The Work of a Data Analyst
In one Midwestern city, a group of physicians considers whether or not they can maintain their practice. Auto plants have shut down, factory workers have lost their jobs, and healthcare benefits have long-since expired.
Many residents of the area have no healthcare, so for these doctors, treating them is like working pro bono for their neighbors. Even more of their patients now qualify for public aid, however, so treating them is increasingly like working pro bono for the government.
Very likely these physicians could be hired on at the local hospital, becoming employees of big medicine, should they choose to take down their own shingle. But for now, they are weighing their options, crunching the numbers.
And they're relying on me to give them data.
I’ve never been much of a numbers person.
That’s why most people are surprised when I tell them I'm a data analyst. I work with decimals and dollar signs all day long. But being a data analyst is less about the numbers and more about the stories they tell.
Recently, I walked into a coworker’s office and handed her two invoices from one of the medical practices for whom our company provides billing services. “These were entered incorrectly,” I told her.
She stared at me.
I went on: “The charges marked $0 also need to be entered, but they weren’t.”
She continued to stare, then said, “Right . . . but why do you have them?”
“You’d be amazed what I can find!” I told her. Because it wasn’t like it was my job to monitor her team’s work flow. But it was my job to analyze the data. And when the total patient count in our billing system didn’t match the total patient count we had received from our client, the analysis began.
First, it was two numbers: 458 and 456. But then it was two days, two providers, two patients, two individual charges overlooked by one medical billing specialist who wasn’t trained properly or who was just having a bad day.
It started with the numbers, but they ended up telling a story. And I was the narrator.
The questions I'm asked all day long by account executives and clients aren’t questions you would ask a numbers person. They are questions you would ask a story teller: who (patient demographics), what (procedure utilizations), when (month by month tracking), where (zip code mapping), why (collection percentages), how (revenue projections)?
The difficult side of data analysis as storytelling, however, is that the numbers often tell more than one story, or they tell a different story than expected.
Our CEO and I recently were looking at some reports for a client, trying to determine the financial impact of a recent fee increase. She and I were looking at the same numbers, but we were hearing different stories. I heard a straightforward one: since the revenue began to increase the same month the prices changed, there was obviously a connection. She heard a more nuanced one: sure the price change and the revenue increase happened at the same time, but so did some other economic and practice management factors, ones that were harder to quantify.
We agreed that those particular facts told too many stories to rely on them alone. So we gathered more data. And then some more. Eventually, all the numbers together told us what we needed to know.
That’s how data works. One statistic alone often tells a different story than looking at several pieces together. A client’s payer mix might tell us the story of a local economy or the changing trends in healthcare. Looking at a single patient’s insurance history might tell us the story of an aging man who recently retired and is now widowed. The numbers can tell us these things.
That’s why I often tell people that they need to trust the numbers. All of our hypotheses and hunches can usually be proven or disproved when we let the numbers tell us their own story. We can’t, however, always trust ourselves to hear or understand the story they're telling. So I also tell people this, “These numbers only say what they're saying.”
I’m still not much of a numbers person, but I do like to hear what they have to say. That’s the surprising work of a data analyst.
Image by SamahR. Used by permission via Flickr. Post by Charity Singleton.