Visual extrapolation of linear and nonlinear trends: Does the knowledge of underlying trend type affect accuracy and response bias?

Abstract

The purpose of these experiments was to examine the ability of experienced and inexperienced participants to predict future curve points on time-series graphs. To model real-world data, the graphed data represented different underlying trends and included different sample sizes and levels of variability. Six trends (increasing and decreasing linear, exponential, asymptotic) were presented on four graph types (histogram, line graph, scatterplot, suspended bar graph). The overall goal was to determine which types of graphs lead to better extrapolation accuracy. Participants viewed graphs on a computer screen and extrapolated the next data point in the series. Results indicated higher accuracy when variability was low and sample size was high. Extrapolation accuracy was higher for asymptotic and linear trends presented on scatterplots and histograms. Interestingly, although inexperienced participants made expected underestimation errors, participants who were aware of the types of trends they would be presented with made overestimation errors. © Springer Science+Business Media B.V. 2008.