Malcolm Gladwell made his bones arguing for the 10,000 rule in his mega best seller Outliers. Gladwell posited that virtuosos across fields shared a common attribute that was at least correlated to, if not causal in their success—they practiced for 10,000 hours or more to master their craft. This was welcome news to parents everywhere. As a dad, I can tell you that every father loves three things above all else: the 16th hole at the PGA Phoenix Waste Management Open, taking stuff to the dump (and especially those two magical weekends in the spring and fall when the dump is free), and yelling at their kids to practice.
Outliers was good news. It meant that something all parents wanted to be true, hoped was true, prayed was true, actually was true. Practice makes perfect! Gladwell freed middle-aged people everywhere to helicopter parent as aggressively as they chose. Why? Because, science!
Now, as with any success story, there was a bit of a backlash. Gladwell’s argument, boiled down, was that quantity has a quality all its own1. Critics note that while this might be true, it is true only for skills that are specific and precise. If you want to be a pro golfer and one day publicly degrade yourself at the Waste Management Open, or if you want to be a virtuoso violinist playing Carnegie Hall, then the 10,000-hour rule probably applies. But if you want to be a great writer, not so much. 10,000 hours banging away at a keyboard will make you a virtuoso typist, but typing endlessly won’t magically transform you into Hemingway. This has implications for justice systems, by the way, because the analog of just hiring more cops or just doing more of anything isn’t enough to create change without purpose.
This brings up the second critique, which is that it matters what you do when you practice. Experts in practicing tend to argue that practice only matters if it is focused on the bits you do poorly. Or to put it into a Mike Tyson aphorism, you should only count the sit-ups that hurt, because they are the only ones that count.
So what does this have to do with violence and victimization? Quite a bit.
The Purpose Paradox
Researchers tend to care a lot about causality, but less about causal processes. This is what leads to findings like 10,000 hours of practice are necessary to be a virtuoso, without defining what it means to practice.
Researchers tend to care a lot about finding a mechanism that satisfies post hoc, ergo propter hoc—after this, therefore because of this. After 10,000 hours of practice, and because of 10,000 hours of practice, you are a virtuoso. ‘Because’ does a lot of heavy lifting in that sentence though and researchers fight endlessly about whether something happened ‘because’ of something else. But I wonder if we shouldn’t argue just as much about this as we do about because. What is in this that matters? What is the process by which some intervention causes a change? What is it about practice that matters?
One way to think about this, at least when dealing with people and their behavior, is to think about what the purpose of the intervention is. Why do we suppose that doing this thing will cause that thing to change? What is the purpose of practice?
Thinking about this is not an academic exercise. In fact, thinking about this question turns an academic exercise into a practical exercise. More specifically, the most practical exercise is asking the question, what is the purpose of the intervention? Putting correlations in the data aside, what makes the causal relationship plausible?
Routine Activities of At-Risk Youth
Suppose we are tasked with looking for ways to reduce violence and victimization at the national level. One place to start is to look at a lot of historical trend data and look for patterns that match the ups and downs of violence. There are a lot of examples of trends that match. I don’t mean to pick on the theory that legalization of abortion in the 1970s led to the great crime decline of the 1990s but I do wonder if this hypothesis didn’t come about simply because the historical trends match so well. I’ve never heard a sufficient explanation of why legalized abortion would lead to less crime two decades later.2
There is a better way to go about this, one that focuses on the process and purpose of the hypothesized drivers of crime and violence.
So, a better place to start is with young people, who are disproportionately more likely to be both a victimizer and a victim. It seems natural to me to start by thinking about what daily life is like for young people who live in places with a lot of violence and how could that daily life be improved in ways that reduce their risks.
Thinking these thoughts naturally leads you to think a lot about schools, since that is a place where young people spend a lot of time or could spend a lot of time. A lot is written about secondary schools and how they can be improved, but there is much less attention on the post-high school experience. There is also a lot of focus on training young people for employment in lieu of college. But for those who do not continue past high school, I worry that too much of this approach is about getting young people jobs but not about launching careers. For simplicity, I am thinking of a job as work without a lot of intrinsic purpose, whereas purpose is built into a career.
Having more young people on a career trajectory strikes me as one plausible causal mechanism for less violence and victimization. So how do we launch more careers?
One path to a career is college, but college is out of reach for many of the young people who live in places with a lot of violence. A bridge to college and a career does exist in the form of 2-year public colleges, what we colloquially call community colleges. So, let’s investigate what is happening—and has been happening—with community colleges and how that correlates with violence and victimization.
