EPI writing about numbers and data for the non-mathematically minded

September 15, 2014 7:16 PM

This is my first year in public health. I am writing a report describing data trends and prevalence rates regarding maternal and child health issues. I am trying to be a little creative in my writing. I feel as though I keep restating the same phrases, "the rate was", "a higher percentage of Asian women" , etc. Any tips on writing about data? For example, one in 4 children, one in ten women, etc. Any resources you can point me to in this type of writing? Basically, writing about numbers and data for the non-mathematically minded :)
posted by TRUELOTUS to Education (9 answers total)

I would say read how other people have described those trends. Go on google scholar and search for the highest cited papers with keywords that are similar to your key words and then read how their paper was formatted. Otherwise if you feel its too tedious put all those rates in a table and move on
posted by z11s at 10:01 PM on September 15, 2014

One strategy is to only describe the points that are most relevant to your larger point or most interesting for other reasons, and use a table or graph to present everything. You can also try to use comparative terms that make your numbers more meaningful (twice as high, half as frequently, etc.).
posted by MadamM at 10:28 PM on September 15, 2014

One good way to describe prevalence and trends is to use examples. e.g.

"10% of women suffer from depression. This means for every 100 women, on average 10 of them will be depressed"

Using actual numbers which let people visualise rates can help. It can also be less misleading than using percentages. e.g.

"Triefolane has been seen to raise that rate by 50%. This means that for 100 women on Triefolane, we would expect an average of 15 to be depressed"
posted by Cannon Fodder at 11:48 PM on September 15, 2014

If you're writing an annual report for a public health office, make it look and sound as much like its predecessors as possible, because the people using it will kinda hate you if they can't look in the usual place to find this year's update of the numbers they need. Revamping report design and tone is usually a bureaucratic committee process, not one that happens on the fly. So that's easy.

For a one-off report, you can take a bit more liberty. As MadamM notes, you don't need to comment on all the numbers in your tables and figures - pick out the most striking observations and write them up simply. "The prevalence in African-American infants was twice that in Whites." "Less than half of the low-income group was vaccinated. Infections increased fourfold between 2006 and 2014." "Disease risk in French children was 20% that in Dutch children." And, yes, "One of four children spontaneously combusted."

(I had a fight once with a boss about "the majority" - even though it literally means "more than 50%", she felt it suggested more than that. So maybe avoid "the majority". Or maybe avoid my boss.)

(And be careful with "rate". Some pedantic epidemiologists grind our teeth at the phrase "prevalence rate" because "rate" is properly used to describe only measurements per unit time. Whether point or period, prevalence is a proportion, a dimensionless ratio, because all the units, including the time unit, are the same in the numerator and the denominator and thus cancel out.)

If your setting isn't actively advocating something, be careful to avoid editorializing (which includes attributing character to your numbers - shocking increases, surprising results, steep declines, surge in incidence, and so on. A "steady decline" is acceptable - there's only so many ways to describe slopes - but avoid drama unless you're supposed to be remarking on the drama.)

Unless you're writing for a lay audience, your readers are prepared for a certain amount of repetition. It's easier to read a dull report than one that jars the reader by the epidemiologic equivalent of thesaurus abuse. And you definitely don't want to impart a tone to your report that doesn't reflect the policy of the place you're working.
posted by gingerest at 11:57 PM on September 15, 2014 [6 favorites]

I work with epis to make their writing palatable, and to clearly communicate epidemiology and other health science to a broader population.

You're right that in a lot of writing about data, you're going to be recycling the same types of sentences over and over again. Here are some tips:

• Use plain language when you can. Write how you would speak. You don't need to vary how you say this population had a higher rate of whatever than that population - if you do this too much, then you're bound to tie some sentences into unparseable knots.
• What's the most important point? Use it as a header for the section. Then, edit the entire arc of your headers for logical flow.
• Don't forget that you're not just writing about the data - you're writing about what the data tell us about maternal & child health issues. Don't get stuck on prevalence and sample sizes. Translate that to what it tells us about the health of the population in question.

Here are a few more resources that I find very helpful:

• CDC's Simply Put. This is a long document that you can keep on your desk and use as a resource. I do. It goes beyond what you are talking about, and you may discover some really useful techniques. For example, people tend to parse numbers more easily than percentages. "Roughly 1 out of every 7 (14.2%) people" is much better than "14.2 percent of respondents."
• CommunicateHealth's email bulletin We Love Health Literacy. Short, weekly emails on how to write better science communication, and the evidence behind it.

There is a lot to this whole field of communicating data well, and I'm happy to chat further about it if you'd like.
posted by entropone at 6:55 AM on September 16, 2014 [4 favorites]

Response by poster: Excellent advice! So glad I posted. Thank you!!
posted by TRUELOTUS at 8:41 AM on September 16, 2014

Two small comments:

I had a fight once with a boss about "the majority" - even though it literally means "more than 50%", she felt it suggested more than that. I agree with the boss, though you are right in a narrow sense. In most data situations (except voting, certainly) it is clearer and more correct to call 51% "half" than a "majority" when we are talking about sampled data.

Careful about percentages: I see a lot of situations where the base is unclear, and note that "three times" is better than "300%."
posted by lathrop at 9:06 AM on September 16, 2014

Try to get some popular science articles or books into your reading pile - I personally love the Best American Science and Nature writing series of books, and magazines like Discover and Seed are great too. NPR's Planet Money podcast will also give you some great samples of making tricky math topics make sense to people with no background in your field who need to understand your reports.

The more stuff you read that is targeted towards making science topics accessible towards laypeople, the better sense you will get of how to do the same thing yourself. I'd be cautious of relying too much on Google Scholar for examples, because most academic papers are specifically written for others in the field, and their formatting and jargon can be inaccessible to people whose training is in a different area.
posted by augustimagination at 2:10 PM on September 16, 2014

I read The Chicago Guide to Writing About Numbers for a stats class in grad school and it was really helpful for this.
posted by anotheraccount at 2:52 PM on September 16, 2014 [1 favorite]