00:00 - My interests in educational inequality really go back to my undergraduate days, some 25 years ago when I was working in classrooms and studied a d tracking program that supported disadvantaged students to achieve in high school and attend college.

00:17 - And then, as I pursued graduate studies at the University of Wisconsin, I continued these interests in educational inequality and also developed interests in causal inference. And when I was writing my thesis in graduate school, I uncovered this interesting pattern about how students with a low likelihood to attend, in this case, selective colleges have these really large returns and even the largest returns to attending than those with a higher likelihood, the more advantaged students. So I kept pursuing that, that field of study

00:50 - as a postdoc at the University of Michigan, I started working with a colleague Y and we looked at the effects of college completion on earnings, and again, saw this pretty striking pattern, a pattern. We started calling negative selection that students unlikely to complete college reap these really large rewards. And then in my early years at UCLA, I explored other outcomes, family formation pattern, civic participation, and saw some similar, patterns

01:20 - of, for these really large returns. And I learned a great deal from my UCLA students. So students generally expressed many benefits of completing college, but it was my first generation students, students of color, students who really overcame the odds against 'em to attain these high levels of education, who invariably kind of told me that college changed the course of their life. And many of them also told me they wanted to go back

01:45 - to their communities and their youth. And that motivated some work on civic participation. So the book was this effort to kind of compile and advance, the work that I'd been doing on variation and effects of college for, over 20 years now. And, and I wanted to show the ways that college is worth it, particularly for those students on the margin who may be choosing whether or not to attend and who face, you know, a lot of hurdles along the way. So my goal was to assess how long-term benefits

02:16 - of four year degree completion varies across the population and to look at a large range of outcomes, socioeconomic, family level, social assistance, civic outcome measures. So I draw on nationally representative longitudinal survey data from the Bureau of Labor Statistics. This is the National Longitudinal Surveys Program. and I draw on two cohorts from the National Longitudinal Surveys and then look at the effects of college completion on these long-term life outcomes.

02:48 - and so I conduct statistical analysis to assess the differential effects of completing degrees and comparing those who completed a four year college degree to those who didn't, and, and, and look at these various outcomes. And I use a variety of statistical approaches, matching machine learning models to, to do this. and over the years I've worked with fantastic collaborators, both faculty and a lot of my graduate students to kind

03:18 - of inform these analyses and make the book what it was. First, as I was saying about the, the National Longitudinal Surveys program that's been ongoing for more than 50 years. So we have these great longitudinal data that looks at these two cohorts born in their early 1960s and early 1980s. And then I'm able to follow them because this has been ongoing for so long to follow these individuals over time

03:44 - and see how their lives unfold after they, either complete college or don't complete college. and then, so to assess the effects of completion, we wanna draw on these counterfactual comparisons of different outcomes. So the outcome when an individual completed college and the out kind of the, the outcome, when that same individual ideally did not complete college, that would be our, our best way to assess the effects.

04:11 - And in theory, we could observe, you know, the outcome had someone gone and not gone to college. But in reality, of course, only one state is observed. So we, we turned to these statistical approaches to estimate the unobserved counterfactual pathway. What would've happened had that person not, completed a college degree. So that's where I turn to these matching analyses and machine learning approaches to estimate effects. And then look how those effects vary across different

04:40 - subgroups of students and matching is pretty intuitive. We, we take someone who graduated from college and compare their life outcomes to someone who had the same estimated likelihood of graduating from college, but didn't actually complete a degree for whatever reasons, and, and then estimate the effects based on all these match pairs. So we take thousands of match pairs and look at the average outcomes, across 'em, then

05:06 - to make the, the book more, and even more intuitive and appealing. And, and with all these quantitative analyses, I constructed 12 matched vignettes to describe some of the narratives of both the college and non-college graduates. So this gave, you know, some names to these people from the National Longitudinal Survey data and how their lives looked similar before college and how they diverged after college.

05:32 - And you know, what it, what it kind of looked like. So the problem, that that's ideally how we wanna do things in, in public and in the press. And in a lot of discourse, people kind of routinely make statements without reference to any counterfactual. So in the book, I kind of consider a lot of forms of what I would call a kind of faulty reasoning related to the value of a college degree. So for example, you get these articles like,

06:01 - too many students are going to college or are too many students going, and that college may not be the best option for, for all students. Usually the basis for concluding that a lot of people in, in a lot of these cases, let's say they, they refer to people who didn't score high on some threshold on a test or something that they shouldn't pursue a college degree, is how they compare to students with a high test score. They don't draw on counterfactual comparisons of

06:28 - what these students would be doing had they not gone to college and what the value is, you know, by, by making those types of comparisons. And they usually don't even look at the outcomes associated with the degree. Maybe it's the experience within college itself or, you know, a year out of college, so they're doing this or that, and not the long term life trajectories and across a range of different domains. Others have argued that

