Colonial organisms vary in the degree to which they are individuated at the colony level, i.e., in the degree to which the colony constitutes a unified whole, as opposed to a group of independent lower-level entities. Various arguments have been offered suggesting that evolutionary change along this continuum may be biased, that increases may be more probable than decreases. However, counterarguments can be devised, and the existing evidence is meager and inconclusive. In this paper, we demonstrate how the question can be addressed empirically by conducting a test for bias in a group of stenolaemate bryozoans, the cyclostomes. More specifically, we suggest three criteria for colony individuation: degree of connectedness among lower-level entities (in this case, zooids), degree of differentiation among lower-level entities, and number of intermediate-level parts. And we show how these criteria can be used together with a phylogeny and ancestral-state reconstruction methods to test for bias. In this case, results do not unambiguously support any single interpretation but are somewhat supportive of a null hypothesis of no bias in favor of increase.As part of the demonstration, we also show how results can be transformed into a quantitative estimate of an upper limit on bias. Finally, we place the question of bias in a larger context, arguing that the same criteria and methods we employ here can be used to test for bias in other colonial taxa, and also at other hierarchical levels, for example, in the transitions from free-living eukaryotic cells to multicellular organisms.