Investigation of Two Preliminary Analysis-Altering Elements: Ordinate Scaling and DPPXYR

Area of Science:

• Behavior Analysis
• Research Methodology
• Data Visualization

Background:

• Accurate interpretation of graphed data is crucial in behavior analysis.
• Graph characteristics, such as ordinate scaling and data point display (DPPXYR), may influence visual analysis.
• Previous research has not consistently examined the impact of these specific graph manipulations on analyst judgments.

Purpose of the Study:

• To conceptually replicate findings on the impact of ordinate truncation and DPPXYR on visual analysis.
• To investigate whether manipulating graph characteristics affects analysts' confidence in identifying functional relations.
• To determine if these manipulations influence the estimated magnitude of treatment effects.

Main Methods:

• 27 visual analysts evaluated 48 ABAB graphs from datasets with null, small, moderate, and large effects.
• Graphs were manipulated by altering ordinate scaling and DPPXYR.
• Two mixed-effects models analyzed the impact of graph characteristics on confidence and effect magnitude estimation.

Main Results:

• Graph manipulations did not consistently impact analysts' confidence in determining a functional relation.
• Mixed findings were observed regarding the effect of graph manipulations on the estimated magnitude of treatment effects.
• The study systematically varied graph presentation to assess its influence on data interpretation.

Conclusions:

• Visual analysis confidence appears robust to common graph manipulations.
• The estimated magnitude of treatment effects may be susceptible to specific graph presentation choices.
• Recommendations for standardized graph construction and future research directions will be provided.