Balanced scales make agreement rates independent of trait strength in the sample

Evidence

YYea-saying is hard to measure on unbalanced scales such as the F Scale because you cannot tell if the percentage of “agrees” versus “disagrees” in the answers merely reflects the strength of the trait in that sample. But speaking just in terms of content, agreement should basically equal disagreement on properly balanced scales, whether the sample is high, middling, or low on the trait. — Bob Altemeyer, The Authoritarian Specter, 802