You certainly have a valid point, and I am one of the many guilty folk.  But the answer to your question also depends on what exactly are the questions posed in the study, and what are the specific methods. In some cases, a simple log-transformation will solve the problem. In other cases, it may be a good idea to run both separate analyses for social and other insects, and combined ones, too see whether the overall  patterns are not simply driven by the former.

I guess if you only estimate pure abundance (biomass) you can easily get biased results. But if you have the option to specify taxa until a certain level you sure should include both, as diversity and species richness estimators are considered good bioindicators too. It also might be interesting to compare both, and maybe identify one as a better indicator for certain factors.

Thanks Serban and Julian for the discussion.

Let me make a point.

We must consider that use an organism as bioindicator knowing only his Order, or Genus without any clues about their real role in the soil ecosystem can be little useful. I figure it out after read some papers about woodlice, in special this: 

Zimmer, M., and W. Topp. "Relationships between woodlice (Isopoda: Oniscidea) and microbial density and activity in the field." Biology and Fertility of Soils 30.1-2 (1999): 117-123.

Another one (this with ants) that open my eyes: DEL TORO, Israel; RIBBONS, Relena R.; PELINI, Shannon L. The little things that run the world revisited: a review of ant-mediated ecosystem services and disservices (Hymenoptera: Formicidae). Myrmecological News, v. 17, p. 133-146, 2012.

>12000+ ants described and the vast majority with only the taxonomy "defined", no more that this. So if the Formicidae signal starts to blink on the dashboard the pilot will not know what is out of service, he will only be sure that a led is working on the board.