NFIRS Representativeness 2011 final

U.S. Fire Administration

NFIRS
Representativeness
Study
October 2011

Table of Contents
ABOUT THE NATIONAL FIRE INCIDENT REPORTING SYSTEM............................................1
BACKGROUND AND PURPOSE OF AN NFIRS REPRESENTATIVENESS STUDY ................ 3
PAST NFIRS REPRESENTATIVENESS STUDIES ....................................................................4
SCOPE AND METHODOLOGY .................................................................................................5
RESULTS ..................................................................................................................................5
CONCLUSION ............................................................................................................................6
RECOMMENDATIONS...............................................................................................................7

About the National Fire Incident Reporting System
The National Fire Incident Reporting System (NFIRS) is a voluntary, all incident based data
collection system. Section 9b of the Fire Prevention and Control Act, Public Law (PL 93-498)
specifies that the U.S. Fire Administration (USFA) develop standardized data reporting methods
and encourage and assist State, local, and other agencies, public and private, in developing and
reporting information. NFIRS was developed specifically to address these two requirements.
All fire departments in the United States and the U.S. Territories are eligible to participate in
reporting to NFIRS on a voluntary basis. Participating fire departments collect a common core of
information on an incident and any casualties that ensue by using a common set of definitions. In
a few States, the data may be collected using paper forms, however, most of the data are
submitted electronically. Local agencies forward the completed NFIRS modules to the State
agency responsible for NFIRS data. The State agency combines the information with data from
other fire departments into a statewide database and then transmits the data to the National Fire
Data Center (NFDC) at USFA. Data on individual incidents and casualties are preserved incident
by incident at local, State, and national levels. Once limited to fire incidents only, NFIRS now
encompasses all incidents to which the fire department responds—fire, emergency medical
services (EMS), hazardous materials (Hazmat), and the like. 1
For 2009, more than 1.1 million fire incidents were reported to NFIRS by approximately 69
percent of all fire departments in the United States. Based on data from the National Fire
Protection Association (NFPA), there was an annual average estimate of 30,175 fire departments
in the United States for the years 2007 to 2009. For this same three year period, 23,747
departments reported fires to NFIRS in at least one of these three years. This is about 79 percent
of the departments estimated to be in the United States. NFIRS has seen approximately an 85
percent increase in the number of departments reporting fires between 1999 and 2009 (the last
year for which NFIRS data were available at the time this study was undertaken). In 1999, there
were 11,067 departments reporting fires to NFIRS and in 2009, there were 20,421 departments
reporting fires. Some of this increase can be explained by the number of fire departments
receiving Assistance to Firefighters Grants. If a fire department is a recipient of one of these
grants, then NFIRS participation is required. 2

1

U.S. Fire Administration’s, Fire in the United States 2003-2007, 15th Edition, October 2009.
From the Assistance to Firefighters Grant Program guidance, if the applicant is a fire department, the department
must agree to provide information, through established reporting channels, to NFIRS for the period covered by the
assistance. If a fire department does not currently participate in the incident reporting system and does not have the
capacity to report at the time of the award, the department must agree to provide information to the system for a 12month period that begins as soon as the department develops the capacity to report. See
http://www.fema.gov/firegrants/docs/2010_AFGguidance.pdf.

2

1

NFIRS data are used extensively at all levels of government for major fire protection-related
decisions. At the local level, incident and casualty information are used for setting priorities and
targeting resources. The data collected are particularly useful for designing fire prevention and
educational programs and EMS-related activities specifically suited to the real emergency
problems the local community faces.
At the State level, NFIRS is used in many capacities. Many States mandate NFIRS reporting
departments for their own purposes. One valuable contribution is that some State legislatures use
these data to justify budgets and to pass important bills on fire-related issues such as sprinklers,
fireworks, and arson.
Many Federal agencies, in addition to USFA, make use of NFIRS data. NFIRS data are used, for
example, by the Consumer Product Safety Commission (CPSC) to identify problem products and
to monitor corrective actions. The Department of Transportation (DOT) uses NFIRS data to
identify fire problems in automobiles, which has resulted in mandated recalls. The Department of
Housing and Urban Development (HUD) uses NFIRS to evaluate safety of manufactured
housing (mobile homes).The USFA uses the data to design prevention programs, to order
firefighter safety priorities, to assist in the development of training courses at the National Fire
Academy (NFA), and for a host of other purposes. 3
It is important to note that NFIRS data are used by USFA and other Federal agencies to derive
national fire data estimates. USFA does not use NFIRS data to produce State or county level fire
estimates.

3

U.S. Fire Administration’s, Fire in the United States 2003-2007, 15th Edition, October 2009.

