What Are You Counting? SOC 3333 · Crime Lab · Midwestern State University

Part 2 asks a question that sounds simple and is not. How much of your offense happens, and how does anyone know? Every number you will ever cite about crime was produced by an instrument, and every instrument was built to see some things and not others. Before you can report a rate, you have to be able to say what got counted, who did the counting, and what the counting missed. That is the whole skill of this part, and it is the one that keeps paying off through Parts 3, 4, and 5. Nothing here is recorded and nothing is graded.

Your offense has at least two definitions, and they do not match

Part 2 asks you to define your offense legally and conceptually. Those are different jobs, and students who treat them as one job lose points in both. Here is what each one is for.

The legal definition is what a statute says. It has elements that a prosecutor must prove, it varies between states, and it changes when a legislature changes it. It exists to decide whether a specific person can be convicted. It was never designed to be a research category.
The conceptual definition is what researchers mean when they study the thing. It is usually broader than the statute, because researchers want to study a behavior rather than a charge. It has to be stable enough that two studies in two states are studying the same phenomenon.
The gap between them is a finding, not a nuisance. When the legal definition is narrower than the conceptual one, official counts will be lower than the behavior. When a state broadens a statute, recorded offenses can rise while behavior does not move at all. Chapter 2 called this a measurement artifact. In Part 2 it is one of the most useful paragraphs you can write.

Then there is the third question nobody asks

What is the unit being counted? This one decides whether two numbers can even be compared, and it is the most common source of confusion in submitted work.

You matched 0 of 6.
Say the unit out loud every time you cite a number. Not "burglary was 269" but "269 burglaries per 100,000 residents, recorded by police." Not "identity theft rose" but "reports of identity theft to the FTC rose." The unit and the source belong in the sentence, because without them the number cannot be checked and cannot be compared to anything.

Every number came from somewhere. Name the somewhere.

Eight real statements about crime. For each, work out which kind of instrument produced it. This is the core skill of Part 2, and the exam asks it in this exact form.

You named 0 of 8.
The five instruments, and the one question that separates them. Ask who had to do something for this number to exist. Police records need someone to report and an officer to record. A victim survey needs a household to be sampled and a person to answer. A self-report survey needs an offender to admit. Agency records need a case to enter a specific bureaucracy. A single study needs a researcher to go and look. Each of those requirements is also a filter, and the filter is what the instrument cannot see.

The dark figure, for your offense specifically

Chapter 2 gave you the dark figure in general. Part 2 asks you to size it for one offense. Start with the number that makes the point, which is how much of all violent crime ever reaches the police at all.

Percentage of violent victimizations reported to police in 2024, as measured by the NCVS.
70%
Violent victimizations reported
47.9%
Property victimizations reported
30.5%

Bureau of Justice Statistics, Criminal Victimization, 2024. These figures come from the legacy NCVS instrument, because BJS is releasing estimates from a redesigned survey separately. Retrieved August 2026.

Now put those two numbers next to a police-recorded count. If fewer than half of violent victimizations reach the police, then a police-recorded violent crime count is a count of something narrower than violent crime. It is not wrong. It is a count of reported and recorded violent crime, and that is a different quantity with its own trends and its own biases.

Reporting rates are not the same for every offense, and the reasons are specific. Work through the four questions below for your own offense.

One. Is there a victim who knows they were victimized? For robbery, yes and immediately. For identity theft, often months later. For price fixing or environmental violations, frequently never. Where the victim does not know, no victim survey and no police report can exist.
Two. Is there a reason to report and a reason not to? An insurance claim requires a police report, which is why motor vehicle theft has an unusually high reporting rate. Fear of the offender, shame, immigration status, and distrust of police all push the other way, and they push hardest for exactly the offenses that matter most.
Three. Does the offense have an institutional gate? Some offenses are recorded because a bureaucracy has to touch them: a hospital, a bank, a regulator, a school. Where a gate exists, the gate becomes your data source and the gate's rules become your bias.
Four. Does anyone survey for it? The NCVS asks about a specific list of personal and household offenses. If yours is not on that list, the NCVS cannot help you no matter how common your offense is.
Write this as a paragraph, not as a hedge. A weak Part 2 says the data has limitations. A strong one says which limitation, in which direction, and by roughly how much. Direction matters most. If your instrument undercounts, every trend you report is a trend in counted offenses, and a rise could mean more offending or more reporting. Say which one you think it is and why.

