Edward J. Fox, PhD

Data Are Not the Same as Evidence

Indian health financing research uses many kinds of data: federal budgets,
administrative enrollment records, IHS user population estimates, claims,
survey estimates, Census data, and program reports.

Each source can be useful. But each source was created for a particular purpose.
The researcher must first understand what the data actually measure before
using them to answer a financing question.

Start With the Research Question

Before choosing a dataset, write down the question the data must answer.
A question about population is different from a question about enrollment.
A question about enrollment is different from a question about utilization.
And a question about utilization is different from a question about spending.

IHS User Population Data

IHS maintains official user population measures through its National Data Warehouse.
The IHS system distinguishes among Indian Registrants, Active Indian
Registrants, and Indian User Population Estimates.

IHS describes the User Population as an official count of registered and active
users within each IHS Administrative Area. The annual official reports provide
separate columns for Indian Registrants, Active Indian Registrants, and Indian
User Population Estimates.

This distinction is important because a financing analysis that uses an IHS
user population denominator is answering a different question from one that
uses a Census AI/AN population estimate.

Active Users Are a Research Choice

The term “active user” can be useful when the research question concerns people
actually using the IHS system. But the researcher must still identify the exact
IHS definition and reporting year.

IHS publishes annual user population estimates, and the official reports can
change as data are revised or as reporting rules are applied. A study should
therefore preserve the original year and source rather than treating a current
user count as though it represented every historical period.

Measure	Useful for	Important caution
AI/AN population estimate	Population-level analysis	May include people who do not use IHS services.
Indian Registrants	IHS registration context	Registration is not the same as recent service use.
Active Indian Registrants	Understanding active participation in the IHS system	Use the IHS definition and reporting year.
Indian User Population Estimate	Official IHS user-population reporting	Do not automatically equate it with the Census AI/AN population.

Administrative Data and Survey Data Answer Different Questions

Administrative data are generated when programs operate. Enrollment files,
claims, encounter records, and IHS reporting systems can provide detailed
information about people and services within those systems.

Survey data, such as Census and American Community Survey estimates, are designed
to describe populations more broadly. They can be valuable when the research
question concerns the population outside a particular health program.

The two types of data should not be treated as interchangeable simply because
both contain a variable called “American Indian” or “Alaska Native.”

Why Definitions Matter So Much

AI/AN identification can differ across datasets. A source may use race,
ethnicity, tribal affiliation, enrollment, program eligibility, or an
administrative definition.

The Census and an administrative health program may therefore produce different
population counts without either dataset being incorrect.

The researcher needs to decide which definition matches the question and then
state that choice clearly.

Medicaid Data Add Another Layer

Medicaid research can involve enrollment data, claims, expenditures, managed-care
payments, provider information, and state administrative reports.

An enrollment count can tell us how many people are covered. It does not tell us
how much care they used or how much Medicaid paid for that care.

Likewise, a spending total may describe payments under a particular accounting
definition without identifying the number of AI/AN people who received the
services.

This is why your earlier Medicaid research questions—enrollment, utilization,
reimbursement, and population measurement—fit naturally into this segment.

CMS–IHS Data Matching

One of the most useful approaches to Indian health financing research is to connect
administrative systems rather than relying on a single dataset.

CMS and IHS data can be used together to examine relationships between people
served by Indian health programs and participation in Medicare or Medicaid.
The purpose is not simply to produce another enrollment count. The larger goal is
to understand how financing programs intersect with the Indian health population.

This approach also illustrates a broader lesson: data matching can answer
questions that neither source can answer alone.

Watch the Denominator

Many apparently different financing estimates are actually denominator problems.
If the numerator is IHS spending and the denominator is the broad AI/AN population,
the result is different from using IHS active users.

The same issue appears in Medicaid. A spending estimate divided by all AI/AN
residents is not the same measure as spending divided by Medicaid-enrolled AI/ANs.

A good research table should therefore show the numerator and denominator
explicitly rather than reporting only a per-person result.

Do Not Add Numbers Just Because They Look Compatible

The most dangerous step in multi-source financing research can be the final
calculation. Two numbers may both be labeled “spending” but refer to different
populations, services, years, or payment arrangements.

Before adding two numbers, ask:

Build a Data Dictionary

A useful research practice is to create a simple data dictionary before
performing the analysis.

For every variable, record the source, year, definition, population, unit,
and any important exclusions. This makes it easier to detect differences
before they become errors in the final estimate.

Variable	Record
Population	Definition, source, year, geographic scope
Enrollment	Program, eligibility definition, year, source
Utilization	Encounter or service definition, year, source
Spending	Appropriation, payment, reimbursement, or expenditure; year and source
Provider	IHS, Tribal, Urban Indian, outside provider, or other classification

What Makes a Strong Financing Estimate?

A strong estimate is not necessarily the one with the most decimal places.
It is the estimate whose definitions are clear, whose sources can be identified,
whose calculations can be reproduced, and whose limitations are acknowledged.

In Indian health financing, that discipline is particularly important because
the system crosses federal, state, Tribal, and private financing structures.

Questions to Keep in Mind

From Data to Policy

The course has now moved from the financing structure to the evidence needed
to measure it. The final segment asks what researchers can do with that evidence.

How can enrollment, spending, utilization, and financing data be translated into
a clearer understanding of Indian health policy?