2

Background and Purpose of an NFIRS Representativeness
Study
In the past, the Office of Management and Budget (OMB) has questioned why the NFIRS data
collection should be a census of incidents for all departments rather than a statistical sample.
USFA has investigated the possibility of sampling and the issues surrounding it. While there are
certain advantages to a statistical sampling methodology, USFA has not undertaken a sampling
approach to fire data collection for several reasons. As previously mentioned, USFA is directed
by Public Law to develop a standardized incident data reporting method and assist local and
State agencies in reporting incident data to this system. Because NFIRS is used at the local,
State, and Federal levels, abandoning it for a statistical sampling method would adversely impact
State and local fire department data collection and the NFIRS standard, which is also used
internationally. Additionally, much of the cost burden of the current NFIRS data collection is
carried by the State NFIRS operations. USFA has relied heavily on States as cooperative partners
in bearing the costs and resources of maintaining the system, and the States have relied on USFA
to shoulder the development costs. Switching to a sampling method would mean USFA would
need to acquire additional funding and personnel to design and maintain a new sampling system,
leaving the State and local entities to wholly cover the cost of a data collection system.
OMB directed USFA to complete a study regarding the bias in NFIRS due to nonresponse. As a
result, USFA completed this study to identify the percentage of fire departments that responded
to the NFPA survey and also reported fire incidents to NFIRS to aid in examining the
representativeness of NFIRS.
In addition, based on OMB’s instructions for previous terms of clearance for the NFIRS form
renewal, this study identifies how USFA might continue efforts to encourage even more
participation in the all incident based data collection system.

3

Past NFIRS Representativeness Studies
Prior to this current representativeness study and at the direction of OMB, USFA has conducted
several analyses to examine the extent of bias in NFIRS due to nonresponse. Brief summaries of
the past studies are described as follows.
In a joint study effort in 2005, USFA and NFPA examined the biases in NFIRS participation,
specifically whether the fire experience of NFIRS-reporting departments differed
systematically from the fire experience of other non-reporting departments within the same
population. Results based on data from 1997 and 2002 indicated that there were differences in
total fire loss estimates derived from NFIRS reporting departments and non-NFIRS reporting
departments; however, the degree of difference was not great enough to merit adjusting current
scaling methodologies. 4
In the fall of 2008, USFA undertook a study of the NFIRS data set to examine the potential bias
in NFIRS due to fire department nonresponse. As a result, USFA completed an analysis to
identify fire departments that did not participate in NFIRS, characteristics of these departments,
and whether their nonresponse impacted the representativeness of NFIRS. Undertaken on a
regional and county basis, the review provided insight into what, if any, adjustments could be
made to minimize the impact of possible reporting bias on the fire loss estimates. States of
particular concern for nonreporting were located in the Northeast and West regions of the
country where the average rates of reporting were about 72 percent for each of these regions. By
contrast, the Midwest region had an estimated 87 percent reporting rate. 5 Based on the findings
of the study, USFA concluded that NFIRS reporting departments were representative of the
universe of all fire departments in the United States.

4
5

U.S. Fire Administration’s, Fire in the United States 2003-2007, 15th Edition, October 2009.
U.S. Fire Administration’s, Fire in the United States 2003-2007, 15th Edition, October 2009.

4

Scope and Methodology
To complete this NFIRS Representativeness Study, USFA matched the NFIRS database to
NFPA proprietary data to determine the percentage of departments responding to the NFPA
survey that also reported fires to NFIRS. Upon a special data request, NFPA provided USFA
with a file of fire departments that responded to the 2009 NFPA Survey of Fire Departments for
United States Fire Experience. This survey is based on a stratified random sample of fire
departments where the strata are based on the size of community protected. A total of 2,732
departments responded to the survey. Estimates derived from the survey are based on ratio
estimation methods. Additional information regarding the NFPA survey methodology is
described in NFPA’s report, Fire Loss in the United States During 2009:
http://www.nfpa.org/assets/files/PDF/OS.fireloss.pdf.
The file of 2,732 fire departments provided by NFPA included the fire department name and
address. To match the NFIRS database to the NFPA data, the State and Fire Department ID
(FDID) fields are required to be able to identify unique fire departments within each State.
Because NFPA does not collect the FDID number as part of their annual survey, USFA
conducted a clerical process to add the FDID field to the NFPA file of fire departments.
Because NFIRS is a voluntary system, departments that report fires in one particular year may
not report fires in the next year. As a result, USFA used the 2007-2009 NFIRS databases in this
analysis to account for fluctuations in the year to year reporting of fire departments. In fire data
analyses at the national level in general, USFA uses at least three years of data to account for
variations in fire department participation each year.
The State and FDID fields were used to determine the unique number of fire departments that
reported fires during this time period and to match against the NFPA file of departments. As
noted earlier, from 2007 to 2009, a total of 23,747 departments reported fires to NFIRS.

