AISP Toolkit Feb25 2025 - Flipbook - Page 18
Scale: Administrative data can allow for a population view, rather than a sample, and is
therefore less vulnerable to certain forms of bias, such as nonresponse.
Time & cost: Data reuse is less time- and resource-intensive than collecting new data.
Limitations
Reusability: Administrative data are collected to meet operational and reporting needs and
may not accurately represent the concepts or outcomes an analytic plan aims to measure.
FOUNDATIONS FOR COMMUNITY INVOLVEMENT
Quality: Data quality issues are common, including missing data and insu昀케cient data
documentation, leading to issues of reliability and validity.
Depth: These data are often one-dimensional and may need to be paired with qualitative
and other forms of data in order to address deeper questions about causal relationships,
client experiences, contextual factors, etc.
Access: Gaining access is often di昀케cult and time-consuming.
Risks
Privacy disclosure: Any transfer of data includes the risk of data being accessed
improperly.
Misuse of data for research and evaluation: Without su昀케cient data documentation,
analysts may misuse or misinterpret data.
Replicating structural racism: Administrative data are collected during the administration
of programs and services for individuals in need of social supports. These data include
people who are disproportionately living in poverty, who, as a result of the historical legacy
of race in America, are disproportionately Black, Indigenous, and people of color. Seeing
data as race-neutral is inaccurate, and such views could lead to system-level data usage
that unintentionally replicates structural racism.
Harming individuals: Certain uses of administrative data carry particularly high risks of
causing personal harm. These include uses that provide case workers, service providers,
teachers, law enforcement, etc., with personal information that could lead to biased
treatments, punitive action, or lengthened system involvement.
Harming communities: Use of administrative data, especially when mapped or otherwise
represented spatially, can create or deepen community stigma. When analysts fail to
understand and acknowledge the discriminatory practices and structural causes of
disparate outcomes by race or geography, they risk using administrative data in ways that
perpetuate harmful, de昀椀cit-based narratives.
While the particulars are important, we have identi昀椀ed broad categories of use on the risk vs.
bene昀椀t matrix below. For example, projects that involve low risk and high bene昀椀t, such as a
longitudinal program evaluation, indicator projects, or generating unduplicated counts across
programs, are generally a good idea and an easy starting point for collaboration. Conversely, projects
that are low bene昀椀t and high risk, such as sharing data from menstrual cycle tracking apps, using
social media content for predictive policing without the opportunity for public comment, or police
14