The Case for Direct Investment in Young People
Public, two-year colleges tend to be free or low-cost, tend to be widely available (including to people who live in dangerous neighborhoods), and tend to focus on building either the cognitive skills that students do not acquire from lower-performing high schools, or providing career-specific instruction. Commendably, the Brookings Institution has focused a lot on community college. One Brookings report points out:
[C]community colleges serve a large share of students who are low-income or first-generation college students, as well as those who are older or from underrepresented minority groups. Given their low tuition and widespread accessibility, community colleges are key to expanding higher educational attainment and boosting skills in this country, which can provide an important avenue for upward social mobility.”
The Brookings report also answers the critical question, ‘Does community college improve student outcomes?’
People who earn an associate degree are rewarded in the labor market. One recent study indicates that an associate degree yields a causal earnings premium of about 30% over a high school degree. Those who begin at a community college and successfully transfer to a four-year degree-granting institution also earn higher wages and greater economic security, given the earnings premium associated with a bachelor’s degree. However, degree completion rates among degree-seeking students at community colleges tend to be quite low.”
The Rise and Decline in Community College Enrollment
Let’s start with the big-picture trends in community college enrollment. First, let’s look at the really big picture, which is how many high school graduates there are. This is not a perfect measure, since you can go to community college any time after high school, but it’s instructive to see the general flow into the college-eligible pool. Notably, there are two elements to high school graduates—(1) the number of teenagers and (2) the graduation rate. The teenage population appears to have peaked around 2020 and is declining slowly. At the same time, high school graduation rates are creeping upward. The net effect, according to this graph from Cognia, is a slightly increasing pool of graduates through 2034 and then a long slow decline.
Let’s turn next to community college enrollment. You would expect that enrollment generally follows the high school graduation population, but it does not, and this turns out to be really important in understanding how community colleges can become a more effective tool in the fight for purpose for young people and against violence and victimization.
Note: Data from 2001-2022 are from the National Center for Educational Statistics, from two series: https://nces.ed.gov/ipeds/TrendGenerator/app/trend-table/2/3?trending=row&valueCode=4&rid=1&cid=14 and https://nces.ed.gov/programs/digest/d13/tables/dt13_303.70.asp. Data were available for 1970, 1975 and 1980 and the intervening years were imputed using a linear moving average.
The trend in community college enrollment needs little explanation. Enrollment increased steadily until 2008 when enrollment accelerated for a few years. Enrollment peaks in 2010-2011 and then begins a rapid decline which continues until 2020 when there is a second even more rapid decline.
Another way to show the data is to show the percentage change in one year of enrollment. The figure below shows several periods of rapid growth with a notable spike in 2008-2009 and a big decline in 2020. Finally, the trend flips in 2023 with a substantial increase.
The hypothesis here is that there is a relationship between community college enrollment and changes in violence and victimization. While a causal relationship is hypothesized, the figure below is not meant as a rigorous test of association. Rather, the idea here is just to put forth the hypothesis. One way to explore that hypothesis is to observe whether changes in community college enrollment coincides with changes in violence and victimization. Here, I present the year-to-year changes in both series for the last fifteen years. The idea is that as the green series (community college enrollment) increases, the orange series (violence rates) decline. Because the relationship is hypothesized to be symmetrical, the hypothesis also anticipates that declines in enrollment will coincide with increases in violence and victimization. The figure below generally supports that hypothesis.
End Note
This is a pretty blunt analysis and I do not presume it to be a substitute for more serious inquiry. The historical trends correlate, and the mechanism—the purpose—of community college is clear.
The point is pretty blunt though—if you think about what opportunities young people who live in places with histories of violence have available to them, and which opportunities could be strengthened to improve their opportunities and thus reduce their risk of violence and victimization, there is a compelling case for community college to be a priority. And once community college is a priority, more investment—sorely needed investments in institutional capacity and capabilities—should follow.
One last point—while there does not seem to be a strong correlation between high school graduates and community college enrollment, there will soon be a declining number of high school graduates. That means that the problem of fewer students enrolling in community college is not only not going to magically resolve itself, but it may also get worse. More needs to happen to create an environment where community college is a viable option, a third choice instead of a false dichotomy of work or four-year college.
A Model for Presenting Data to the Public
Since I have been thinking a lot about trends for this essay, I should note that I have had the opportunity to dive into FRED—the online data portal maintained by the Federal Reserve of St. Louis. It’s a remarkable website, simple to use, comprehensive, and incredibly useful. On the other hand, the Crime Data Explorer maintained by the FBI is none of those things. I gave some thought to what principles underlie the FRED data and how they compare to the CDE. What emerged is very straightforward—FRED imposes remarkably few limitations on the user and that is the key to its success. The FBI Crime Data Explorer (CDE)—and most other social science data websites—impose their values on users and the experience suffers. Here are the principles:
1. Trends, not short-term change. A clear, simple focus on time series data to generate and compare trends. The FBI focuses on recent data, as do most government websites which default to comparisons of this year compared to last year, or this quarter compared to last quarter. Very few things in the world can be well understood in short time frames.
