250 Chapter 6. Attitude Formation To summarize briefly, research on order effects indicates that when a stimulus of a set of adjectives and subjects evaluate the person person is described in terms after all adjectives have been presented, a primacy effect is usually obtained. Vari- ous explanations have been offered to account for this effect, including change in meaning, attention decrement, and discounting. Studies designed to test the ex planations have found that when subjects are induced to pay attention to all ad jectives in the list, the primacy effect tends to disappear and is sometimes re- placed by a recency effect. This finding, however, does not appear to provide clear evidence in favor of one explanation as opposed to another Expectancy-Value Analysis of order Effects From our point of view, the finding that order of presentation influences attitude mplies that when a given list of adjectives is presented in different orders, differ ent belief systems are formed. In other words, order of presentation may result in the formation of different beliefs, may influence belief strength, or may affect eval uations of the adjectives. Which of these processes is responsible f a given order effect can be determined only when measures of beliefs and attribute evaluations are available. One possible interpretation of order effects from the point of view n expectancy-value formulation is related to the recall of information pre sented. In most research in impression formation, subjects have no prior beliefs about the hypothetical stimulus person, and it seems reasonable to assume that the beliefs they come to hold are primarily determined by the information they can recall about the person. This does not mean that they will believe everything they recall or that the beliefs they do form will be held with equal strength. Moreover, subjects may form additional inferential beliefs about the stimulus person on the basis of these initial beliefs In our discussion of serial order learning in Chapter 5 we noted that items ap pearing at different positions in a list are not equally likely to be recalled. Now assume that for some reason subjects are better able to recall adjectives appearing early in the list. If the informational and inferential beliefs they actually form are based on the adjectives they can recall, a primacy effect will obtain. A similar argument can be made for a recency effect. When early and late adjectives are recalled equally well, neither primacy nor recency are expected. Research on serial order learning has shown that words at the beginning and end of a list tend to be better recalled than words in the middle. Further, there is usually a slight tendency ords at the beginning to be recalled better than words at the end, suggesting