Investigations into community diversity can be affected greatly by the scale of the analysis. On Pennsylvanian carbonate surfaces of Kansas (Plattsburg and Oread Fms.) and Texas (Winchell Fm.), two scales of community diversity were analyzed: 1) fine-scale, generic population abundances sampled from one-meter-grid point-counts, and 2) broad-scale, relative guild proportions in lithologic units. At the broader scale, 70% of community pattern could be explained by the presence or absence and types of bafflers, such as phylloid algae or frondose fenestellid bryozoans (Axes 1 and 2 of Bray-Curtis Ordination). Lithologic units dominated by bryozoan bafflers displayed highest guild diversity and evenness in communities, followed by lack of bafflers, and lastly, least diverse in phylloid-algae dominated communities. Lithology (grainsize, percent siliciclastics) and region (Texas vs. Kansas) appeared to have less influence on guild diversity. At the fine scale of individual samples, only 37% of taxon variation between samples could be explained by baffler presence or identity. Small scale variation in generic diversity and community composition may have been more greatly affected by taphonomic or biotic and environmental parameters, such as predation and competition intensity, flow energy, substrate, circulation, depth, or simply lateral variation. Analyses using taxon abundance alone may not explain all processes controlling patterns at local and regional scales. The apparent greater explanatory power of the analysis when using more inclusive categories could be a function of "noise reduction", but may equally be an oversimplification of community processes.