The big idea
Behavior analysts make most of their decisions by looking at graphs, so a graph is a working tool, not decoration. A good graph makes the relation the reader needs to see obvious: whether behavior changed when the condition changed, how fast responses are piling up, or when in the day problems cluster.
Different questions call for different graphs. The equal-interval line graph is the default for showing behavior across time. Bar graphs summarize and compare. Cumulative records show rate as slope. Scatterplots show patterns. Semilogarithmic charts show proportional change.
Breaking it down
Equal-interval line graph
On an equal-interval line graph, the same distance along an axis always stands for the same amount. Its parts:
- Horizontal axis (x-axis or abscissa): time, usually sessions or days.
- Vertical axis (y-axis or ordinate): the measure of behavior, labeled with its unit, such as responses per minute or percentage of intervals. It usually starts at zero.
- Phase change lines: vertical lines marking where one condition ends and the next begins. Dashed lines often mark smaller changes within a condition.
- Condition labels: short, descriptive names centered above each phase, such as Baseline or Token system.
- Data points and data paths: points joined by lines within a condition. Do not connect points across a phase change line, and by convention break the path where planned data are missing, such as during an absence.
- Legend and figure caption: needed when more than one data path appears.
Proportion matters too. The vertical axis is usually drawn shorter than the horizontal one (about two-thirds as long is a common guideline), so changes are neither flattened nor exaggerated.
Bar graph
A bar graph shows values for separate categories or conditions as bars. Use it to summarize: the mean rate in each condition of a functional analysis, problem behavior by setting, or each staff member’s integrity score. It cannot show trend or session-to-session variability, so it should not be the only display used to decide whether an intervention is working.
Cumulative record
On a cumulative record, each point adds the current session’s count to all the counts before it.
- The slope shows rate: the steeper the line, the faster the responding.
- A flat stretch means no responses.
- The line can never go down.
Daily counts of 3, 5, 0 and 4 are plotted as 3, 8, 8 and 12. The flat step from 8 to 8 is the day with no responses. Cumulative records came from the operant laboratory, and in applied work they suit totals that build over time, such as words mastered across a school year.
Scatterplot
A scatterplot shows how individual measures fall in relation to two variables. In research, it can show whether two measures go together, such as hours of sleep and aggression the next day. In practice, the term usually means a time-of-day grid (Touchette, MacDonald and Langer, 1985): rows are time blocks, columns are days, and each cell is marked for how much behavior occurred. Patterns across days suggest when to observe and what to assess (F.5).
Semilogarithmic charts
On a semilogarithmic chart, the vertical axis is logarithmic, so equal distances show equal ratios: going from 2 to 4 per minute looks the same as going from 20 to 40. The Standard Celeration Chart plots count per minute on such an axis across calendar days. The slope of the data is celeration (C.3), which makes it easy to compare proportional change across learners or behaviors that run at very different rates.
Seeing it in practice
| The question | Graph | Why |
|---|---|---|
| Did Dev’s requests increase after functional communication training began? | Line graph with a phase change line | Shows level, trend and variability session by session |
| What was the mean rate of problem behavior in each analysis condition? | Bar graph, alongside the line graph | Compares conditions at a glance |
| How many sight words has Lucía mastered this year, and is her pace steady? | Cumulative record | The total and the pace (slope) are both visible |
| At what times of day does Mr. Tremblay, 81, shout in his care home? | Scatterplot grid | Shows clusters by time of day across weeks |
| Is Ana’s reading fluency growing proportionally faster with a new practice routine? | Standard Celeration Chart | Equal ratios look equal, so growth rates can be compared |
Common mix-ups
- Reading a cumulative record like a line graph. A flat line means zero responding, not steady responding. Decreasing behavior shows as a flatter slope, never a falling line.
- Connecting data across phase change lines. It blurs the comparison between conditions.
- Using a bar graph for decisions over time. Means hide trend and bounce.
- Truncated or stretched axes. A vertical axis that starts high, or is drawn very tall, can make a small change look dramatic.
- Two meanings of scatterplot. It can be an x-y plot of two variables or a time-of-day grid; read the context.
- Reading log spacing as equal amounts. On a semilog chart, equal distances mean equal ratios.
Exam tip: For any cumulative record question, think slope. Steeper means faster, flat means no responses, and the line never drops.
Quick recap
- The equal-interval line graph shows behavior over time, with labeled axes, phase change lines, condition labels and data paths that break between conditions.
- Bar graphs summarize and compare categories but hide trend and variability.
- Cumulative records: slope equals rate, flat means none, and the line never decreases.
- Scatterplots show relations between two variables or patterns by time of day.
- The Standard Celeration Chart uses a log scale, so its slopes show proportional change.
Further reading
- Cooper, J. O., Heron, T. E., & Heward, W. L. (2020). Applied behavior analysis (3rd ed.), chapter 6.
- Touchette, P. E., MacDonald, R. F., & Langer, S. N. (1985). A scatter plot for identifying stimulus control of problem behavior. Journal of Applied Behavior Analysis, 18(4), 343-351.
Key terms
- Line graph (Equal-interval line graph, Equal-interval graph)
- A graph on which equal distances along each axis stand for equal amounts, with time (sessions or days) on the horizontal axis and a measure of behavior on the vertical axis. Data points are joined within each condition and separated by phase change lines. It is the most common graph in behavior analysis.
- Bar graph
- A graph that uses bars of different heights to show values for separate categories or conditions. It summarizes and compares well but hides trend and session-to-session variability, so it should not be the only display used for decisions over time.
- Cumulative record (Cumulative graph)
- A graph in which each data point adds the current count to all previous counts, so the line can rise or stay flat but never fall. The slope shows the rate of responding: steeper means faster, and a flat stretch means no responses.
- Scatterplot (Scatter plot)
- A graph showing how individual measures are distributed in relation to two variables. In applied practice the term often refers to a time-of-day grid recorded across days, which shows when a behavior tends to occur and helps guide observation and assessment.
- Standard Celeration Chart (SCC)
- A semilogarithmic chart that plots count per minute on a logarithmic vertical axis against calendar days, so equal proportional changes look equal. The slope of the data shows celeration. It is the main display used in precision teaching.
- Phase change line (Condition change line)
- A vertical line on a graph marking the point at which one condition ends and another begins. Data paths are not connected across it, and each phase is labeled above the data.
Check yourself
4 questions on C.10. Choose an answer to see whether you're right and why.
Show the answer
Answer: D. A cumulative record adds each session's count to the running total, so the line rises only when responses occur. A flat stretch means nothing was added: no correct responses. Steady responding would appear as a steady upward slope, and a cumulative record can never slope downward.
Show the answer
Answer: B. Each cumulative point is the running total: 4, then 4 + 6 = 10, then 10 + 0 = 10, then 10 + 5 = 15. Plotting the raw counts (4, 6, 0, 5) would make it an ordinary line graph. A running total can never fall back to 0 or 5.
Show the answer
Answer: A. A line graph with a phase change line shows each session, so the parent can see level, trend and variability before and after the plan. A bar graph of means hides session-to-session changes. A time-of-day scatterplot answers a different question, and an unlabeled cumulative record would not mark the condition change.
Show the answer
Answer: C. Data paths are broken at phase change lines so each condition's data can be judged on their own and then compared. Joining points across conditions blurs that comparison. The measure of behavior goes on the vertical axis and time on the horizontal, and condition labels go above each phase on the graph itself.
Adds a step to your study record. Saved in this browser only.