06:55 - low income students are mismatched on college campuses relative to high income students. Again, not drawing on counterfactuals affirmative action policies have, in some narratives, hurt, claim that they hurt students by placing them in situations where they can't compete. And, and all these kinds of debates, the, the same comparisons are habitually invoked that between the achievement of different college students with different backgrounds

07:22 - or races, it's not between, you know, these, these counterfactuals and what we'd see long term and how these, these lives unfold. and then a lot of debates kind of point to the students are struggling to find stable jobs or they're in debt. and even contend that it's, it's, it's obvious that, you know, colleges don't guarantee success, have seen quotes like that. and there's often this, this implication that students

07:49 - who can't afford to pay tuition and who are taking on debt shouldn't be attending college. And there's, again, there's no mention of kind of long-term benefits to college estimated by counterfactual trajectories. And so that's just a, a very common, you know, reasoning that's invoked in, in popular discourse that that doesn't really get at the value of degrees. and the, the, the final one that kind of comes up a lot is that college degree returns are

08:17 - decreasing across generations or that, you know, students are filling jobs that previously didn't require a degree, or they're working at Starbucks after, you know, a year after school or something. Same kind of problem, not addressing counterfactual. And we know that returns to college are increasing, not decreasing. And, you know, many jobs have changed requirements because of increasingly complex tasks

08:41 - associated with these positions. So even if returns had been decreasing, it doesn't invalidate the return, to the degree. And because if we consider what people would be doing in the absence of a degree who don't hold a college degree, their, their prospects are increasingly limited. so in each case, these types of comparisons kind of obscure the logic of causal effects and, and don't always describe, you know, the long term outcomes

09:10 - and look at the value of college with a counterfactual lens. and so, so what are the effects of, of college across, students? So throughout the book, I ask not only how a four year degree increases wages and income over the career, but how it also reduces low wage careers, job instability, low skilled work poverty, how it reduces social assistance and increases civic engagement.

09:39 - And, and evaluating this dis differential response to four year completion, find that the benefits for unlikely college graduates are, are very large. So its college reaches a more diverse population. It kind of entails more diverse benefits. And so for disadvantaged use, I observe these increases in wages, household income, civic and political participation, but also find that completing college circumvent a lot of disadvantage.

10:07 - Unemployment, low wage work, job instability, family poverty, social assistance, and the patterns are very similar across the two cohorts. So we're seeing the same kind of, set of, set of results for the more recent cohort who were, were at the time in their thirties. It's not that college equalizes everything. so I, I find that it equalizes a large range of outcomes, especially those disadvantaged ways in which it disa, you know,

10:36 - circumvents that disadvantage. Those look very equal across college graduates. But for things like who marries whom and household income, I see a lot of large benefits for the more advantaged students as well. And some of those are even larger. So college doesn't, you know, completely eradicate, social background as we wouldn't expect it to. But we see these, these really large returns for the more disadvantaged youth would always say everyone benefits.

11:05 - There's, there's kind of broad benefits and there's notably large benefits for advantage college graduates. Those are saying for those for kind of family formation, household income and wages. You know, we see a lot of benefit there, but for a range of these other outcomes, disadvantaged students benefit more. And there we see the negative selection really coming into play for all those things I mentioned.

11:31 - now all the labor market insecurity, poverty, social assistance, and so forth. And so the kind of the sum of the findings that present throughout suggests that completing college has larger benefits for those, those outcomes for disadvantaged graduates. And a lot of this is very consistent with a large literature showing these, These large benefits for unlikely graduates. Universities have really long held an ideal

11:59 - that higher education should extend beyond traditionally served students. That the kind of task of demo democratic society is to preserve the ideal of liberal education and to extend it as far as possible to all, all members of the population. Yet we're at a place where college completion rates in the US are, are problematically low, we have more students attending, but the completion rates have,

12:24 - have stagnated to some degree. And so those who contend that college attendance and completion have limited benefits, kind of reinforce that it, that inequality and, and harm the lives of people who might have gone to college and experienced these benefits, but, might have been swayed otherwise. and especially those disadvantaged youth and, and those who contend that there are limited benefits, those, those arguments tend to be flawed

12:55 - because a college degree is especially benefit, beneficial for, for these, students, unlikely to complete degrees. So I think investment in higher education benefits are collective wellbeing or democratic society. There's greater long-term earnings, more tax revenue, less reliance on public assistance, higher levels of volunteering. All that indicates that public investment in higher education for students from more disadvantaged backgrounds. These yields, these kind of far-reaching societal benefits.

13:26 - If we frame education in terms of individual rewards, it's more challenging to kind of garner widespread public support and expand to a more diverse population. I think we should demand a society based on this idea of a public good, a collective good, rather than one based on individualism and scarcity and acknowledge the widespread benefits collectively. If we invest in more students, obtaining four year degrees.

Faculti AI can make mistakes