RBA: Results-Based Accountability Explained

Prompts for This Topic
On This Page

Results-Based Accountability (RBA) is a framework from Mark Friedman that separates the well-being of whole populations from the performance of programs, and measures programs with three questions: how much did we do, how well did we do it, and is anyone better off.

Results-Based Accountability (RBA) is a framework for deciding what to measure and how to act on it. Mark Friedman built it around one split: population accountability is about the well-being of whole populations in a place, and performance accountability is about how well a program, agency or service system works for its clients. The three questions "How much did we do? How well did we do it? Is anyone better off?" are the performance side, and they are the quickest way into the framework.

RBA is common in US state and local government and human services. This page covers the two accountability types, the terms, the three performance measure questions with a worked example, and where RBA sits next to a logic model or a logframe.

The Two Parts of RBA

Population accountabilityPerformance accountability
AboutThe well-being of whole populationsThe well-being of client populations
ForCommunities, cities, counties, states, nationsPrograms, agencies and service systems
Question it answersWhat quality of life do we want for people here?Is this program working, and is anyone better off?
ExampleChildren succeed in schoolA tutoring program for third graders

This split is why RBA feels different from a logframe. A program can run well and still not move a community condition, because many partners affect that condition. RBA keeps the two apart on purpose and then links them: a program's performance measures should line up with, and contribute to, a population result.

The Terms RBA Uses

RBA fixes the meaning of three words that many programs use loosely. These definitions follow the Connecticut training deck built on Friedman's work.

TermMeaningExample
ResultA condition of well-being for children, adults, families or communitiesChildren born healthy
IndicatorA measure that helps quantify the achievement of a resultRate of low-birthweight babies
Performance measureA measure of how well a program, agency or service system is workingPercent of participants who get living wage jobs

A population result is built from three parts: a population, a geographic area and a condition of well-being. "Families in our county live in stable, affordable housing" has all three. Results and indicators belong to the population side; performance measures belong to the program side.

RBA Performance Measures: The Three Questions

Performance accountability sorts every program measure into one of three questions:

  1. How much did we do? The quantity of effort: clients served, activities delivered.
  2. How well did we do it? The quality of effort: timeliness, staff training, cost per unit, client satisfaction with the service.
  3. Is anyone better off? The effect on clients: the number and percent who changed.

The deck shows this as a grid with quantity and quality down one side and effort and effect across the other. The first two questions measure effort (the means); the third measures effect (the ends). Each question can be answered with a count (#) and a percent (%).

For "Is anyone better off?", the deck groups measures by what changed in the client:

  • Skills or knowledge, such as parenting skills
  • Attitude or opinion, such as attitude toward drug use
  • Behavior, such as school attendance
  • Circumstance, such as working or in stable housing

Worked example (illustrative)

The following is an illustrative scenario, not a documented program. A county job training program in the US Midwest sorts its measures this way:

QuestionMeasure
How much did we do?Number of persons who received training
How well did we do it?Cost per person trained; percent of intakes completed within two weeks
Is anyone better off?Number and percent of graduates in jobs that pay a living wage 12 months after exit

The first two rows and the training and unit cost measures follow the job training example in the Connecticut deck; the timeliness measure and the 12-month follow-up are illustrative. Note that the program reports both a number and a percent for the third question. A count of 40 graduates employed means little until you know whether 40 is 25 percent or 80 percent of those trained.

Choosing headline measures

A program rarely needs every measure it can think of. RBA asks you to pick a few headline measures per question, keep the rest as secondary measures, and list measures you wish you had as a data development agenda.

How RBA Uses the Measures: Turn the Curve

RBA does not stop at reporting. For a population indicator or a program performance measure, a group shows the baseline (history and forecast), discusses the story behind it, names partners with a role to play, and agrees what works and what to propose. The deck summarizes this as "turn-the-curve thinking: talk to action." If you use RBA for a program, the practical step is to report each headline measure as a trend over several periods, not a single data point.

One caution from the same deck: data are a proxy for the condition you want, not the condition itself. Using more than one indicator for a result lowers the chance you mistake the proxy for the real thing.

RBA Next to a Logic Model or Logframe

These two comparisons are this site's own reading, not Friedman's:

  • "How much did we do?" and "How well did we do it?" are close to the outputs and delivery quality in a logic model. "Is anyone better off?" is close to its outcomes.
  • The population results in RBA are broader than a program's outcomes and sit above the logframe's purpose level. A logframe states one program's results; RBA asks you to name the community result the program is contributing to.

For a framework used more widely in international development, see results-based management and the results framework. For the matrix and its indicators, see the logframe (US readers: the logic model template), and for a causal story, a theory of change. To draft measures from your own program description, use the design a results framework prompt and check every number and measure yourself.

Common Mistakes

Counting effort and calling it results. A table of clients served answers only the first question. Add at least one "Is anyone better off?" measure.

Reporting a count without a percent. The count shows size, the percent shows quality or effect. Report both where both make sense.

Holding a program accountable for a population result. A community rate moves with many partners. Report it as a shared result and hold the program to its own performance measures.

Mixing the terms. Using "outcome", "target" and "indicator" for the same thing is what RBA's fixed definitions are meant to prevent. Write the definitions on page one of your plan.

Sources

This page follows the Connecticut General Assembly training deck "RBA 101 for State Agencies" (2013), which is built on Mark Friedman's Results-Based Accountability, and Clear Impact, which supports the framework. Friedman's book Trying Hard Is Not Good Enough (2005) is the original text; its page numbers are not cited here.

Did this page help?

Need help with this work?

See services