When I couldn’t get the data I wanted on homeless health care, I set out to gather it

The story started before I even knew what the story was. An encampment where individuals experiencing homelessness had built a small community had just been bulldozed. I was following the story as breaking news but was interested in the long-term impacts on people’s lives.

Aside from the complicated housing process, I was also interested in how this eviction impacted people’s access to health care since the mobile health care teams would use the encampment as a home base when providing services each week. Without the encampment, nurses were reporting it was hard to find their patients.

I tried to get data from the two local medical clinics that provided street outreach to track where their efforts were focused and how that changed following the encampment eviction. I also wanted deidentified data on every patient contact for the past year. Those records were mostly not subject to public records laws as the agencies were nonprofits and largely weren’t required to provide those specifics to any governmental agency. They only offered summarized data reports.

I also tried to obtain public records that showed a log of encampments that state agencies had evicted to compare patterns with health care outreach across the state. Our lawyers our still fighting that months-long public records battle. We are slowly winning.

But ultimately, I had no data for my data project. So, under the advice of my fellowship advisor Paul Overberg, I created my own.

I built a 23-question survey designed to assess the health care needs of individuals experiencing homelessness in the Coachella Valley.

It’s relatively simple in concept but extraordinarily hard to execute. Cushion time should be budgeted for unknown challenges.

To start, I had to determine how many people I needed to interview for the data to hold any truth. I used this calculator to figure out I needed to interview 200 individuals experiencing homelessness within our region. Our population size was based off the total number of homeless individuals in our region per the most recent point in time count and we wanted a 5% margin of error.

Then, I created the survey. This is where you realize how much power you have in collecting information that has never been collected before. You have this one opportunity. What questions should you ask that you won’t be able to ask again? I started with the answers I was trying to find and worked backwards. I wanted to know what health issues were most prevalent among individuals experiencing homelessness. I wanted to know if having an easy-to-access free medical clinic and/or mobile health clinics led to fewer unnecessary emergency room visits. I wanted to know if individuals were insured or uninsured. I answered those questions and so many more I didn’t know I had.

A 23-question survey is, quite frankly, a very long survey. I wasn’t just blast-emailing this out for people to click through online at their leisure. I was sitting down and gaining trust in between every question. The survey could take anywhere from five minutes to 45 minutes per person to complete.

Some people wanted to share their life story in between each question. Others wanted to ask me for phone numbers to service providers because they wanted help. And others had a rough day and just wanted to talk about their latest girlfriend problems.

I already had deep roots in the homelessness community from prior reporting, but I also worked with local providers to help me with the surveys as well.

I worked with a local shelter and knocked out a large portion of the surveys during their dinner times. I worked with organizations that did street outreach to find new sources as well. But most of my time was spent walking through the encampments and nooks that I already knew so well. In each encampment I had a few sources I had profiled for prior stories on homelessness. While I did surveys, they would walk around with me just to chat and catch up. But what I found is that those organic relationships led to better survey collection — those sources introduced me to new people, convinced their friends who didn’t want to do the survey that is was important and showed me where other new encampments had popped up. All these partnerships, whether with agencies or with individuals, helped build trust with those I was asking to answer my survey questions. When that trust was built, it led to more truthful answers.

I also had a team of data collectors working with me as well — this was a big undertaking and it’s helpful to understand the resources you’ll need at the start of a project. In this case, the resources I needed from my editors were other reporters in our newsroom to help do the surveys.

Before we headed out, though, I held a quick meeting to teach the team how to do the surveys (you would think it is straightforward; it is not).

This is what I needed everyone to know and what I think is important for others working on gathering data in situations like this to know:

•  It is tempting to write a survey packed with scientific, complex sentences because you were taught that leads to more accurate, scientific answers. That is not the case here. I’ve learned from following along on the point-in-time counts that when simple questions are presented in such a complicated way, the respondents can become confused and unintentionally provide an inaccurate answer.

•  It is tempting to just mark down what people are telling you to get through the survey quickly. But surveyors should take the time to listen in between the lines. Ask more questions and ask the same question in different ways before marking down the final answer. An example, the question “Are you currently experiencing homelessness?” threw people off. We were focused on those who would be considered homeless based on the U.S. Department of Housing and Urban Development. So that would include individuals living unsheltered on the street, living in illegal shacks, staying at an emergency shelter, in transitional or program housing, sleeping on someone’s couch, etc. However, someone who just exited an unsheltered situation and is now living in program or transitional housing is very proud to have made that move and no longer considers themselves “homeless.” In these cases, I asked enough questions to understand that the home they were in was program housing and when I walked away I checked the “homeless” box. I did not want to skew my data or skew that individual’s perception of herself and her pride.

• The same thing goes for medical terminology. Don’t expect every person you survey to know what “diabetes,” “heart disease,” or “infectious diseases” are. Talk through symptoms and daily and weekly struggles to understand what each person’s true health needs are.

•  For these reasons, it is much better to verbally facilitate the survey than to have people fill it out themselves, though that would be much quicker. Talking through questions leads to more accurate results.

•  Finally, be patient and listen. It is easy to write someone off who is angry and argumentative and conclude they don’t want to do the survey. But it is our job to collect the data of the most vulnerable and challenging populations, so be patient. I learned that while some survey interactions appeared on the surface to be fueled by the respondents’ annoyance and anger, below the surface what they really wanted was for me to listen because they felt like no one ever had before.