Big Data Service on Campus
A new effort aims to improve data collection for on-campus fires.

By Paul D. Martin

Until very recently, when a fire occurred on a college or university campus, details about the event were hard to find. When college officials and fire safety professionals wanted to know the details of a campus fire or accurate statistics, they often had nowhere to turn.

Thanks to our work at the Center for Campus Fire Safety (CCFS), that’s all in the past. The Campus Fire Data Project, created by the CCFS and backed by the requirements of the new federal Campus Fire Safety Right-to-Know Act, is a data collection service that provides the statistical information colleges and universities need to make fire and life safety improvements to their campus infrastructure, and also presents a comprehensive and standardized repository for campus-related fire incident data. The service gathers campus fire incident data and is a launching point for an online nationwide campus fire-reporting program, one designed to provide a more accurate and defined picture of fire as it affects colleges and universities across the United States. There has yet to be a thorough study done on this kind of data, and CCFS is now poised to lead the effort with the support and expertise of partners such as NFPA and Underwriters Laboratories, who share our common mission of fire and life safety on campuses.

Initial funding was awarded in April of 2008 through a grant from the Department of Homeland Security’s Fire Prevention and Safety Grant Program. That money allowed us to begin working on the system’s database and web portal, while a team of experienced campus fire officials from CCFS determined what kind of data we wanted to collect. Development continued through 2008 and into early 2009, as we refined the desired data fields with the help of focus groups comprised of end users (campus fire safety officers) from across the country. Officially launched on May 1, the project already has more than 100 colleges, universities, or related fire departments participating. CCFS is working aggressively to expand participation and welcomes organizations who wish to serve as partners.

The Campus Fire Safety Right-to-Know Act is crucial in helping us gather the information we need. This federal legislation became law in August 2008 and requires colleges and universities that maintain on-campus housing to compile fire data, report the data to the federal government, and publish an annual fire safety report to give students, parents, and the public more accurate and current information on fires in on-campus housing. A college’s annual fire safety report, for instance, would include statistics on the number of fires in on-campus housing, the cause of each fire, the number of injuries and deaths, and the value of property damaged. The report would also include a description of fire systems, the number of fire drills, evacuation procedures, education and training programs, and future plans for fire safety improvement, as well as the institution’s policies on appliances, smoking, open flames, and other potential hazards. Colleges are also required under the new law to maintain a fire log that captures specific information about fires that occur in on-campus housing—and it is for this purpose that our data system is perfectly suited.

In the past, organizations such as NFPA would distill information reported by fire departments and include it in one of their annual reports on fires in the United States; for example, NFPA reports that colleges and universities across the United States average more than 3,300 fires per year. While those kinds of numbers provided a big-picture overview of the problem, little additional information about these fires was available. Anecdotal campus-related fire information was also compiled by several organizations, CCFS included, using mainstream media accounts, such as this excerpt from a recent newspaper story on a campus fire: "Several buildings on the…campus were evacuated Friday morning because of a fire in an underground utility tunnel. The blaze started shortly after 8 a.m., and was put out before 11 a.m. A university employee was taken to the hospital for smoke inhalation. The fire has interrupted Internet, telephone, and cable television service in some areas of campus."

Reliance on this type of media information to portray the state of campus fire safety, while it may contain some details that provide insight into the fire, is far from comprehensive, and it is neither an efficient nor a definitive method of data collection. College officials or local fire departments now working with the CCFS can input detailed fire incident data for all campus fire incidents, not just those in on-campus housing as mandated by the Campus Fire Safety Right-to-Know Act. Incident information collected directly at the source using an all-inclusive report form, with defined fields and uniform criteria, results in a broader range of fire data for the campus, a much-improved incident record, and a more complete and accurate national picture of the campus fire issue. The CCFS system lets participating institutions store their incident records as part of the larger nationwide database, and to review, sort, and analyze their specific data and even print individualized incident reports. All reports are archived online and can be downloaded and/or printed at any time by authorized users.

A few key guidelines are in place to ensure the integrity of the data. Only one account is allowed per college, fire department, or government agency; all entries are made based on CCFS-defined data fields; and CCFS will not publish or release campus-specific incident data. The full database is secure and is only available to CCFS program administrators, though participants have full rights and control over the data they input for incidents under their jurisdiction. Account holders have the ability to run reports and measure statistics for their account, while CCFS retains the ability to provide overall statistics and will report the data with a respect for campus anonymity. That means it will not publicly report any specific institutional information, but only statistics based on broad criteria, such as national, regional, campus size and type, and demographic characteristics.

This information will be used to better identify the fire problems campuses face, including causal factors, compliance with fire safety plans or polices, and even protective systems performance. From this data, we will be able to create new training programs that will be available for use by campus fire officials, programs that will help further mitigate the risk of fire on college campuses. We expect to begin evaluating the national data in early 2011, with the intent of identifying emerging trends or areas of concern. Similarly, by using the CCFS data system, individual colleges will be able to perform local reviews of their own incident data to help guide their campus fire safety programs.

We invite colleges and universities to participate, as well as fire departments that serve colleges and universities. Registration is easy and can be done at www.campusfiredata.org. Applications are reviewed by CCFS staff to assure they meet the participant criteria and to avoid duplicate accounts, and upon acceptance—usually within 72 hours—a college can be on-line and entering data. Everyone’s participation will help us better identify, and minimize, the fire problems on college campuses.

The Vacancy Problem, Redux
More statistical examination of the problem of fire in vacant homes.

