The Huge World Of Arrests: Why The War On Drugs Matters In Mass Incarceration, Part 4.

By Xenocrypt

A simplified narrative holds that “the war on drugs” is responsible for all of “mass incarceration”. Recently there’s been some push-back against this narrative, arguing that the influence of the war on drugs on mass incarceration is exaggerated. In this series, I’ve been trying to push back against that, and emphasize the ways in which drug offenses really did drive “mass incarceration” along several metrics.

In the first part of the series, I wrote about new court commitments — basically, the set of people who have gone to prison:

In the second part of my series, I wrote about the distinction between prison admissions and the prison population — which statistic you choose to look at can give you seemingly very different conclusions:

In the third part of this series, I wrote about felonies and the set of people who have been convicted of a felony — a much larger group than the set of people who have gone to prison:

But just as only a fraction of people with a felony conviction have gone to prison, the set of people of people with a felony conviction is, in turn, only a small fraction of another much larger universe: the set of people who have been arrested.

In this part of the series, I’ll talk about just how huge the world of arrests is in absolute terms, and (tentatively) discuss the roles drug offenses might have played in the growth and level of arrests. I will also even more tentatively discuss historical and international comparisons, although there is even less and even worse data here than either for felonies or prison sentences.

The Huge World Of Arrests:

What gets the most attention in “mass incarceration” is mass incarceration — the people in prison right now, at any given time. There are mostly people serving sentences of over one year, usually more, in state or federal prison:

In previous articles in this series, I noted that the large “stock” population of violent prisoners obscures how most of the “flow” through prisons consists of non-violent offenders, often serving shorter sentences.

Similarly, the large “stock” population of prisoners, of all those people in state and federal prison serving felony of sentences of over a year, obscures how much of the “flow” through the criminal justice system consists of simple arrests. Arrests, that is, people being arrested, maybe spending a day in a county jail, maybe less or more, maybe being convicted of a misdemeanor, maybe not, with only a small fraction going on to a felony conviction, let alone a state or federal prison.

In sheer human magnitude, arrests are easily the biggest thing the criminal justice system does (perhaps unless you define traffic enforcement as “criminal”). We’ve already seen that state felony sentences cover a much larger population that the “prison system” as such. For example, in 2006, the last year with all three numbers, there were 492,315 sentences to new prison terms compared to 1,132,290 state felony sentences…and 14,380,370 arrests, according to the FBI’s Uniform Crime Report (UCR) projections. Even 2017 saw 10,545,985 arrests. (And with only 610,000 people in local jails at any one time, the average time behind bars from an arrest is about three weeks, with the median presumably much less.)

Again, the vast majority of arrests do not lead to a prison sentence directly or even to a felony conviction. Relatedly, most people who go through the criminal justice system do not get trapped cycling in and out through more and more serious levels, inevitably going from misdemeanors to felonies to short prison terms to long prison terms.

The vast majority of people who enter the criminal justice system stay at the level of arrests, misdemeanors, and short jail terms. The effect the system has on them might be far less severe than it has on the people who end up with felony convictions and prison terms, but there are a lot more of them.

What Is The History Of Arrest Rates?

What is the historical context of 11-14 million Americans arrested a year? This is little-understood even within the “criminal justice discourse”. Just how little-understood can be seen in the following:

Last year, Roger Lancaster published the following in “Jacobin”:

This is unfortunately wrong in several ways which went unnoticed, presumably owing to the popularity of the “hyper-policing” narrative. I believe the 1960 numbers are actually an unweighted count from a large but incomplete sample of police agencies covering 109 million people, or about 60% of the United States at the time:

It was only in 1970 that the UCR numbers in these reports were weighted to include a projection of national arrest counts. UCR reports from 1960 to 1979 can be seen on the Internet Archive, while UCR reports from 1980–2014 can be seen on a BJS data tool, with more recent reports accessible individually on the BJS website.

(Note: After some consideration, I’m going to make use of all of these reports, including the unweighted samples, but I’ll note when the rates come from an unweighted sample. In my opinion there’s reason to think that weighting would not change the rates much, but I am not sure and not a real expert, so use caution.)

