Interpreting Numbers in Context

By Karen Sternheimer

In the age of big data, one of the most important—and overlooked—skills that training in sociology provides is the ability to interpret numerical data. Being statistically literate is important for so many reasons, not the least being that it ultimately can help you find a job. Even if you aren’t a statistician or data analyst, knowing how to understand numbers can give you a leg up among the math phobic in many professions.

You don’t have to fall in love with equations or mathematical theory to become skilled at interpreting data. The most important thing to keep in mind is that numbers tell a story, and your job as an interpreter of data is to figure out what story they are telling, and share that story with others.

Telling a good story begins with helping your reader or listener understand why the story matters. Why should they care? That’s easy when it comes to data on crime, as most people have at least some trepidation about becoming a crime victim. So let’s use some data from the FBI’s 2016 Crime in the United States report, compiled from the FBI’s Uniform Crime Reporting Program (UCR) to illustrate how to tell a story (in this case, a true story) using data.

The story begins with the data collection process. We have to know about where the data come from in order to provide clear details. In this case, the UCR comes from crimes reported to police, which are then reported to the FBI from more than 18,000 agencies, using clearly defined definitions of crime so that the information can be compiled into a comprehensive report.

So if a crime is not reported to the police, it will not be part of the story that the UCR data are telling. The UCR includes information on arrests, but not regarding whether those arrested were ever tried or convicted.

From the most recent data available, we can see that there were 1,248,185 violent crimes reported to police in 2016 nationwide. First, we need to understand the definition of “violent crimes.” The FBI includes homicide, rape, robbery, and aggravated assault in this measure.

How many were in each category? How many of those were murders? According to Table 1, of that number, 17,250 were classified as homicides. That is a large number, but one that is much smaller (fortunately) than the total number of violent crimes.

Next, we need some context. The million plus violent crimes seems like a lot, but what does it mean in conjunction with other information? The raw number of crimes, or crime volume, is very limited in its usefulness in telling a clear story. The main story these numbers tell is wow, that’s a lot of violence! Let’s think about how to tell a more nuanced way.

Percentages are one way to provide more context. Of all the violent crimes reported to police in 2016, 64 percent were aggravated assaults, about 27 percent were robberies, 8 percent rapes, and murders were less than 2 percent. Suddenly the story changes when we learn that the most serious crimes (rape and murder) are fortunately the least common of the violent crimes measured.

Rates are another useful measure to understand numbers in context. While percentages help us to compare the number of violent crimes within each category, rates are measures that can help us understand a number in the context of the larger population. Remember that gigantic number of 1,248,185 violent crimes reported to police? We need to know the context of the population size to make sense of this information.

In 2016, the U.S. population was estimated to be 323,127,513. To calculate the violent crime rate, we need to compare the crime volume to the population, or divide 1,248,185 by 323,127,513, which gives us .00386, a very small number that is hard to make sense of on its own.

But if we multiplied that number by 1,000 or even by 100,000, then it becomes easier to interpret. Multiplied by 1,000, we would have a violent crime rate of 3.86 per thousand, meaning that out of every thousand people in the population, 3.86 were victims of violent crime.

Because serious violent crime is often so rare, we typically multiply by 100,000 so we would say that there were 386 victims for every hundred thousand people in the population. This is especially useful for homicide; if we divide 17,250 by the population size of 323,127,513 we get .00005338. Multiplied by 100,000, we find our rate is 5.3, meaning that for every hundred thousand people in the population, about 5 were victims of homicide.

Once we have rates, we can understand long-term crime trends. Because rates adjust for population changes, the growth or decline in population size is accounted for, unlike when we look at crime volume alone. This allows us to compare how rates have changed in the short-term and long-term.

If we look at data on violent crime over the last twenty years we can see that violent crime and homicide are up from 2015, but still lower than ten years ago, and even lower still from twenty years ago. The FBI website has easy access to data from 1995, and we can compare violent crime rates even going even further back if we want to.

Hopefully you can see how looking at just one or two numbers is a lot less meaningful than presenting them in context, using percentages and rates. Once you get comfortable interpreting numbers in context, you can do your own calculations using ratios and calculating percentage increases and decreases. The stories are endless, and endlessly interesting.

Comments

[researchers]

If as students, researchers, and professors, we collectively plan to address social issues, then we must learn to work with our neighbors outside of the academic setting. One way to prevent research fatigue and ensure that researchers are not engaging in studies that may harm participants is to follow the sociologists code of ethics:

[See article]

A very useful and informative information provided by you.