The impact of task properties feedback on time series judgmental forecasting tasks

This study evaluates the impact of task properties feedback on the time series forecast accuracy of four different judgmental forecasting processes. Specifically, we test the impact of providing information on time series data patterns and degree of noise level to knowledgeable subjects to interpret this information. Ninety-eight subjects were used as the source of the individual and three-person group forecasts for eight artificial time series with varying patterns and noise levels. Our findings show that such task properties feedback leads to improvements in forecast accuracy for all forecasting processes tested, particularly for high noise series. This is true for both individual and group judgmental forecasting processes, as well as combination forecasts. These findings have important implications for business practitioners who continue to rely on judgmental forecasting processes. The information provided to subjects in our study is such that it could readily be obtained as output from most statistical software packages. Our findings imply that all judgmental forecasting processes could benefit by relying on this type of cognitive aid as an input to their judgments.