The interaction between target numerosity and visual cue size approached significance [ F (4,112) = 2.39, p = 0.055] implicating that the estimation bias induced by convex hull differed in size between target numerosities.
Post hoc paired samples T-tests showed that this bias was significant for numerosity 12 [ t (1,28) = −16,89, p <0.001], numerosity 28 [ t (1,28) = −2.77, p = 0.01], numerosity 36 [ t (1,28) = −2.71, p = 0.01] but marginally significant for numerosity 44 [ t (1,28) = −1,99, p = 0.056] and not significant for numerosity 20 [ t (1,28) = 0.2, p = 0.84].