g2, se2, sk2, t2, cf3, and c3. Version two tests consisted of g3, se3, sk3, t3, cf4, c4, g4,

se4, sk4, t4, cf5, c5, g5, and se5. Version three tests consisted of sk5, t5, cf6, c6, g6, se6,

sk6, t6, cf7, c7, g7, se7, sk7, and t7. Models were presented to respondents in the

questionnaires in the orders above.

 To randomize the questionnaires, first they were put in to piles based on test type

and version. Second, the test one versions were pulled from their piles in a random order

and placed into one combined pile for test one. The same was done for test two. This

way, respondents were less likely to have the version, thus the same set of models, as

someone sitting next to them was.