Studies and Experiments Towards a Computer Interview Acceptable to the Naive User

This thesis is concerned with concepts and procedures involved in the development of programs by which computers may interview people. Based on the hypothesis that the style of the questions is very important in a computer interview, an analysis of four medical interviewing programs was attempted. Two variables emerged from this analysis namely Encouragement (Encouraging Phrases) and Chattiness (Chatty Phrases). Six experiments were then conducted to test the usefulness of the above two variables in a computer interview. The results of the six experiments were as follows: a) The use of random Encouragement and Chatfiness; seems to have a significant effect on. interviewee's acceptance of a computer interview. b) There appears to be an optimum for the amount of Encouragement and Chattiness used in a computer interview. c) Random Chattiness seems to have no effect on interviewee's acceptance of a human interview, while random Encouragement might have a negative effect. d) For a computer interview random patterns of Encouragement and Chattiness seem as good as more planned patterns of the two variables. e) The accuracy of the derived information and the interviewees' acceptance of a computer interview on a general subject may be as good as that for a human interview. From these results the original hypothesis is seen to be too limited and in its place a conceptual analysis of the essential features of computer based interview programming is developed. "