Exploring the impact of visual representations of variation on informal statistical inference

This experiment investigated how different methods of data visualisation can influence informal inference. Two hundred and five student participants with an introductory statistics background completed an online randomised survey. Participants were presented with a scenario that required them to make inferences about the difference between the life of two battery brands, based on a plot visualising the data. The plot type and the dataset visualised were manipulated based on the degree to which a plot depicted variability in the data (bar, box and dot) and whether or not the plot depicted data sampled from populations where each brands' battery life was equal or unequal. Both manipulations were found to significantly impact the informal inferences drawn by participants. The experiment also discovered an interaction between plot type and brands' battery life equality. Plots that depicted variability (box and dot) led to more correct inferences when the plots depicted data sampled from unequal populations, but fewer correct inferences when they displayed data sampled from identical populations. The findings build upon previous research by demonstrating that different methods of data visualisation can affect informal inference. These findings have important implications for statistics education and data visualisation.