Trend dominance

People prefer for things to get better over time when evaluating series of outcomes presented in graphs, even at the expense of substantial overall welfare. We refer to this empirical regularity as trend dominance, and demonstrate it across domains including economic growth, environmental outcomes, and the COVID vaccine rollout. We apply a succinct measurement instrument to empirically calibrate trend dominance, characterizing individual-level variation in how much total welfare individuals sacrifice in exchange for increasing trends. Across several experiments conducted on a NORC probability sample as well as convenience samples, we present evidence that trend dominance does not reflect genuine preferences. Trend dominance is, at least in part, a product of respondents struggling to identify the total benefits presented in a sequence and assuming trends continue beyond the plotted sequence. Media organizations, policymakers, and public health authorities routinely use sequence plots to illustrate forecasts and projections, but people’s evaluations of these charts often do not reflect preferences. If people internalize these (ill-considered) evaluations, or if the evaluations bias their behavior, the significance of these distortions extends to affecting behavior directly. Designers of graphs should take into account the biases people bring to visual presentations of over-time data.