Welcome to our segment on attitude questions. Attitude questions are very popular in a lot of surveys, in particular consumer satisfaction surveys or the evaluative surveys work with attitude questions a lot. In this first segment, we'll talk about attitudes more generally and then we'll talk about context effects and scales that you can use to measure attitudes in later segments. Attitudinal questions are questions about things that are not verifiable by external observation or records, at least in theory they aren't. And, you know, it's important that you sort of distinguish attitude questions from other questions about subjective psychological states. So, an attitude has this evaluative component. If you read this example below, you'll see we asked people here how strongly they agree or disagree with a particular statement and the statement being here, "People in my neighborhood generally get along with each other." That's a little different than asking them about something that's knowledge related, a cognitive component, and more a question about a belief, like, "How healthy is pizza on the following dimensions?" Furthermore, it's different from behavioral intentions that have the strong behavioral component. For example, "In the next six months, how likely is it that you will buy a car?" The last two, the cognitive components and the behavioral components, beliefs and behavioral intentions, are a little easier to ask. Attitudes have a few more issues that we are going to talk about. So, the traditional view on attitude is that an attitude is an enduring positive or negative feeling about some person, object, or issue. Classic reference here is the work by Petty and Cacioppo from the 80s. In other places described as a psychological tendency that's expressed by evaluating a particular entity with some degree of favor or disfavor. Often, you know, there would be an inclusion of a direction, positive versus negative, or a feeling that is implied and measured unidimensionally. Now, implicit in these traditional views is that attitudes are pre-existing. So you can, you know, pull out a drawer in your head and you can retrieve an attitude. Fazio called this automatic activation of attitudes. That requires that they're somewhat stable, that they're independent of the context and independent of time in the way people answer, and that they are predictive of behaviors and resistant to persuasion. And that's the reason why they're popular for questionnaires, you know, you can ask about these attitudes and you hope that you can make behavioral predictions from these attitudes. However, there's a problem with this traditional view, because in surveys at least, when we ask them. So methodologists found big differences in numbers depending on how such attitude questions are asked. So, here are two examples. 56% say that the US should not allow public speeches in favor of communism, if asked that way in a survey. And 39% say that US should forbid public speeches in favor of communism. Conceptually kind of the same but big differences in the numbers. Likewise, 25% say that the Federal government is spending too little on welfare, whereas 65% say that the federal government is spending too little on assistance to the poor. Again, big difference between the two and conceptually sort of the same thing. That shouldn't be the case if attitudes are stable and not effected by question wording, context, or things of that nature. So, for that reason, alternative views have developed. There's a nice title by John Zaller (1992). We have that on the reference list. It's called "Making it up as you go along." So, captures the spirit here. And in it he writes that individuals do not typically posses true attitudes, rather they construct opinion statements on the fly based on whatever considerations are momentarily salient. Norbert Schwarz has written a lot on attitudes and attitude measurements. In a 2007 piece he wrote about attitudes being "conceptualized as evaluative judgments formed on the spot rather than a trait-like disposition." And that's important because it A) effects whether you want to measure those attitudes in the first place. And if so, how you measure the attitudes. So let's recap. In these alternative views, people don't have stable summary evaluations stored in memory. If you ask for one, people will construct them on the fly. So they give you top of the head responses on whatever happens to cross their mind. And it also means that tomorrow you will get a different answer. And small change in the wording or the question order can get you very different responses. People also will report attitudes to things that don't even exist. And that's an interesting feature, true in general for questionnaires, we've sort of by the age of two or three lost the ability to not answer questions, so usually you will get an answer no matter what the question is. Tourangeau, Rips, and Rasinski, and in particular Rasinski, have written a lot on these in their various pieces. Here Rasinski distinguishes between attitudes, seen as unobservable, global evaluation of objects, and attitude expressions, which are specific responses to a specific question asked at a particular time in a particular way. So any single attitude measure will be an imperfect reflection of the underlying attitude. Can see this as a little bit of a reconciliation about these different views. So there's something, you know, inherent in a person that you might be able to measure, but how it is expressed is a very different thing. So in this Rasinski (2011) piece he, as I said, tries to resolve the divergent views and says that some attitude expression almost entirely constructed on the spot, just like Zaller says. Other attitude expression, they are retrieved almost entirely intact from memory. Most attitude expressions are some combination of the two. That means they're constructed by a chief ingredient and the construction is a fairly stable overall summary evaluation retrieved from memory, that is the actual attitude. But, all these questionnaire design components that we talk about, they go into this formation of the response. Norbert Schwarz, in the 2007 paper, that is part of the course notes, he resolves these divergent views in a slightly different way. He translates stable attitudes, strong attitudes, context effects, and nonattitudes into something that matches this more modern view on attitudes. Take a moment and retreave from the reading or imagine how you would translate that in the context that, if you conceptualized the construction, elevation, and judgements on the spot, drawing on what is available to the respondent. [BLANK_AUDIO]. So, stable attitudes, that means a similar evaluation is arising from the respondent drawing on similar inputs or the same chronically accessible information. If you see strong attitudes, that means the respondent is drawing on highly accessible information. So, you know, if a answer comes out quickly, that might be highly accessible. Context effects that you see, it means that respondents are drawing on contextual information when constructing evaluations. And nonattitude means that we have the same number of conflicting information accessible to respondent at the same time. All of this has implications for questionnaire design. You have to watch out for context effects, due to question wording, prior and later questions, self administered questions. Means, the presentation of the entire questionnaire because you don't have that much control over what comes first and what comes later. In interviewer administered questions, you have to include interviewer characteristics, behaviors, and introductions to the survey in the set of context effects that you have to worry about. And the same is true for external factors, such as weather or mood. They have shown to influence respondents' answers as well. The other issues, for example, the specific evaluations, or, you know, versus global evaluations, agree versus disagree scales, filtering of don't know and hypothetical questions. And we will talk about these other issues in the next segments.
