The 'qualitative' pole of (Q-squared) combined methods has been defined in mainly residual ways as 'non-numerical' or 'noneconomics'. There is need, instead, for a critical social theorization of qualitative methods. Evidence from a development program for chronically poor, tribal, northern Orissa is used to examine the communicative action of 'participatory assessment' (PA). PA assumes that 'group' and 'visual' synergies can challenge the power relations that restrict communication and poor people's emancipation. The authors' ethnographies show that participants sequestered information from PA village seminars. Although well trained, the PA organizers increasingly ignored cultural context and substituted universalized techniques that produced only quantities of noncontextualized attitudes. The core PA routines therefore gave misleading results; they mistakenly replaced substantive accounts of communication in relation to lifeworlds with abstract seminar techniques. To obtain more reliable results, methods of 'embedded' economic anthropology were used instead to assess poverty.