When to use it
Use ABC recording during the descriptive assessment part of a functional behavior assessment (FBA). It captures what happened just before a behavior (antecedent), the behavior itself, and what happened just after it (consequence). Patterns in these records help you form a hypothesis about what maintains the behavior.
ABC data shows correlation, not cause. A consequence that often follows a behavior is not necessarily the one maintaining it. Use ABC data to form hypotheses. Where appropriate and safe, confirm them with a functional analysis conducted or supervised by a qualified behavior analyst.
Two common formats:
- Narrative (continuous) ABC: write a brief description of each A, B, and C in your own words. It is flexible and good early in an assessment, but it takes more time and depends heavily on the observer’s objectivity.
- Structured ABC: use pre-set codes (see the key below) so observers check options instead of writing. It is faster and easier to summarize, but you may miss events the codes don’t cover.
Many teams begin with narrative recording and move to structured codes once common antecedents and consequences are known.
How to fill it in
- Define the target behavior first. Write the topography, onset, offset, and exclusions at the top of the sheet. Record only behavior that fits the definition.
- Record the date, time, and activity for each entry. Setting and activity patterns, such as “always during math,” are often the most useful finding.
- Antecedent: write what was happening immediately before the behavior, in observable terms. Include who was present, what was said or asked, and what changed. Write “teacher said ‘Put the blocks away’”, not “client didn’t want to clean up.”
- Behavior: describe what the client did, how long it lasted, and how intense it was, in observable terms. Write “hit peer on arm twice with open hand”, not “got aggressive.”
- Consequence: write what happened immediately after, including what people said or did, and whether a demand was removed, an item was given, or attention changed. “Nothing changed” is also a consequence worth recording.
- Optional: note setting events such as poor sleep, illness, missed medication, or a schedule change. These can change how strong a consequence is as a reinforcer on a given day.
- Record as soon as possible. Entries written at the end of the day are much less accurate.
Common mistakes:
- Writing interpretations (“he was frustrated”, “she wanted attention”) instead of observable events.
- Recording only the most dramatic episodes, which skews the pattern.
- Leaving the consequence blank when “no one responded.”
- Observing only one setting or one staff member, so the pattern reflects that context rather than the client.
The sheet
Client:Observer:Setting:
Target behavior and operational definition:
| Date | Time | Activity / setting | Antecedent (what happened just before) | Behavior (what the client did) | Consequence (what happened just after) | Duration / intensity | Initials |
|---|---|---|---|---|---|---|---|
Optional structured code key. Write the code in the cell instead of, or in addition to, a narrative note. Adjust the codes to fit the client.
| Antecedent codes | Consequence codes |
|---|---|
| DM: demand or instruction given | AT: attention given (reprimand, comfort, talking) |
| DN: told “no” or access to item denied | TG: item or activity given |
| RM: preferred item or activity removed | ES: demand removed, delayed, or reduced |
| AD: adult attention diverted to someone else | RD: redirected to task |
| TR: transition between activities | NC: no change or no response |
| AL: alone or no activity available | BL: blocked or physically redirected |
| OT: other (describe) | OT: other (describe) |
Summarizing the data
After you have enough entries across relevant settings (many clinicians aim for at least 10 to 15 episodes, but there is no universal number), tally how often each antecedent and each consequence appears.
Percentage of episodes with a given antecedent or consequence = number of episodes with that code divided by total episodes, times 100.
Worked example: 20 ABC entries for screaming.
- 13 of 20 entries follow a demand (DM): 13 / 20 = 65%
- 12 of 20 entries end with the demand removed or delayed (ES): 60%
- 4 of 20 entries end with attention (AT): 20%
This pattern is consistent with an escape hypothesis. It does not confirm it. Next, compare with a conditional-probability view: how often was a demand followed by screaming, compared with demands that were not followed by screaming? A scatterplot or an experimental analysis can then refine the hypothesis.
Tips
- Write consequences exactly as they happened, even if the response didn’t follow the plan. Accurate records help more than tidy ones.
- Collect ABC data across the times, people, and settings where the behavior does and does not occur. Knowing where it doesn’t happen is informative too.
- Train observers with a few practice entries and compare their records. If two people describe the same episode very differently, tighten the definitions or switch to structured codes.
- If the behavior is dangerous, such as severe self-injury, aggression, or elopement, the team should have a written safety plan in place before and during data collection. Do not change safety responses just to see what happens.
- For families, a simplified three-column version (before, behavior, after) is often easier to keep up with than the full sheet.