The stories groups tell: campaign finance reform and the narrative networks of cultural cognition

For the hierarchy--egalitarianism dimension, agreement with egalitarian affiliated statements received a − 1, while disagreement received a + 1; similarly, for the same dimension, hierarchy affiliated statements received a + 1 for agreement and a − 1 for disagreement. The scores were then aggregated producing a single hierarchy score for each potential participant ranging from − 6 to + 6, where a strongly negative score denoted an egalitarian orientation, while a strongly positive score denoted a hierarchical orientation. The same process was conducted for the individualism--communitarianism dimension. Agreement with individualism affiliated statements received a − 1, while disagreement received a + 1; similarly, communitarian affiliated statements received a + 1 for agreement and a − 1 for disagreement. The scores were then aggregated producing a single summative communitarian score for each potential participant ranging from − 6 to + 6, where a strongly negative score denoted an individualist orientation, while a strongly positive score denoted a communitarian orientation. To test H3, the CC scores were weighted by respondents’ response to the screening question (Mildly Agree = x 1.0; Agree = x 1.25; and Strongly Agree = x 1.50). The idea being that if the issue is seen as more important to an individual, the more likely they are to have developed at least a cursory impression and are able to convey this to the group.

The top quartiles had scores ranging from + 4 to + 6, while the bottom quartiles had scores ranging from − 4 to − 6 for each of the two dimensions. While 10 participants accepted the invitation to participate, actual attendance ranged from 8 to 10 individuals.

These data were collected in conjunction with the Cultural Cognition Project at Yale University and the Edmond J. Safra Center for Ethics at Harvard University. We thank the Edmond J. Safra Center for Ethics for their generous grant which made collecting these data possible.

Note that such an approach contrasts the more conventional way to conduct focus groups where efforts might be made to moderate or divert a few dominant focus group participants (e.g. Krueger and Casey 2000; Morgan 1996, 1997) from dominating the conversation. Indeed, such forceful personalities were allowed to thrive in our focus group environment, when they emerged, as they were likely to be powerful cultural foils for the other focus group participants.

Scores higher than .80 are generally considered acceptable (Lombard et al. 2002).

To address the possibility that a few strong personalities exerted disproportionate influence on the composition of the group narrative Spearman’s Rho was utilized to correlate the CC score of each individual with the percentage of narrative elements that individual contributed during the discussion. CC scores computed for 37 participants ranged from 10 to 18, with a median score of 15. The individual contribution was computed by dividing the total number of coded narrative elements the individual used divided by the total number of narrative elements identified in the focus group (pre- and post- calculated separately). The correlation coefficient for the pre-information focus group conversation was .112, with a 1-tailed significance of .254 (p < .05). The coefficient for the post-information portion was .089 with a 1-tailed significance of .299 (p < .05). Therefore an individual’s cultural cognitive score and desire for reform did not result in increased contribution of elements to the group narrative.

Some tangential observations are worth mentioning. No focus group discussed the topic of campaign finance in specific or technical policy terms. In fact, there was widespread confusion towards campaign finance policy in general. However, group members did speak passionately about their feelings towards elections, towards politicians and others, and used examples often to illustrate their points. Members made mention of current events, politicians and sports players to describe their feelings toward more general topics.

When using the Kruskal–Wallis H, the non-parametric equivalent of ANOVA, the initial result will simply indicate whether there is a statistically significant difference among the several categories tested. If a significant difference is discovered, further analysis of the differences between each of the groups using a Mann–Whitney U test is recommended. However, as the original significance value is p < 0.05, then this p value needs to be divided by the number of group comparisons to establish the 0.01 value for the results of the Mann–Whitney U test, this is known as the Bonferroni correction.