Search Results
Questionnaires can be useful for evaluating persuasive texts. However, the researcher has to be careful in choosing an experimental design and dependent variables when using them. There are two widely used true-experimental designs: the pretest-posttest design and the posttest-only design. The pretest-posttest design has the attraction of measuring the persuasiveness within the same subject. However, this apparent advantage turns out to be a major disadvantage. Results show that pretesting leads subjects to stick to their initially expressed attitude thereby obscuring persuasive effects.
The choice of the dependent variable is another important decision when evaluating persuasive texts. One can measure persuasiveness at the beliefs level (Does the reader accept the argument?), at the attitude level (Does the reader get a favourable impression of the described object?), and at the intention level (Does the reader intend to perform the advocated action?). Although these levels are functionally related, results show that changes at one of these levels do not automatically imply changes at one of the other levels. Therefore, when evaluating persuasive texts it is safer to measure all three dependent variables.
When performing complex tasks, one has to perform a number of hierarchically related actions. When writing an instructive text for such a task, one has to translate the hierarchical relations into linear text. This translation can follow a breadth-first or a depth-first approach. The breadth-first approach gives all instructions at one level of the hierarchy before giving all instructions at a lower level. The depth-first approach gives all instructions for one branch of the hierarchy before giving instructions for the next branch.
A first exploratory study was determined whether people have a breadth-first or a depth-first conceptualisation of the task. The results showed a preference for the depth-first conceptualisation. A second experiment investigated whether the depth-first approach led to a better task performance. However, the results showed that the breadth-first approach yielded a more accurate plan and was appreciated more by users. The implications of these results for models of processing written instructions are discussed.