TANF agencies generally track basic caseload trends over time, such as total families or families of certain types. These aggregate reports can be used to measure increases or decreases in the total caseload and to a lesser extent to forecast future caseloads. Moving beyond simple totals to more detailed longitudinal analyses can significantly increase the potential of data to improve caseload understanding.

Basic Aggregate Caseload Reports Longitudinal Caseload Analyses

Track total numbers of cases and recipients active by month, quarter, or year

May present aggregates for certain subpopulations (by case type, household size, location, etc.)

Illustrate increases and decreases in total caseloads (or certain subpopulations) over time

Do not require historical record of disaggregated data (i.e. individual records): can be updated by adding new totals to existing reports

Present a broader range of numbers to contextualize caseload size, such as entries, exits, and returning cases by month, quarter, or year

May present aggregates for certain subpopulations (e.g. case type, location), and can also identify cases that move across subpopulations

Demonstrate whether caseload size changes are due to the frequency of new cases coming in, former cases returning to the rolls, or a change in the rate with which cases close

Require longitudinal disaggregated data: historical individual- or case-level records at the month, quarter, or year level that can be linked over time by some kind of case or individual identifier

Longitudinal analyses give TANF agencies a deeper knowledge of system trends, supporting more accurate forecasting and more informed service planning. They can uncover turnover in a caseload of a consistent size; the policy differences between a caseload comprised mostly of the same cases lingering month after month versus a caseload with large numbers of new cases and large numbers of cases leaving each month is significant.

As compared to basic aggregate caseload reports, longitudinal analyses have exciting potential as a way for agencies to better inform policy and practice. However, their increased flexibility requires more complex data, and analysts have to make a number of different decisions to tailor the final numbers to specific policy questions. For agencies that have no prior experience with these types of analyses, taking the time to move through these decision points can be daunting. Even for agencies that have established practice in these areas, there are often questions about the right design decisions to make, or the implications of structuring analyses in different ways.

This article is the first in a series of articles that seek to organize a body of best practices and key challenges identified by state practitioners and researchers conducting longitudinal caseload analyses. Where possible, we recommend data structures and analytic designs that can answer the broadest range of policy questions as simply as possible. We also highlight common sticking points and places where different decisions can significantly impact the ultimate analytic interpretation. Expect to see these articles posted on our website in the next couple of months:

Organizing TANF Data for Caseload Analyses presents a data model and publicly available code that structure historical, disaggregated records for maximum analytic potential.

Considerations for TANF Caseload Analysis Design outlines key decision points and questions to consider in approaching longitudinal caseload analyses, including best practice recommendations from real agency applications.

Sample TANF Caseload Trend Analyses demonstrates the potential results from these analyses in more detail and discusses potential interpretations.

To see an example of the type of insights you can get from caseload dynamic analysis, click here.

These articles are based on a series of ongoing research and technical assistance efforts from Chapin Hall at the University of Chicago. For more information, contact Robert Goerge: rgoerge@chapinhall.org.

Many of the resources presented on these pages were developed with the support of the Family Self-Sufficiency Data Center, Grant Number #90PD0272, funded by the Office of Planning, Research, and Evaluation in the Administration for Children and Families, U.S. Department of Health and Human Services. The contents of these pages are solely the responsibility of the authors and do not necessarily represent the official views of the Office of Planning, Research, and Evaluation, the Administration for Children and Families, or the U.S. Department of Health and Human Services.

Where possible, we recommend data structures and analytic designs that can answer the broadest range of policy questions as simply as possible