Results
The NFPA file was initially matched to the 2009 NFIRS file of reported fires by State and FDID
fields to determine the percentage of departments responding to the NFPA survey that also
reported fires to NFIRS. This automated match resulted in 2,155 NFPA survey respondents that
also reported fires to NFIRS. That is, 78 percent of the NFPA respondents also reported fires to
NFIRS in 2009. The 577 departments that did not match were examined further to see if they, in
fact, had reported to NFIRS. A manual comparison of these departments was completed and an
additional 80 departments were identified as reporting to NFIRS. Then an automated match of
the remaining 497 nonmatching NFPA departments was undertaken to determine if these
departments reported fires in NFIRS in 2007 or 2008. An additional 98 departments and 52
departments were found to be reporting fires to NFIRS in 2008 and 2007, respectively. Thus, of
the 2,732 NFPA survey respondents, a total of 2,385 departments (87 percent) reported fires to
NFIRS from 2007 to 2009.
5

Conclusion
Based on the following, the USFA concludes that the distribution of participants is reasonably
representative of the entire Nation, even though the sample is not random. There is no known
major bias that will affect the results.
• 87 percent of the 2009 NFPA survey respondents were also determined to be
reporting fire incidents to NFIRS from 2007 to 2009;
• NFIRS collects about one million fire incidents, or about two-thirds of all fire
incidents each year;
• Data collected are reasonably well distributed geographically and by size of
community; and,
• In 2009 alone, more than 18,000 additional departments (i.e., in addition to those
responding to the NFPA survey) reported fires to NFIRS.
Most of the NFIRS data exhibit stability from one year to another, without radical changes.
Results based on the full data set are generally similar to those based on part of the data, another
indication of data reliability. Although improvements could be made—the individual incident
reports could and should be filled out more completely and more accurately than they are today
(as can be said about most real-world data collections as large as NFIRS), and all participating
departments should have the same reporting requirements—the overall portrayal is a reasonably
accurate description of the fire situation in the United States. 6
It is important to reiterate that USFA along with other Federal agencies do not use NFIRS data to
derive State level fire estimates. NFIRS data are used to show the fire problem at the national
level. Because the findings in USFA’s NFIRS Representativeness Study show a very high
percentage of the NFPA respondents are also reporting fires to NFIRS, fire departments across
the country, appear to be well represented in NFIRS. Therefore, USFA concludes that NFIRS
reporting departments are representative of the universe of all fire departments in the United
States. Additionally, there is no bias in NFIRS due to department nonresponse at the national
level that would have meaningful impact on national level estimates.

6

U.S. Fire Administration’s, Fire in the United States 2003-2007, 15th Edition, October 2009.

6

Recommendations
Based on this NFIRS Representativeness Study, it is recommended that USFA continue to work
with NFIRS State Program Managers to encourage fire department participation in NFIRS.
USFA proposes that NFIRS State Program Managers collaborate with nonreporting fire
departments to encourage participation by emphasizing the importance of fire data analysis and
how fire departments can use their data to show the need for funding and additional resources.
Additionally, the U.S. Congress authorized and funded (beginning FY2009) USFA to develop
Enhancements to NFIRS. These include a simplified NFIRS web-based reporting interface and a
data warehouse for generating output reports for use in analyses. These improvements make
reporting and accessing the NFIRS data much easier for fire departments.
USFA completed and deployed the new web-based data entry tool in the summer of 2010. The
Data Entry Browser Interface (DEBI) is a one purpose tool for use by the fire service to
document incident information within NFIRS. While the functionality is the same as the NFIRS
client Data Entry Tool (DET) that has been available for use for many years, DEBI allows entry
of incidents using a standard web browser, eliminating the need to download, install, and
configure client software.
The development of a flexible NFIRS data warehouse with comprehensive data mining
capabilities was completed in July of 2011. It is scheduled for deployment to national, State and
fire department NFIRS users in three phases beginning in October of 2011. The data warehouse
allows NFIRS users to access and report on nationally collected data with significantly increased
functionality over the current report generation tool. The data has been transformed into a
custom schema that greatly increases the speed of report generation and data access. NFIRS
users are also now able to generate reports using data from other departments and States, which
was not previously possible.
All States will be encouraged to develop programs to increase participation. State Program
Managers will also be encouraged to continue to work with departments currently reporting to
enhance data quality through training programs. USFA offers several NFIRS training courses for
fire departments which include the NFIRS Program Manager course, NFIRS 5.0 Self-Study, and
the Introduction to NFIRS 5.0. The NFIRS Program Manager course enables participants to
promote, support, and manage NFIRS data collection successfully. The NFIRS 5.0 Self-Study
(online) course provides an overview of the data collection system, its modules and data
conversion issues. The Introduction to NFIRS 5.0 course emphasizes how to use standardized
forms to achieve uniformity in incident and activity reporting. This training program is designed
specifically to support local fire service organizations, and assists them in providing data both to
management and decision makers, as well as to the State uniform fire reporting system.
Additionally, USFA’s NFIRS Support Center offers a consolidated national help desk to provide
technical support regarding all aspects of NFIRS.

7