A reporter digs into to California’s mobile homes, finding a black hole of data and a worrying lack of oversight

When I set out to write about mobile home parks in California, one of the least covered forms of housing in the state’s housing crisis, I had a sense that the problems were bad and regulation was lax. The question was: Was that true? And how do you actually prove that? 

The California State Auditor had published a report in 2020 that found the state housing department, which is in charge of regulating these homes, had not conducted full, parkwide inspections at more than half of the state’s nearly 5,000 mobile home parks in a decade, and hadn’t stepped foot for any reason in about 300 parks.

But the statistic only garnered meaning when you could show that these parks warranted closer inspection, a point advocates for park residents and even owners made to me in background interviews. That’s why I applied for the Data Fellowship from the Center for Health Journalism. I was hopeful that with state data and analysis tools like R, I could diagnose the real dimensions of the issue. 

Of course, it was not that simple because at the heart of the issue stood a lack of data.

I set out on my investigation by requesting from the state housing agency a data dictionary that could detail what was in its database on mobile home parks. Most of it was redacted, so I spoke with an official at the department to learn what I could realistically request and did so. I then asked for a list of parks that had received a parkwide inspection since 2014, and the date of inspection. In parkwide inspections, a park inspector looks to make sure the entire park is safe, including both individual lots and common areas. If they find problems, the home or park owners are put on notice to fix them.

I then requested a full list of complaints about parks starting in 2019, when the official confirmed the data was available, and information about those complaints, including what park they came from and the nature of the complaint. During an investigation in response to a complaint, an inspector might visit a park and notice other problems outside the allegation, but they are not required to investigate them, the supervisor of the State Auditor’s report said. And resources at the housing department to deal with mobile home parks were scarce. 

The idea I came up with alongside my senior fellow was simple: We would go through every recent complaint about parks in California, devise a system to rate the severity of the problem based on how much it affected the health and safety of residents, and check whether the park had been previously or subsequently checked in full or not. The hypothesis: residents at parks that were going too long without an inspection were exposed to more serious health risks. 

But the data never came.

I received a list enumerating complaints from park residents about their parks. They identified the parks they were talking about and the dates, but it lacked any sort of detail on the subject of the complaint. This posed a problem because people could be complaining about a crooked fence or a gas leak, and I had no way to know. I was told that fulfilling the request for complaint narratives, which I filed in the fall of 2022, would only be complete in August 2023. I canceled the request because the project was due in March.

This would be one of the themes I identified in my reporting: The state lacked clean data to identify risk in parks. The pattern held whether that was, as I had requested, readily accessible information on what residents alleged was going wrong at any given park, or other data such as the condition of infrastructure, property values, rent prices or even the age of the park itself. (The state had a dataset on the rough construction date, but said it was incomplete.) The housing department during interviews said the inspection system “as prescribed by law and regulation, is adequate to ensure the minimum health and safety is maintained in all the parks in the state.” 

My senior fellow and I tried to fill in the gaps by joining the datasets we did have using a magic wand known as R, a programming language I started learning as part of the data fellowship. R allowed me to do something I couldn’t do with Excel: merge different datasets using a common denominator — in this case, the park ID number. I first joined the two datasets on inspections and complaints with a third dataset containing basic information on existing parks, such as the number of homes and the park address. That provided information on how many parks had gotten no complaints and at least one inspection, or plenty of complaints with no inspections. Still, it was hard to know what to do with that data on its own.

That’s where the community engagement element kicked in, which meant getting out from behind the computer and talking to residents about their experience. I found, first, that the residents at this long list of parks the state was in charge of had little idea about the state’s role in oversight. After visiting multiple parks near me and connecting with residents all over the state through phone interviews, one woman told me she went to city hall, then to the fire department, and only afterwards filed a complaint with the housing department about a problem at her park. That squared with what advocates around the state were telling me: People didn’t know how or whom to report complaints to, and even when they did, they feared that doing so would land them in trouble, so those in more vulnerable positions avoided it. 

That unveiled an even bigger gap in the data I had on complaints: It was possible that the most serious problems were never even making it into the system. So I leaned into interviews with residents and advocates to better illustrate those data gaps. I also found through interviews with experts other types of data to illustrate the problem. For example, a government-commissioned survey in Colorado that showed two years after the state relaunched its complaint-driven inspection program, more than 70% of residents said they hadn’t heard of it. 

Along similar lines, I also set out to find data on parks in general that would illustrate why they were worth paying attention to. That’s how I learned that parks offer some of the cheapest housing (in a state with one of the worst housing crises in the country) for residents who tend to be older and poorer than the average renter. They also weren’t designed for permanence and were mostly built in the 1960s and 70s. As a result, climate change disasters such as flooding, wildfires and extreme heat are more likely to impact mobile homes than any other housing type. (These facts ring true across the country, if you’re looking for the “so what” of mobile home parks in your city or state.) Residents have dirtier drinking water, and less reliable access to it on average at mobile home parks in California. 

My data work still proved highly fruitful. I interviewed my dataset to land on a subject, and the question seemed obvious: Which park had among the most complaints filed against, and had it gotten a full inspection? 

The answer was a Stockton park with 45 complaints, and it had indeed gotten a full inspection. The plot twist was that despite heavy government intervention, conditions were nonetheless brutal. Residents dealt with sky-high piles of trash, a faulty sewage system, and later, no water or electricity for weeks. I never would have found it if not for my data work, but I filled in the data gaps with state documents detailing the complaints and inspections, court documents and repeated interviews with residents, advocates and local and state officials. 

As always, I relied on experts to put the story subject in context. What did this say about what could be happening at parks the state wasn’t visiting regularly? As one longtime housing policy consultant put it: “Parks that are that bad probably represent 50 or 60 parks in the state. But there’s the next level (of parks) that are not quite as bad. But if they don’t get inspected for 10 years, how bad will they be?”