Persuasion header image
The Nature of Attitudes and Persuasion

The Yale Approach

Congruity Theory

Cognitive Dissonance Theory

Social Judgment/ Involvement Theory

Information Integration Theory

Theory of Reasoned Action

Elaboration Likelihood Model


Information Integration Theory

Information Integration Theory
Suggestions for Changing Attitudes
Information Integration Theory
This theory was developed, and extensively tested through a variety of experiments, by Norman Anderson (
1971, 1981a, 1981b, 1991; see also Fishbein, 1967). Information Integration theory explores how attitudes are formed and changed through the integration (mixing, combining) of new information with existing cognitions or thoughts.

Information integration theory considers the ideas in a persuasive message to be pieces of information, and each relevant piece of information has two qualities: value and weight. The value of a bit of information is its evaluation (favorable or unfavorable) and the weight is the information’s perceived importance. For example, Steve tells Sarah that Joe has a ponytail. The value of this information is whether Sarah thinks a ponytail (for Joe) is good (attractive) or bad (unattractive or inappropriate). The weight is how much that friend’s hair style matters to Sarah. If it does matter (has some weight) and if Sarah thinks it is good for Joe to wear a ponytail, then this piece of information inclines Sarah to have a favorable attitude toward this friend.

However, Sarah’s new attitude would also depend on what she thought about Joe before she learned about Joe’s new hair style. If she previously had a favorable attitude toward Joe, her attitude would remain favorable. It could be come even more favorable, especially if she thought hair style was very important (if this information had a larger weight) and if Sarah really, really liked pony tails (if the information had a high positive value). On the other hand, if Sarah used to have an unfavorable attitude toward Joe, this new information probably wouldn’t change her attitude from unfavorable to favorable. It could mean that her new attitude wasn’t as negative as before, especially if this new information had a large weight and a high positive value.

On the other hand, it is possible that Sarah doesn’t think men should wear ponytails. This would mean that the new information had a negative value. Again, Sarah’s new attitude would depend on three factors: her original attitude, the value of the new information to Sarah, and its weight. If she liked Joe before she learned about his ponytail, she might like him less (have a less favorable attitude). Her attitude is most likely to change if men’s hair style is important to her (has weight) and if she has a very unfavorable feeling about ponytails on men (value). If her initial attitude was unfavorable, finding out about Joe’s new hair style would have a tendency to make her new attitude even more unfavorable. If the weight of this new information was high and the value was very unfavorable, Sarah’s attitude could become noticeably more negative.

This, Information Integration Theory states that when we obtain new information (often from persuasive messages), those new pieces of information will affect our attitudes. They won’t replace our existing attitudes: If Sarah began with an unfavorable attitude toward Joe and she likes ponytails on men, she won’t all of a sudden have a strong positive attitude toward Joe. However, when we learn new positive information, negative attitudes tend to become less negative and attitudes that are positive are likely to become somewhat more positive.

Furthermore, Information Integration Theory tells us that each bit of information has two important qualities, weight and value. Both factors influence our attitudes. Information that is (1) high in value, highly favorable (or highly unfavorable), and (2) high in weight (is very important to us) will have more influence on our attitudes than information low in value or weight. Information with low value (slightly favorable or slightly unfavorable) and low weight will have the least influence on our attitudes.

Therefore, new information is mixed, combined, or integrated with existing information to create a new attitude. However, information can be combined in more than one way. One important question is whether new information is added to existing knowledge, or whether it is averaged into it. Consider this simple example. Bob has a pretty favorable attitude of +3 (on a scale of -5 to +5) toward a certain automobile. If he learns a new piece of information (say, it has chrome wheels) that is slightly favorable for him, say a +1, what will his new attitude be?  If he adds +1 and +3, then Bob’s new attitude will be more favorable than his existing attitude, a +4. On the other hand, if Bob averages the new and old information his new attitude should be less favorable, a +2 (1 plus 3 is 4, divided by 2 pieces of information, equals an average of 2).

Some people believe that the adding model is best. But what happens if one has several pieces of new information, all valued at +3 (again, on a scale of -5 to +5). If Bob is told four new pieces of information that he values at +3 each, his attitude would be +3 (his initial attitude) +3 +3 +3 +3, or +15. But if the attitude scale goes from -5 to +5, he can’t possibly have an attitude of over +5. And research shows that in situations like this one Bob’s final attitude wouldn’t even be +5.

If adding doesn’t work, does this mean that information is combined by averaging?  If he starts with a +3 and learns four new pieces of information, all valued at +3, averaging this information (+3, the initial attitude, added to +3 +3 +3 +3 and then divided by 5) would produce a final attitude of +3. But surely if Bob learns several new favorable pieces of information about this car his attitude would become somewhat more positive. And, again, the research shows that in these kinds of situations Bob’s final attitude would be higher than +3.

Many tests have been tried to decide this question but the evidence does not clearly support either adding or averaging models. In my opinion, this is true is because human beings aren’t computers or calculators. I certainly agree that people do combine new information and old to create new attitudes. However, I do not believe that people assign numbers to pieces of information or perform mathematical calculations (adding or averaging) to figure out their new attitudes. I think that formulas should be considered to be approximations of what human beings do without numbers. To make a formula work, we have to put numbers into it and combine those numbers in some way (adding or averaging them). These theories and formulas do come close to predicting our attitudes, so they are useful. But we shouldn’t be surprised if these formulas do not predict exact attitudes. I think it is enough that they can come close.

I kept these examples about Sarah and Bob simple. But many attitudes are complex and often we have both positive and negative ideas about people or cars. A attitude toward a car that is favorable overall may be made up from both favorable (affordable, sporty, nice color, anti-lock brakes, fast, handles well) and unfavorable (too little cargo room, no CD player, poor gas mileage). For the overall attitude to be favorable the positive ideas must be more numerous or have higher weight and value than the unfavorable ideas (or be all three: positive ideas are more numerous, have higher weight, and higher value than the negative ideas).

<Previous Page | Next Page>