BA420 Unit 5 Visual Analysis Paper Example

Reading Sofia's Graph With a Protocol Instead of a Hunch: A BA420 Unit 5 Visual Analysis Paper

[Student Name]

Purdue University Global

BA420: Methods of Behavior Analysis

Unit 5 Visual Analysis Paper

[Instructor Name]

[Date]

The writer, the school, the student and every data point are composite and illustrative; the article and books cited are real.

The Data

Graphs are where my data end up, and this unit taught me to read them more carefully. Sofia, a seventh grader, completes her work well once she starts but takes a long time to begin. Her plan measured latency, the minutes from the teacher's instruction to her first written mark, during independent work in English. The district behavior analyst designed a withdrawal design to test a simple intervention: a laminated checklist taped to Sofia's desk listing three steps, write your name, read the first question, write one word.

The phases ran as follows. In baseline, eight sessions, Sofia's latency ranged from 3.0 to 5.5 minutes, averaging 4.2, with a slight upward trend in the last three sessions. In the first checklist phase, eight sessions, latency dropped to 2.8 minutes in the first session and then to between 0.6 and 1.5 minutes, averaging 1.1. In withdrawal, six sessions, latency rose to between 2.9 and 4.4 minutes, averaging 3.6. When the checklist returned, eight sessions, latency fell to between 0.5 and 1.4 minutes, averaging 0.9.

Why a Protocol

My first reading of the graph, done in a few seconds, was simple: latency went down with the checklist, up without it and down again. That impression was correct in general, but Wolfe et al. (2019) argue that visual analysis is more reliable when analysts use a systematic protocol, because unstructured judgments vary between analysts and can overlook important features of the data. Their protocols guide the analyst through within-phase and between-phase characteristics and ask, for each comparison, whether the data show a basic effect.

Within Each Phase

Level, the average or typical value, was high in both baseline and withdrawal and low in both checklist phases. Trend, the direction of change, was flat to slightly increasing in baseline, slightly decreasing in the first checklist phase, increasing in the first few withdrawal sessions and flat in the second checklist phase. Variability, the spread of data around the level, was moderate in baseline and withdrawal and low in both checklist phases. Cooper et al. (2020) emphasize that stable baselines are needed before a phase change, because changes against a variable or trending baseline are harder to interpret. Sofia's slight upward baseline trend worked against the expected direction of change, which strengthens confidence when latency then dropped.

Between Phases

Immediacy asks how quickly the data change after a phase change, usually by comparing the last three points of one phase with the first three of the next. Overlap asks what proportion of data in one phase falls within the range of the adjacent phase. Consistency asks whether data in similar phases look alike.

The protocol made me look at the first checklist session, which my eye had skipped over because the overall pattern looked so clear.

For the first phase change, from baseline to checklist, immediacy was moderate rather than strong: the first checklist session, at 2.8 minutes, fell within baseline's range, and the large drop came in the second session. Overlap was one point of eight. For the second change, from checklist to withdrawal, immediacy was strong: latency rose above two minutes in the first withdrawal session and stayed high, with no overlap. For the third change, from withdrawal to checklist, immediacy was strong again, with no overlap. Consistency was high: the two baseline-like phases resembled each other, and so did the two checklist phases.

Demonstrations of Effect

A withdrawal design with four phases provides three opportunities to show an effect. The protocol counted three demonstrations: a reduction in latency with the checklist, an increase when it was withdrawn and a reduction when it returned. The first demonstration was somewhat weaker because of the delayed change. Many standards for single-case research call for at least three demonstrations of effect at different points in time to support a claim of experimental control. Sofia's graph meets that standard, though the delayed first change is worth noting.

Building the Graph

Visual analysis depends on a well-built graph. Following standard conventions, Sofia's graph shows sessions on the horizontal axis and latency in minutes on the vertical axis, with zero at the bottom of the vertical axis, so changes are not exaggerated. Each phase is separated by a vertical dashed line and labeled at the top, and data points are not connected across phase lines, so the eye does not read a continuous trend through a condition change. The scale stays the same across the whole graph. Kazdin (2011) notes that visual inspection is most persuasive when changes are large, immediate and consistent across replications, and least persuasive when baselines are unstable or when effects are small relative to the variability in the data. A misleading graph can make a weak effect look strong or hide a real one, so the team checks each graph against these conventions before anyone interprets it.

Agreement Between Analysts

To test the protocol's value, the district behavior analyst and I analyzed the graph separately, each using the same protocol, and then compared our ratings. We agreed on every within-phase characteristic and on the number of demonstrations. We disagreed only on whether the first phase change showed a moderate or a weak immediate effect. Without the protocol, our unstructured judgments a week earlier had differed on whether the first change mattered at all.

Why the First Change Was Slower

The protocol's flag on the first phase change prompted a question. Sofia's teacher remembered that on the first checklist day, Sofia had asked what the checklist was for and spent part of the work time reading it. By the second day she used it without discussion. A novel stimulus, not a failure of the intervention, likely explained the slower start. That kind of explanation needs to be noted, not used to dismiss the data, and it suggests introducing a new checklist with a brief explanation the day before it is used.

Conclusion

A systematic protocol turned a quick impression of Sofia's graph into an explicit analysis. Level, trend and variability within phases, and immediacy, overlap and consistency between them, supported three demonstrations of effect, with a delayed first change the protocol caught and the teacher's observation explained. Wolfe and colleagues' argument for structured visual analysis proved itself in a single student's graph.

References

Cooper, J. O., Heron, T. E., & Heward, W. L. (2020). Applied behavior analysis (3rd ed.). Pearson.

Kazdin, A. E. (2011). Single-case research designs: Methods for clinical and applied settings (2nd ed.). Oxford University Press.

Wolfe, K., Barton, E. E., & Meadan, H. (2019). Systematic protocols for the visual analysis of single-case research data. Behavior Analysis in Practice, 12(2), 491-502. https://doi.org/10.1007/s40617-019-00336-7