On making forecasts from binary sequences: Uncovering implicit cues

The purpose of this article is to examine the statistical characteristics of binary sequences with the aim of uncovering the implicit cues that people use when making forecasts of what comes next. Information theory was used to quantify these statistical characteristics. In 2 experiments people were presented with 100 intact sequences of 20 Xs and Os and simply asked to forecast whether the 21st event in each sequence will be an X or an O. Multilevel logistic regression models were used to estimate the odds associated with these forecasts under different experimental manipulations. In a third experiment people judged the forecastability of sequences in a paired-comparison task. The results from the first 2 experiments showed that third-order redundancy (i.e., information provided by knowledge of the preceding pairs of events) was the most salient cue influencing forecasts. Experiment 3 showed that judgments of forecastability were based on this cue as well. When examining intact sequences with the goal of forecasting what comes next, people are more sensitive to higher-order transitional probabilities than has been previously suggested. (PsycInfo Database Record (c) 2021 APA, all rights reserved).