After Ferguson, a noticeable gap in criminal-justice statistics emerged: the use of lethal force by the police. The federal government compiles a wealth of data on homicides, burglaries, and arson, but no official, reliable tabulation of civilian deaths by law enforcement exists. A partial database kept by the FBI is widely considered to be misleading and inaccurate. (The Washington Post has just released a more expansive total of nearly 400 police killings this year.) “It’s ridiculous that I can’t tell you how many people were shot by the police last week, last month, last year,” FBI Director James Comey told reporters in April.

This raises an obvious question: If the FBI can’t tell how many people were killed by law enforcement last year, what other kinds of criminal-justice data are missing? Statistics are more than just numbers: They focus the attention of politicians, drive the allocation of resources, and define the public debate. Public officials—from city councilors to police commanders to district attorneys—are often evaluated based on how these numbers change during their terms in office. But existing statistical measures only capture part of the overall picture, and the problems that go unmeasured are often also unaddressed. What changes could the data that isn’t currently collected produce if it were gathered?

In one sense, searching for these statistical gaps is like fishing blindfolded—how can someone know what they don’t know? But some absences are more obvious than others. Bruce Western, a professor of sociology at Harvard University, cited two major gaps. One is the racial demography of arrests and criminal records. An estimated 65 million Americans, or roughly 25 percent of the U.S. population, have a criminal record of some kind. But the racial makeup of those records isn’t fully known. “There are estimates, but with [65 million] people in the FBI criminal record database, we have no systematic knowledge of their demographics,” Western told me.

There may be many missing statistics from the realm of policing, but even greater gaps lie elsewhere. Thanks to the FBI’s Uniform Crime Reports, police departments might actually be one of the better quantified parts of the criminal-justice system. Prisons also provide a wealth of statistics, which researchers have used to help frame mass incarceration in its historical and demographic content. The Justice Department’s Bureau of Justice Statistics maintains an annual report on the size of the U.S. prison population. The report includes state-by-state demographic statistics like inmate ages, races, crimes committed, and other crucial data for researchers and policymakers.

But while current prison statistics give a good sense of the size and scale of mass incarceration, they provide little information on conditions inside the vast constellation of American prisons. Perhaps the most glaring gap is solitary confinement. No one knows exactly how many people are currently kept in isolation in American prisons. “There are estimates, but no official count nationwide,” Western said.

The most commonly cited figure comes from the Vera Institute for Justice, which estimates that around 80,000 people are in solitary confinement at any given time in the United States. But precise statistics remain elusive despite the substantial policy implications. There is abundant evidence that prolonged isolation can be mentally and psychologically traumatic. Justice Anthony Kennedy told Congress in April that solitary confinement “literally drives men mad.”

For data to be relevant, it must be gathered broadly and repeatedly. 80,000 people is a significant number, but even the total is accurate, it’s analytically useless without context. For example, does this figure represent a decline or rise in solitary confinement compared to five years ago? How does it vary from state to state? What are the racial, gender, and age breakdowns of those 80,000 people? For how long are they kept in isolation, and why?

Another major gap in prison statistics is the number of non-sexual assaults behind bars. Although Congress mandated the collection of sexual assault statistics with the Prison Rape Elimination Act in 2002, prisons are not required to report ordinary assaults to the Bureau of Justice Statistics. “We have anecdotal evidence and some data in individual states that violence is rising in prisons, but we don’t have good, national database collection on that,” said Marie Gottschalk, a professor of political science at the University of Pennsylvania.

Statistical shortcomings like these have long vexed researchers studying the American criminal-justice system. Gottschalk and Western both contributed to a landmark report on mass incarceration published by the National Academy of Sciences last year. Among its findings, the report listed a number of fundamental gaps in official statistics about American prison conditions:

No comprehensive national data are routinely collected on even the most basic dimensions of the nature and quality of the prison experience, such as housing configurations and cell sizes; the numbers of prisoners who are housed in segregation confinement and their lengths of stay and degree of isolation; the amount of out-of-cell time and the nature and amount of property that prisoners are permitted; the availability of and prisoners’ levels of participation in educational, vocational, and other forms of programming, counseling, and treatment; the nature and extent of prison labor and rates of pay that prisoners are afforded; and the nature and amount of social and legal visitation that prisoners are permitted.