Where to actually look

These are the national collections most likely to have something on your offense. Each card says what it holds and what it cannot tell you. Start with the first two, because between them they cover most street offenses from both sides.

Two habits worth building now. First, always find the original collection rather than a news article about it, because the news article will not tell you the unit of count. Second, write down the data year, the geography, and the collection name the moment you take a number. You will need all three for the citation, and finding them again later takes longer than writing them down took.

Descriptions drawn from each agency's own documentation, retrieved August 2026. Coverage and the most recent data year change, so check the site rather than this page for the current release.

What to do when the data is not there

Some offenses are counted well. Some are counted badly. And some are not counted at all, by anyone, at the national level. If your offense is in the third group, that is not a problem with your topic. It is the most interesting thing about it, and Part 2 is where you say so.

The authority to cite. A 2023 National Academies report on the national data infrastructure states plainly that some crimes, such as corporate or environmental crime, are not measured in either data source, meaning neither the UCR nor the NCVS. It lists fraud against government agencies and environmental crimes as offense types measured in neither. The NCVS is a household survey, so by design it does not measure crimes against businesses and organizations. The UCR counts what police record, and most corporate harm never reaches a police department at all.

Why these offenses fall through

The victim often does not know. A person overcharged by a cartel, exposed to an illegal emission, or underpaid through wage theft frequently never learns that anything happened. No survey can ask about an event the respondent does not know occurred.

The response is regulatory, not criminal. Corporate wrongdoing is often handled by an agency through a fine or a consent decree. That case leaves a record at the agency and no record in any crime statistic, which means the more effectively an agency handles it, the less it appears in crime data.

The unit is not a person. Crime statistics count incidents, victims, and households. Much corporate harm is diffuse: a small loss spread across millions of people, or a health effect that appears years later in a population rather than in an event.

So what do you do instead?

One. Say the absence explicitly and cite it. Write that no national victimization or police collection measures your offense, name the two that would be expected to and explain why each one does not, and cite a source for the claim. That paragraph is worth more than a page of numbers about something adjacent.

Two. Move to enforcement data, and label it as enforcement data. Regulatory agencies, federal prosecutors, and sentencing bodies publish counts of cases they brought. Those counts are real and they measure enforcement activity, not offending. A drop in cases can mean less wrongdoing or a smaller enforcement budget, and you cannot tell which from the count alone.

Three. Use complaint data, and label it as complaint data. Consumer complaint systems capture what people report to them. A rise in complaints can mean more offending, more awareness, or an easier reporting form.

Four. Use individual studies for what the national systems miss. A single well-designed study of one industry or one city may be the best evidence that exists on your offense. That is a legitimate answer. Report the study's scope honestly and do not generalize past it.

The sentence that earns full credit here. Something close to: no national collection measures this offense, for these specific reasons, so the best available evidence is X, which measures Y rather than the offense itself, and here is what that means for any trend I report. If you can write that sentence about your offense, you have understood Chapter 2 better than the numbers themselves would have taught you.
This applies well beyond corporate crime. Wage theft, most environmental violations, elder financial abuse inside families, and harms that fall on people who cannot safely contact the police all have the same structure. So does any offense where the victim is an institution rather than a household. If your offense is in this category, say so early in Part 2 and build the rest of the paper on that footing.

Before you submit Part 2

Work down this list with your draft open. Click a line when it is true. Nothing here is recorded.

0 of 14 checked
That is the list. Part 2 is the part that makes Part 3 possible, because you cannot describe a pattern until you know what produced the numbers the pattern is made of.