The Rise And Fall Of Arrests Over “Mass Incarceration”:

What do we see when we combine those sources into a (more or less) apples-to-apples comparison of arrest rates of covered populations? That it’s not just the absolute number of arrests that’s been much larger than the absolute number of prison terms — but also the changes in arrests, both the rises and falls.

The growth in prison sentences after 1970 or 1980 was considerable, with immense human consequences. You’ve probably seen a graphic similar to this one in nearly every article on “mass incarceration”, including mine:

But even that dramatic explosion of prison sentences is, quite literally, dwarfed, when put on the same scale as the rise and fall of arrest rates.

The arrest rate rose from the early 1960s to 1990 or so, peaking at about 50% more than the apparent 1960 rate, before declining again. (The drop in arrest rates even since 2009 is pretty significant, which is another reason that “Jacobin” quote is misleading.) In terms of growth rate, this is not as obviously striking as the growth in the prison sentencing rate or the incarceration rate, which increased by a factor of 2 or 3 or more.

But the sheer numbers of “extra arrests” are striking in absolute terms. There have been perhaps as many as 130,000,000 — one hundred and thirty million — “extra arrests” relative to the 1960 baseline, far outstripping the 7 million or so “extra prison sentences” over the mass incarceration period. In other words, most of the increased punitiveness of the “mass incarceration” period was in arrests, not in prison sentences. However that seems to have changed more recently,

Side note: I don’t want to graph “crimes” and “arrests” in the same figure because a great many arrests, as I’ll shortly illustrate, are not for what you would necessarily count under “crimes”. But many graphs of crime rates show a not dissimilar rise-and-fall pattern as the UCR arrest rate graph:

Although many an arguably deceptive graphic has tried to imply that the US criminal justice system has basically ignored the fall in crime rates:

I think this is basically cherry-picking — from UCR numbers, at least, it would seem that arrests rose and fell more or less along with serious crime. And arrests are most of what the criminal justice system does.

The (Ambiguous) Role Of Drugs In The Rise And Fall Of Arrests:

So arrests rose over the first part of the “mass incarceration period” along with prison sentences, if not as sharply. But unlike prison sentences, arrests then declined, apparently back to around the original baseline.

What explains this inconsistency? The UCR arrest data does have offense categories which have changed little over the past fifty-seven years, although as we’ll see this is both a blessing and a curse.

Here’s a stacked area graph of arrest rates by offense from the UCR data:

As you can see, while the aggregate of arrests rose and then fell more or less back to the original baseline, the arrest rates for the more serious offenses — violent, property, and drug offenses — remain elevated, in part because of drug arrests in particular.

Can we quantify exactly how much drug arrests, in particular, contributed to the growth of arrests over the “mass incarceration” period? Unfortunately, I think this is a difficult question. With prison sentences and felony convictions, there was growth across all available categories, so it was straightforward to at least say what percentage of the growth was from drug offenses. With arrests, it’s a little different.

Here’s the same information as the above stacked area chart, but broken down by offense:

(Hopefully you can see why I’m tentatively comfortable mixing up the 1960–1969 unweighted samples with the 1970–2017 national projections — they line up pretty closely in both trend and in adjacent years, offense by offense.)

Arrests in many categories — most notably “drunkenness” but also others — declined either over basically the whole period, or at least from the circa-1990 peak of total arrests, perhaps as you would expect and hope in a period of declining crime rates.

But drugs are different. It is pretty immediately apparent from that graph that drug offenses have dropped less from their peak, and show less sign of dropping at all, than pretty much any of the other offenses do.

This certainly suggests that a large share of “extra arrests” are or have been arrests for drug offenses, or indeed that drug arrests are doing a lot to prop up the criminal justice system in a low-crime era.

But it also leaves a lot ambiguous. Do the same kinds of circumstances that once led to an arrest for “drunkenness” now lead to an arrest for “drugs” — carrying greater risk of serious sanction, but not changing the total number of arrests much? If drug arrests were forced to drop, would the same people be arrested just as often but for “disorderly conduct”?

Maybe not— maybe drug arrests really are purely “extra punitiveness” layered on top of the system. It’s hard to answer those questions from this kind of information.

