reporting with numbers


Make clear comparisons

Audience research reveals that some kinds of quantitative comparisons are easier for people to make than others. Helping news users accurately comprehend numbers and statistics means being clear about the use of these.

Here are some things you can do to clarify comparisons:

1) Tell audiences whether differences are meaningful


A report on the battle for Nevada and Arizona’s secretary of state election states that “neither contest has a clear leader, with Arizona likely voters splitting 49% for Mark Finchem, the Republican, to 45% for Adrian Fontes, the Democrat, within the poll’s margin of error.”


A report on the race for control of Iowa’s third U.S. House district explains that “forty-seven percent of voters say they would vote for a Republican, while 44% say they would vote for a Democrat. The poll’s margin of error for the district is 6.9%.”


When polls show roughly equal numbers of respondents supporting different options, it does not make sense to use phrases like “slight majority” or “slim lead” to explain these numbers. This is especially the case when the results are within the margin of error, which is certainly the case in the second example. Here, it would be better to simply state that “polls are showing that roughly the same number of people support Democratic and Republican candidates.” The reported difference is too small to be meaningful, and it is simplest to just convey that fact.

2) Provide absolute numbers when making a comparison. Don’t use only relative numbers


A report using data from a Gallup survey states that “young people are now more than twice as likely to report smoking marijuana compared to cigarettes,” and also notes the exact percentages of respondents who reported use of each of these substances (12% for cigarettes, and 26% for marijuana).


A report discussing popular support for various Mexican political parties summarizes a poll with the words “MORENA has more than twice the support of its nearest rivals,” but does not indicate what percentage of survey respondents selected each option.


It is always best to include both absolute and relative numbers. Formulations like “twice as many people prefer X to Y” may seem clear, and easy to understand. But leaving out absolute numbers can be misleading. There’s a big difference if you’re talking 6% versus 3%, and 60% versus 30%—even if these are both “twice as many.”

3) Make comparisons in the same format. Don’t jump between different ways of presenting numbers when relaying poll results.


One news report on a poll compared three answers to a single question: “Among U.S. adults, 45 percent say their finances haven’t really changed, and another 18 percent said their situation actually improved compared to last year. But 37 percent said they have taken a hit.”


A headline about illness in schools began, "One third of teachers and 40% of students absent…"


It can be tempting to write numbers in different ways to decrease repetition, but it requires a lot of cognitive effort to compare them. Specifically, it’s easier to compare 33% and 40% than one third and 40%, or even one-third and two-fifths. People have the easiest time comparing numbers in a similar format. Fractions are only easy for our brains to compare if they all have the same denominator. (Having the same numerator can help a little bit, but not as much.)

Provide feedback.