Cohen's convention puts a "large" association at w = 0.5, and Cramer's V is w scaled by sqrt(min(levels) - 1), so the same large association is a smaller V in a bigger table: 0.50 for two batches or two groups, 0.35 for three, 0.29 for four (Cohen 1988). A fixed V of 0.6 admitted a large association in every table.