Visual Analysis of ABA Data

Visual Analysis of ABA Data

“Behaviour Analysts employ a systematic form of examination known as visual analysis to interpret graphically displayed data." - Cooper, Heron and Heward (2007, p.149)

Recording Data for Analysis

When running Applied Behaviour Analysis interventions, data is continuously collected on the targeted behaviours because it allows those running the programmes to identify whether interventions are working or not.

This data could be the percentage of correct spellings in a test, or the amount of requests made by a student to take a break from work, or the amount of time a student spends out of his seat in a classroom.

The types of data that can be collected are numerous but the purpose of collecting data is to allow those running the interventions to ‘maintain direct and continuous contact with the behaviour under investigation’ (Cooper, et al. 2007, p. 127).

One of the primary ways this “contact” with the data is maintained is through graphs. There are a number of different types of graphs used within ABA but we have selected only to discuss the line graph as it ‘is the most common graphic format for displaying data in ABA’ (Cooper, et al. 2007, p.129).

When the data is plotted, there are 3 properties that are used to identify what is “going on” with the data; these are the variability, the level and the trend.

Variability

The variability of data relates to how different or “spread out” the scores are from each other. Take the two graphs below that show hypothetical data of the percentage of correct scores on a spelling test over 10 days for 2 children, Jane and Matt.

The plotted data for Jane shows her percentage of correct spellings remains stable around 80%. The data for Matt’s percentage of correct spelling changes or “varies” greatly over the 10 days and does not remain stable at all.

When interpreting the variability of Jane and Matt’s data, you would say that Jane’s responding is "stable" while Matt’s would be regarded as "variable" or even "extremely variable".

In general, if the data has high variability (e.g. Matt’s spellings) it suggests that the teachers do not have control over the teaching method and the tactics being used might need to be changed.

Level

The level of the data relates to the “position” of the data set taken from the Y-axis. Look at the graphs below; in the first graph if the plotted data points fell into the top section they would have a “high level”, if they fell into the middle section they would have a “moderate level” and if they were in the bottom section they would have a “low level”.

You could potentially separate the data levels further into “low-to-moderate” or “moderate-to-high” as in the second graph below.

Looking at the data plotted below in relation to the Y-axis, the level in the phase 1 data set is high, phase 2 is moderate and phase 3 is low. A mean or median line for the data might be used to better visualise the level of the data – this may help more when the data is somewhat variable.

Trend

The trend in the data is the “direction” it is going. For example, in the graph below, the first data set shows an “increasing trend” as the data points are “going up”. The second data set show a “decreasing trend” as the data points are “going down”. Finally, the third data set shows a “zero trend” because the data are not going up or down.

Why Use Graphs?

Graphs make it much easier to interpret and understand the data because they present the information in a visual format. For example , what can you draw from these set of numbers that recorded the percentage of correct spellings by a student over 14 school days:

45%, 46%, 52%, 48%, 58%, 61%, 64%, 75%, 70%, 78%, 75%, 80%, 84%, 90%

You were probably able to tell that the percentages showed an increase – but you had to read each one as you went and refer them back to each other as you went.

Now look at the graph below that depicts the same percentages. You don’t even need to take in the percentage values to immediately recognise that there was a gradual increase in percentage correct, and this is one major reasons why graphical displays of data are so useful.

That is not to say the percentages are unimportant but simply that graphing data can increase the speed at which analyses and interpretations can be made.

This is especially true when taking into account that a child might have a number of different interventions in place within an ABA programme and each one needs to be continuously analysed. Imagine reading 30 sets of numbers compared to the ease of being able to just look at 30 graphs…we know which we’d prefer.

References

• Cooper, J., Heron, T., & Heward, W. (2007). Applied Behaviour Analysis. New Jersey: Pearson Education.
• AutismSpeaks
• Appliedbehavioranalysisedu.org