2. User-designed structure. A clear focus on providing tools to let the user manipulate the comparisons with no structure imposed by FRED and no preference for how the analysis is conducted. The CDE gives the user little control—FRED gives the user all the control.
3. Comprehensive. Complete access to the entire series. The CDE truncates the data series.
4. Timely. FRED does no additional data cleaning—FRED reports whatever the data provider reports. The FBI spends months cleaning law enforcement data. It is not obvious to me what value is added by this, the onus should be on reporting agencies to get the data right at the time of submission—both to the FBI and to the public.
Now, you might ask—didn’t you just attack trend data in the community college essay? Only partially. Valid and reliable time series data is absolutely critical to understanding the causes and consequences of policy change. Trend data alone are a necessary, but insufficient tool. I encourage people to look beyond correlations in trends, but to look beyond the correlations, you first have to look at the trends!
Some Exburances
There are so many interesting things bubbling up these days, that I can’t keep up with it all. And since Twitter suppresses science as an official policy, there is nowhere to have a good discussion. Anyway, here are some brief thoughts and links. Let me know what you think!
Philadelphia’s New Mayor Likes Data
I am a really big fan of Charelle Parker, the new Philadelphia mayor. Look, you are never going to agree on everything with anyone, not even yourself if you aspire to perspicuous thought, but Mayor Parker is smart and leans hard into data and evidence. In a moment where good government everywhere takes a back seat to ideology, it is nice to see a mayor lean into good government! It is budget season in Philadelphia and as is often noted: if you want to know someone’s values, look at their budget. And Mayor Parker’s budget leans hard into getting control of Philadelphia’s generations-long violence problem. It’s the subject for an essay on another day, but the bottom-line is that prevention is the ultimate crime reduction strategy but policing matters, especially when violence and victimization are high. For those skeptical of this approach, I would encourage you to revisit this article on what we know about policing that German Lopez penned while he was at Vox.
Smart Takes from Noah Smith
If you read this Substack regularly, you know I am an admirer of Noah Smith. He is the best policy writer out there right now, merging theory and data to advance effective policymaking ideas, with a notable absence of ideology. This take on the difference between competitive advantage and comparative advantage should get a gold medal. Smith argues that even if generative AI advances beyond human capabilities, it will never replace people and we will still have jobs. The idea is simple: resources are constrained and even if AI is better than people at Thing One and Thing Two, it is probably going to use most of its energy on Thing One leaving Thing Two for the people. The opportunity for AI to advance beyond people may be limitless, but the opportunity costs of using AI are not. Even a super-genius AI has scarce resources.
Making the Financial Markets More Interesting!
Since I seem to be in a gushing mood, let’s gush some more! I am late to the game on this, but Matt Levine is a columnist for Bloomberg and writes a twice-a-week blog on the markets and investing. Even if your entire personal wealth is tied up in your house and your 401K (like almost everybody else), you should read Matt because he will teach you about the markets in a way you won’t find anywhere else. Remember, your brain wants to learn, all the time, at every age, up until the very end. Give your brain what it wants—give it Matt Levine.
Musical Interlude
I drove more than 1,000 miles with two teenagers last week. We listened to a lot of Rainbow Kitten Surprise so now you have to as well! OK, you don’t have to, but here’s a taste (fair warning to Gen Xers, Boomers and Tipper Gore: folks curse pretty freely on records nowadays).
It turns out that this phrase, quantity has a quality all its own, did not originate with Joseph Stalin but probably started instead in the US defense industry in the 1970s. There is also this sort of hilarious nugget from the ancient Greek philosopher Eubulides known as ‘the paradox of the heap’ which argues that a quantitative change in the number of grains of sand leads to a qualitative change in something being a heap or not. h/t Klangable.
The argument for the legalized abortion-crime link usually goes like this: when abortion was legalized (something) (something) fewer unwanted babies (something) (something) less crime when those kids would have turned 20! The implications of the omitted words are disturbing. But even putting that aside, the reality was that abortion was not widely accessible nor cheap in the 1970s, suggesting that abortions were disproportionately available to those who were relatively well-off financially. Which tends to deflate the argument.
I don't know if you've changed up your Substack or I'm just noticing it now, but you're really bringing a lot of value here. Thank you.
Apropos of nothing, I graduated in sociology from UC-Berkeley and one of our legendary faculty, Neil Smelser, had argued that the UC schools had no business providing lower division education and that all students should go to community colleges for two years then transfer to finish at a UC campus. That idea obviously never caught on, but it used to be that the UCs took a bunch of community college transfers. I hope it's still true.
Great piece, John. I have two questions. Do high school attendance trends parallel the trends for community college attendance? Is funding for community colleges declining?