Data display & visual analysis Glossary

Choose a display that matches the question

A line graph connects repeated measurements in order and is common for change over time. A bar graph compares values across categories or discrete summaries. A cumulative record plots an accumulating total, so the line stays flat or rises.

An equal-interval graph uses equal physical distances for equal numerical differences. A semilogarithmic chart uses a ratio-scaled axis so equal ratios occupy equal distances. The standard celeration chart is a standardized semilogarithmic display used within precision teaching.

Evans and colleagues' precision-teaching review describes the system and charting context. A graph type should be selected for the measurement and decision, rather than for the visual effect it creates.

Preserve graph structure

A data path connects successive points that belong together. A condition label names the condition represented in a phase. A phase-change line marks when a planned condition changes.

A scale break signals a discontinuity in the axis. It should be visible and used carefully because it can change apparent distance. Keep units, axis intervals, dates, missing sessions, and condition changes accurate.

The What Works Clearinghouse single-case documentation provides design and evidence guidance for single-case research. It is a research standard, not a universal clinical graph template.

Analyze pattern within a condition

Within-condition analysis examines pattern inside one phase. Level describes the magnitude or position of data. Variability describes how much values fluctuate.

Trend is the direction and rate of change over time. Celeration expresses change in frequency per unit of time, often as a multiplication or division factor on a standard celeration chart.

Inspect the raw points, central pattern, slope, range, outliers, session spacing, and missingness. A mean can hide increasing spread or two different subpatterns. A smooth line can hide that opportunities changed.

Compare conditions with several dimensions

Between-condition analysis compares patterns across phases or conditions. Immediacy of effect examines how quickly responding changes around a condition change.

Data overlap describes the extent to which values from different conditions share a range. Consistency across similar phases asks whether comparable conditions show similar patterns when repeated.

Visual analysis integrates level, trend, variability, immediacy, overlap, consistency, and design context. Wolfe and colleagues' protocol review and Lobo and colleagues' single-case analysis paper discuss systematic visual-analysis methods and quality considerations.

No one dimension decides the result. Low overlap with unstable baseline and simultaneous implementation changes may remain ambiguous. Strong immediacy in one comparison may need replication.

Use an example with the actual denominator

Fictional clinician Harper graphs independent schedule checks per eligible work period for a young adult internship goal. Baseline values across five periods are 1, 2, 1, 2, and 1. After a visual checklist is introduced, the next five are 3, 4, 4, 5, and 4.

The level rises with little overlap and an immediate first-point change. Several alternatives remain: practice, supervisor prompts, easier shifts, or changed opportunity quality. The graph should also show eligible periods, missed periods, checklist availability, and prompt use.

Harper asks the person whether the checklist helps and samples use with another supervisor before changing the plan. The graph supports a decision discussion; it does not prove the checklist caused every change.

Make data-based decisions traceable

Data-based decision making uses defined evidence and decision rules to choose, continue, adapt, or stop action. Record the question, graph version, data window, exclusions, interpretation, client input, decision owner, action, and next review.

The BACB BCBA Test Content Outline includes graphing, visual analysis, measurement, experimental design, and data-based decisions as examination content. It does not validate a graph, causal claim, or plan.

Use accessible colors, shapes, labels, alt text, and tables. Preserve the underlying numbers so readers can verify the display and use another format.

Audit the display before interpreting it

Trace several plotted points back to the source record. Confirm dates, units, opportunity counts, denominators, missing sessions, corrections, and condition assignment. Check that phase lines occur at the actual intervention change and that labels identify every material component.

Review axis ranges and intervals. A compressed range can flatten meaningful variation, while a truncated range can exaggerate small differences. A scale break should be conspicuous. Connecting points across long gaps can suggest continuity that was never observed.

When opportunities vary, graph a rate, percentage with its denominator, or another suitable measure and preserve the raw counts. A line at 80% can mean 4 of 5 or 40 of 50, which carry different precision and exposure.

Have trained reviewers independently analyze selected graphs when conclusions drive high-risk or major plan changes. Record agreement and disagreement by dimension. Consensus after discussion should preserve the initial ratings and rationale rather than erasing uncertainty.

Separate graph integrity, visual-analysis agreement, clinical importance, social validity, and causal confidence. A correctly drawn graph can display weak data, and reviewers can agree on a pattern that has little value to the person.

Explore clinical roles at Finni practices and ask how teams maintain data quality, accessible displays, systematic review, client input, and decision history.