By John R. Hall, Jr., Ph.D.

FPA’s Fire Analysis & Research Division has been analyzing available statistics on home vacancy and fires in vacant homes, and we are finding that some of the patterns that seem to emerge from a first-cut analysis can change significantly with more in-depth analysis. Here are a few of our findings that are worth considering alongside "Empty Threat," the cover story in the September/October issue of NFPA Journal that examined the problem of fire and vacant homes in Columbus, Ohio.

While the 2006 fire statistics show an 11 percent jump in vacant home fires, that figure may be a one-year fluctuation. The cumulative increase in vacant home fires from 2002 to 2006 is 15 percent, with most of that increase coming in 2006. The number of vacant homes was also up by 15 percent from 2002 to 2006. From those facts alone, the home fire problem hasn’t changed. All that has changed is what fraction of home fires involve vacant homes.

When a home becomes vacant, its risk of having a fire actually drops sharply, to about one-third the risk of fire in an occupied home. This finding comes from comparing the vacant home share of homes (12 to 13 percent in 2003–2006) to the vacant home share of home fires (4 to 5 percent in the same period). At the same time, when a home becomes vacant, its risk of having an intentional fire jumps sharply, to about triple the risk of an intentional fire in an occupied home. How can both of these statistics be true?

In occupied homes, intentional fires represented only about 5 percent of all fires in 2003–2006. Nearly all of the remaining 95 percent of fires are started by people, directly or indirectly, using and positioning heat sources in their homes. With no people on-site, those fires drop sharply, and that drives the steep decline in the overall home fire rate when homes become vacant. Even the threefold increase in the intentional fire rate—which translates into the 46 percent intentional share of the new, lower total home fire rate, as cited in the article — doesn’t come close to offsetting the huge decline in all other fires.

The jump in intentional fires in vacant homes is clearly a problem in its own right, even if it occurs against a background of a sharp decline in overall fire risk. It is worth noting that the intentional share of home fires is 46 percent for all vacant homes combined, but is 60 percent for vacant and unsecured homes compared to 36 percent for vacant and secured homes. In other words, it makes a huge difference whether the vacant home is secured. This may also relate to the difference between a vacant building and an abandoned building, a distinction that available databases do not allow us to factor into our statistics. Abandoned properties convey the message that no one cares what happens to them.

That may also explain the huge differences between vacant one- or two-family homes and vacant apartments. Vacant homes account for 5 to 6 percent of all home fires and 7 to 8 percent of all one- or two-family home fires, but only 1 to 2 percent of all apartment fires. A vacant apartment that is not secured may still achieve a certain level of security by being located in an apartment building with many occupied apartments.

Another problem that becomes worse when homes are vacant is fire size. Flame damage spread beyond the structure in 8 percent of the fires in secured vacant homes, 11 percent in unsecured vacant homes, and 10 percent in all vacant homes combined, compared to only 3 percent of all home structure fires. Vacant homes are less likely to have a reported fire than occupied homes, but if a fire occurs, it is more likely to spread. These two effects nearly cancel each other out; 10 percent of a fire rate that is one-third the size of a second fire rate is about the same as 3 percent of that second fire rate. In other words, the rate of fires large enough to threaten adjacent property is about the same for vacant and occupied homes, but those fires represent a larger share of the vacant home fires than of the occupied home fires.

What strategies should we pursue with this problem? It is natural to think first of preventing vacant home fires by preventing home vacancy, but home vacancy is not that new a phenomenon and is not necessarily a bad thing. Home vacancy rates have been drifting upward for a long time, rising from around 9 percent of housing units in the early 1980s to 14 percent in 2008. Some of that increase may have been related to the building of the housing bubble, with associated overbuilding of housing ahead of what was assumed to be a steadily rising demand. Therefore, some of those vacant homes may never have been occupied, while others may have been briefly vacant between the departure of one household and the arrival of another. Some home vacancy is like some unemployment: a natural and unthreatening measure of the gaps in moving from one occupant to another. How do you address home abandonment without interfering in, or being distracted by, the "natural" level of home vacancy?

Most strategies lead back to the home owners. It makes a considerable difference whether the owner is the year-round occupant, a seasonal occupant, an absentee landlord, or the holder of a defaulted mortgage. Local options are different with these different types of ownership, whether the goal is to prevent vacancy or only to secure and protect the property. "Empty Threat" dwelt at some length on the ownership situation in Columbus, and many communities may recognize their own current or impending situations in those observations.

Perhaps the real threat posed by home vacancy is not any direct effect on the fire problem but rather that it can be the first step on a slippery slope. Today’s vacant home can become tomorrow’s abandoned home. If the home is illegally occupied, it may be by homeless people bringing the fire problems associated with makeshift living arrangements or by criminal enterprises such as crack houses. If the home is not occupied and is unsecured, any vandalism, burglary, or arson fires will leave damage that is likely to go unrepaired. Any of these events can create detrimental ripple effects on the entire neighborhood.

The bottom line seems to be that both prevention of vacancy and assurance of security have their places as strategies, but we need to better understand the problem and the impact on the problem of each strategy. More importantly, any strategy’s chances of success trace back to our ability to solve problems through owners. A foreclosed home is owned by a financial institution. It should be possible to force those entities to secure and protect their properties or provide some other workable solution (which may involve demolition in some cases), but it may take changes at the state or national level to give localities the power to compel those actions. At least these are points worthy of further discussion, and it is my hope that these additional statistics will help support that discussion.