One last graph here to illustrate the scale of things. Here are the ~132 million “extra arrests” — the arrests that happened beyond the 1960 baseline rate — of 1960–2017, broken down by offense.

The arrest category with the largest increase is “other offenses”, which might not mean much more than that the categories are out of date (as useful as keeping them the same is for longitudinal purposes!), but “Drug” and “DUI” show notable increases too.

It’s probably best to just say that the overall arrest rate did rise and fall, which means there were a lot of “extra arrests” over this period, with arrests in the “most serious categories” remaining elevated, drugs included. Drug arrests were after all barely a thing in the 1960 survey — under 1% of total arrests.

While details remain unclear, there’s definitely a lot more drug arrests than there used to be, with a lot of human consequences, whether that’s driving a general increase in arrests or whether it’s effectively absorbed from another category.

The Huge World Of People With Arrest Records:

In my posts on prison sentences and felony convictions, I discussed estimates of the total number of people with each on their record, and how that’s changed over time. I want to discuss similar estimates for arrest records. However, as with everything else to do with arrests, the numbers are both much larger and much less clear. There are limitations and nuances depending on the source of the data and the exact question asked. Administrative records or self-reported survey answers? The current percentage of Americans who have been arrested or a projection of how many Americans will be arrested based on a particular set of cohorts? Since most arrests are of relatively young people, and since the arrest rate has changed significantly over time, it all gets very complicated.

What is clear is that a large percentage of Americans have been arrested at least once, and this seems to have been the case for a long time. Famously, in 1967 Robert Christensen wrote that 22% of Americans would be arrested, projecting from a contemporary cohort. (Unfortunately I don’t think this study is actually online anywhere.) A 1986 book chapter goes over a number of other studies and compares the results — work I can’t really evaluate — but they all seem to be around that ballpark if not higher.

More directly, the General Social Survey’s “Arrest” question asks: “ Were you ever picked up, or charged, by the police, for any (other) reason whether or not you were guilty?”. About 11% of respondents said they had in 1973, and about 20% of respondents said they had in 2012.

This is self-reported, however, and might well be a significant under-report. (Is the extent to which it’s an under-report constant over time? I don’t know.) In any given year from 1980 to 2014, you can look at the arrest rates for individual age groups on the BJS, in what’s called an “age-arrest curve”. In 2012, the number of arrests of 19-year-olds was around 12% of the population:

And the number of arrests of this same cohort in 2011, 2012, and 2013 is around one-third of the population. Of course one person might have multiple arrests even in a single year, and certainly in multiple years, but it gives you some sense of the scale of arrests.

A study published in “Pediatrics” based on the National Longitudinal Youth Survey estimated that 30% of Americans have been arrested by age 23, with cumulative arrest prevalences by age tracking above Christensen’s estimates:

Note that the United States is not necessarily unique in having a large share of its population go through the criminal justice system at some point. The definition or even the very practice of “arrests” seems to vary a lot from place to place, but for example, apparently, according to administrative records, “one of three Norwegian boys has been caught for — and charged with — committing a crime”. Compare to perhaps 40% of American males reporting an arrest by age 23. That might be an under-report — America is probably more punitive than Norway even by this metric — but the difference might be of degree more than kind.

Anyway, the data isn’t great. But it seems clear enough that, for a long time, a large percentage of Americans have been arrested, probably an even larger percentage after the “mass incarceration” period. And it also seems like a lot of the “extra arrests” were for drug offenses. So I don’t think it’s a reach to speculate that the “war on drugs” sent a lot of new people through the criminal justice system.


The scale of arrests is enormous. Perhaps 2% of adults have been to prison, 8% of adults have a felony record…and 20%-30% of adults have an arrest record. Again, that means that the vast majority of people who move through the criminal justice system never see a felony sentence, let alone a prison sentence (although presumably nearly all of them spend some time behind bars in local jails, even if for only a day or two).

Drug arrests do not make up a large percentage of that huge flow on their own. But they may well have made up a large percentage of “new arrests” over “the mass incarceration period”, and they may well be playing a significant role in propping up arrest rates in a period of lowered crime rates, similar to what we’ve seen with prison sentences and felony convictions. The war on drugs shouldn’t be discounted.