A plethora of research over the past two decades has demonstrated that citizens in countries around the world dramatically overestimate the size of minority demographic groups and underestimate the size of majority groups. Researchers have concluded that this misestimation is a result of characteristics of the group being estimated, such as level of threat the group poses and the amount of exposure someone has with to the group. However, explanations of this misestimation have largely ignored theoretical models of perception and measurement, such as those developed in classic psychophysics. This has led to interpretations that are at variance with modern theories of measurement. We present a model which combines an understanding of the nature of human estimations with a conceptualization of uncertainty, which extends to accommodate bias. We apply this model to three large-scale datasets collected by the Ipsos MORI research group. Model fits from our approach suggest that to a considerable degree, the errors people make are due to uncertainty rather than bias. These biases are quite different in character from those that other groups have reported. Many of the present biases, furthermore, are shared widely across different countries.