A cognitive consistency theory of attitude change, based on balance theory but formulated as a quantitative model, according to which if a source S and attitude object O are linked by an associative assertion (such as S is an O, S likes O, S shakes hands with O, or S goes with O) or a dissociative assertion (such as S is not an O, S dislikes O, S avoids O, or S criticizes O), then the attitudes to both S and O of the person receiving this message will shift towards a point of equilibrium, the less polarized of the two elements moving proportionately more than the more polarized one. Attitudes towards S and O are measured on a seven-point semantic differential scale from −3 to +3, and the change equation for the source is Change(S) = [| O | /(| O | + | S |)]P, where | O | and | S | are the absolute values (ignoring any negative signs) of the two initial attitudes and P is the pressure towards change—the total amount of change required to produce congruity. Similarly, the change equation for the attitude object is Change(O) = [| S | /(| S | + | O |)]P. For example, if a person has a positive attitude towards the composer Tchaikovsky (+2) and a very negative attitude towards homosexuality (−3), and if the person discovers that Tchaikovsky was gay, then the distance between the two and hence the value of P is 5, and Change(S) = [3/(3 + 2)]5 = 3, which means that the attitude towards Tchaikovsky moves three units in the direction of the attitude towards homosexuality; and by a similar calculation, the attitude towards homosexuality moves two units in the direction of the attitude towards Tchaikovsky; so that the person's attitudes to Tchaikovsky and homosexuality both end up at −1. For a dissociative assertion, the two move away from each other. In the light of empirical findings two correction factors were added to the theory: an assertion constant to account for the fact that the attitude towards the attitude object tends to change more than the attitude towards the source, and a correction for incredulity that reduces the amount of change in both S and O in proportion to their distance apart (for an associative assertion) or their closeness (for a dissociative assertion). The theory was formulated in 1955 by the US psychologist Charles E(gerton) Osgood (1916–91) and the Canadian-born US psychologist Percy (Hyman) Tannenbaum (born 1927).