Gottschalk warned that the absence or underreporting of key statistics could warp public perceptions of inmates. “[If] we’re just collecting inmate-on-inmate data, or just rapes and sexual assaults between inmates, it kind of feeds into those stereotypes, and it ignores the huge amount of violence that’s actually done by staff inside of facilities,” she explained. Violent incidents behind bars, whether sexual or not, are also not regularly included in the FBI’s Uniform Crime Reports. Federal crime data shows an overall decline in violence in the last two decades. Would that trend hold if crimes against roughly 2 million incarcerated Americans were included?

Prisons and police departments may be the most visible parts of the criminal-justice system, but they are not necessarily the most powerful. As judges lost flexibility with the growth of mandatory-minimum sentences during the tough-on-crime era, prosecutors became the most pivotal actors within the criminal-justice process. This rise in influence was matched with a decline in transparency. “Up until the late 1980s and early 1990s, we used to collect more data—not a whole lot of data, but more data—on what prosecutors do,” Gottschalk explained. “[Now] the police are certainly much more accountable than prosecutors are, in terms of the visibility of what they do and the transparency of what they do.”

One prosecutorial tool with little transparency is plea dealing. In 2013, more than 97 percent of all federal criminal charges that weren’t dismissed or dropped were resolved through plea deals. (State-by-state totals are incomplete or unavailable, but often estimated to be similarly high.) For prosecutors, the benefits are clear: Offenders are punished without expending manpower and resources in lengthy trials. But plea deals are also one of the least-scrutinized parts of the criminal-justice system. “In most cases, that’s a complete black box,” Gottschalk said. “It allows prosecutors to have this enormous power without much transparency to the public.”

In the absence of reliable statistics, anecdotal evidence often fills the void. This is especially true when studying racial bias in the prosecutorial process. Prosecutors are not required by law to compile data on racial disparities. They also have little incentive to gather and publish it voluntarily, partly because of resource constraints and partly because of its potential negative implications.

A small number of offices have sought answers nonetheless, often with troubling results. In 2011, Manhattan District Attorney Cyrus Vance invited the Vera Institute to examine his office’s internal records for evidence of racial disparities. The institute’s final report found that disparities plagued every step of the process. Among its findings: Black and Hispanic defendants were more likely to be offered plea deals on misdemeanors that included imprisonment than white and Asian defendants. Would similar reviews of prosecutors’ offices nationwide produce similar results?

Without reliable official statistics, scholars often must gather and compile necessary data themselves. “A few years ago, I was struck at how many police killings of civilians we seemed to be having in Philadelphia,” Gottschalk said as an example. “They would be buried in the newspaper, and I was stunned by how difficult it was to compile that information and compare it to New York and do it on a per-capita basis. It wasn’t readily available.” As a result, criminal-justice researchers often spend more time gathering data than analyzing it.

This data’s absence shapes the public debate over mass incarceration in the same way that silence between notes of music gives rhythm to a song. Imagine debating the economy without knowing the unemployment rate, or climate change without knowing the sea level, or healthcare reform without knowing the number of uninsured Americans. Legislators and policymakers heavily rely on statistics when crafting public policy. Criminal-justice statistics can also influence judicial rulings, including those by the Supreme Court, with implications for the entire legal system.

Beyond their academic and policymaking value, there’s also a certain power to statistics. They have the irreplaceable ability to both clarify social issues and structure the public’s understanding of them. A wealth of data has allowed sociologists, criminologists, and political scientists to diagnose serious problems with the American criminal-justice system over the past twenty years. Now that a growing bipartisan consensus recognizes the problem exists, gathering the right facts and figures could help point the way towards solutions.