An ABA percent of goal obtained calculator compares the observed movement between two phase means with the recorded distance from baseline to a goal. The arithmetic takes one line; establishing which goal belongs in that line can take considerably more care. This worksheet calculates the mean-based form of percent of goal obtained, shows the numerator and denominator separately, and adds a sensitivity row only when two pre-existing goal records conflict. No target is created after seeing the data, and no record is favored because it yields a larger percentage.

Clinicians & ABA Professionals / Data, Outcomes and Clinical Decision-Making.

The question this worksheet can answer

For one adjacent Phase A and Phase B comparison, the tool asks: how much of the recorded distance from the Phase A mean to a specified goal is represented by the observed phase-mean change?

It reports the valid count and arithmetic mean for each phase, the observed mean change, the baseline-to-goal distance, and the mean-based percent of goal obtained. The method is commonly abbreviated PoGO_M to distinguish the phase-mean form from regression-based, trend-adjusted versions.

A defensible ABA percent of goal obtained calculator keeps those components visible so the percentage can be traced back to the chosen goal and observed data.

The output is a ratio expressed as a percentage. Zero means the Phase B mean did not move from the Phase A mean. One hundred means the Phase B mean equals the entered goal. The number is not restricted to that interval: movement away from the goal can produce a negative value, and movement beyond the goal can produce a value above 100.

Those arithmetic meanings do not establish whether the goal is socially important, whether the change is attributable to an intervention, or whether the result should drive a treatment decision. A goal-linked coefficient remains a supplement to the raw graph, design, measurement record, client experience, and qualified review.

Start with the goal record, not the result

Do not derive a goal from the observed Phase B data. Begin with the controlling record that existed before this analysis and capture enough context for someone else to verify it:

  • the exact goal value and the document, version, date, and author from which it came;
  • the outcome definition, unit, opportunity or exposure base, and observation duration;
  • the scale boundary or clinical rationale supporting the goal;
  • whether the same recording procedure applies to the phase data and the goal;
  • the person and stakeholder participation used to establish or review the goal;
  • any applicable consent, assent, access, safety, payer, or organizational limits;
  • the intended decision and the qualified reviewer accountable for interpretation.

A goal can be a meaningful scale endpoint or another justified target criterion. It cannot be a convenient number selected to improve a retrospective result. If no documented and scale-valid goal exists, report ineligible for PoGO_M rather than filling the blank with an assumed ideal.

The goal, Phase A values, and Phase B values must share the same measurement meaning. Ten independent responses per ten opportunities is not interchangeable with ten responses per hour. A percentage of intervals is not interchangeable with a count. If observation opportunities, prompt criteria, response definitions, or session durations changed, resolve comparability before calculation.

Follow the signs all the way through

The formula handles upward and downward goals without a separate sign reversal. When the goal is above the baseline mean, the denominator is positive; movement upward toward it produces a positive numerator. When the goal is below baseline, both the denominator and a movement downward toward it are negative, so their ratio is positive.

Record the intended direction anyway. It lets a reviewer detect an entry error, a goal on the wrong side of baseline, or a result that moved away from the target. If the goal equals the Phase A mean, the denominator is zero and the calculation is undefined. If it is very close to baseline, a small observed shift can produce an unstable large percentage. Show the goal distance rather than hiding that sensitivity.

This page does not clamp negative results to zero or values above 100 to 100. A boundary crossing may be real arithmetic, an entry mistake, a changed measurement rule, or evidence that the recorded goal needs contextual review. Preserve the calculated value and investigate rather than silently editing it.

Copyable eligibility and provenance table

Review itemLocked valueEvidence or unresolved issueOutcome and operational definitionUnit and opportunity or exposure basePhase A and Phase B labelsOrdered valid observationsDocumented goal valueGoal source, author, date, and versionGoal set before this analysis?yes / no / unresolvedSame measurement rule for goal and phases?Goal differs from Phase A mean?Higher or lower direction intendedBaseline trend and variability reviewedMissing, corrected, late, or disputed rowsClient and stakeholder inputAnalyst, reviewer, and review date

PhaseOrderDate or sequenceRaw valueStatusMeasurement or context noteA1valid / missing / invalid / disputedA2A3B1B2B3

An empty observation does not become zero. Keep invalid rows and corrections in an audit trail outside the active numeric list. A new value, boundary, goal record, or measurement definition creates a new calculation version.

Calculate the mean-based index

Let:

  • mean A be the arithmetic mean of valid Phase A observations;
  • mean B be the arithmetic mean of valid Phase B observations;
  • G be the documented same-unit goal;
  • C = mean B - mean A be the observed phase-mean change;
  • T = G - mean A be the baseline-to-goal distance.

The calculation is:

PoGO_M = 100 x C / T

or, written in one line:

PoGO_M = 100 x (mean B - mean A) / (G - mean A)

Stop when T = 0. Flag a small |T| for methods review because denominator sensitivity can dominate the result. Retain full precision and round only the displayed output under a declared rule.

The current SingleCaseES PoGO reference documents this mean-based point estimate and also describes approximate inferential calculations. This worksheet implements only the point estimate. It provides no standard error, p-value, confidence interval, autocorrelation adjustment, or trend correction.

Blank calculation and conflict ledger

OutputPrimary goal recordConflicting record sensitivityCheckValid Phase A countMatches raw rowsValid Phase B countMatches raw rowsMean AMean BObserved change, Cmean B - mean AGoal, GSame unit and ruleGoal distance, TG - mean AMean-based PoGO_M100 x C / TResult relationbelow 0 / 0 to 100 / above 100Goal source and versionEligibilityeligible / ineligible / unresolved

The sensitivity column has a narrow purpose: both goal values must already exist in the record, and the conflict must matter to this analysis. A set of hypothetical goals is not a sensitivity analysis. Nor should an analyst search for the highest percentage or quietly promote one unresolved record without identifying the decision authority.

Mateo's fictional worked example

Mateo is a fictional BCBA reviewing independent responses per ten documented opportunities. Higher values are toward the recorded goal. Phase A is 3, 4, 5, 4, 4, and Phase B is 6, 7, 8, 7, 7. All values are synthetic.

The phase totals and means are:

sum A = 20; mean A = 20 / 5 = 4

sum B = 35; mean B = 35 / 5 = 7

The signed implementation plan lists a goal of 10 independent responses per ten opportunities. The observed change is C = 7 - 4 = 3. The goal distance is T = 10 - 4 = 6.

PoGO_M = 100 x 3 / 6 = 50%

The result means that the observed phase-mean movement equals one half of the recorded numeric distance from the baseline mean to goal 10. Nothing in the 50 percent result establishes intervention effectiveness, half-completed treatment, or identical movement in every observation.

When two goal records disagree

A second signed implementation sheet lists goal 9 under the same unit and response rule. The team has not yet established which version controlled the phase. Mateo does not replace goal 10. He adds a sensitivity row and flags the conflict for resolution.

The numerator remains 3. Under goal 9, T = 9 - 4 = 5:

PoGO_M sensitivity = 100 x 3 / 5 = 60%

StateMean AMean BGoalObserved changeGoal distancePoGO_MPrimary recorded goal47103650%Conflicting record sensitivity4793560%

The observed Phase B data did not improve when the percentage changed from 50 to 60. Only the recorded goal distance became shorter. Until the provenance conflict is resolved, both rows remain visible and the decision note remains open. Selecting goal 9 because it yields the larger percentage would reverse the purpose of the audit trail.

Read values outside zero through 100 without repairing them

Suppose the Phase B mean were 2 while goal remained 10. Then C = -2, T = 6, and PoGO would be negative, showing movement away from that upward goal. If the Phase B mean were 11, PoGO would exceed 100 because the mean moved beyond goal 10. Neither result should be truncated.

For a downward goal, imagine mean A 12, mean B 8, and goal 4. Then C = -4 and T = -8, so PoGO is positive 50 percent. This illustrates why the raw numerator, denominator, goal, and direction should accompany the percentage.

These small examples explain arithmetic only. They do not define an acceptable goal, phase length, progress threshold, discharge rule, authorization criterion, or treatment recommendation.

Put the percentage back in clinical context

The original Ferron, Goldstein, Olszewski, and Rohrer article introduced percent of goal obtained for single-case designs. The current Methods Guide for Effect Estimation and Synthesis of Single-Case Studies describes its eligibility and assumptions, including that a goal is required and that estimates can fall below zero or above 100. The guide also distinguishes the phase-mean form from trend-adjusted approaches.

This tool does not extrapolate baseline trend. It assumes nothing about what the series would have done without the phase change. Inspect level, trend, variability, immediacy, overlap, consistency, measurement quality, treatment integrity, and alternative explanations on the complete graph. A mean shift can coincide with maturation, schedule changes, altered opportunities, a new observer, another service, or other events.

Version 5.0 of the What Works Clearinghouse handbook places systematic visual analysis and repeated demonstrations at the center of single-case evidence review. The handbook does not endorse this worksheet or let one goal-linked phase comparison establish a functional relation.

Goal attainment is also not the same as social validity. Ask the person, using accessible communication, whether the outcome and target remain meaningful, whether benefits and burdens are acceptable, and whether supports protect dignity, choice, assent, ordinary communication, mobility, and safety. Record disagreement instead of converting it into a numeric field.

Current SingleCaseES calculation guidance can support independent reproduction of the named method. That guidance neither selects the goal nor validates a clinical decision. The BACB Ethics Codes page identifies current professional obligations for certificants but does not approve a calculator, goal, or treatment conclusion.

Protect the data and preserve each version

Use fictional or properly de-identified values for education and testing. In an authorized clinical workflow, limit identifiable information, use approved storage, grant role-appropriate access, and preserve source and correction history. HHS provides separate summaries of the HIPAA Privacy Rule and HIPAA Security Rule. Those federal overviews do not determine whether a particular organization is regulated, certify the worksheet, or replace legal and security analysis.

Finalize each version with the raw-data and graph references, phase boundary, outcome rule, goal source, both distance calculations, full-precision outputs, conflict status, client and stakeholder input, analyst, qualified reviewer, decision owner, unresolved limitations, and next review date.

Related resources

Sources