Search

Dillman's Total Design Method (TDM) for mail surveys has proved effectual in attaining high response rates in the U.S.A. The TDM, however, has two drawbacks: its relatively high costs of a mail survey and its susceptibility to response errors. We tested the ability of the TDM techniques to increase response rates. Furthermore, the influence of the TDM on the response error was investigated. In this experiment, the effects of personalization of the cover letter and a final reminder by certified mail were tested on a sample of the Dutch population. If the TDM is followed completely the response rate in the Netherlands is comparable to the response rate in the U.S.A (70%). Omission of personalizing the cover letter or sending a reminder by certified mail results in a significantly lower response rate. Using the TDM in toto does not lead to responses of inferior quality. We find that a reminder by certified mail in combination with a personalized cover letter does not result in more socially desirable answers, more item non-response, or a more deviating response pattern. However, the use of only a personalized cover letter, i.e., without a certified mailing, did result in more socially desirable answers.

Although Statistics is not a very popular course according to most students, a majority of students still take it, as it is mandatory at most Social Science departments. Therefore it takes special teacher’s skills to teach statistics. In order to do so it is essential for teachers to know what students’ attitudes are and what factors influence course outcomes. In this study an answer was sought to the question what the effect is of educational and student factors on course outcomes with respect to introductory courses Statistics at Dutch and Flemish institutions of higher education. The theoretical model for this study was built around the Expectancy Value model that explains achievement related choices. In this model achievement behavior is looked upon as a function of student expectations, motivation, beliefs, goals and the task value. A student will choose the best possible combination in order to obtain the expected success (i.e. pass the course). Motivations and beliefs are influenced by individual factors, such as self confidence and previous experience and by educational factors such as course-organization. In a large-scale field experiment, I collected pretest- and posttest attitude-data of 2,555 students at 11 institutions. For this purpose I used the ‘Survey of Attitudes toward Statistics’, developed by Schau. Additionally a number of expert-interviews were conducted. Besides answering the question above, I tested some innovative statistical tools, i.e. Latent Change Method Effect Models. They analyze true intra-individual change and filter out method effects that are due to differences in the set-up of the field experiment. Additionally, so-called ‘Propensity Related Covariates’ were developed. These covariates contain the combined effect of individual or institutional factors on statistics-attitudes and, in turn, on student achievement. Generally, attitudes toward statistics can be split up into six components: Affect, Cognitive Competency, Difficulty, Value, Interest and Effort. The results of this study show that attitudes on average are located in the top half of the attitude scale, with the exception of Difficulty. They do change over the semester, but only mildly and not always in the direction the teacher would want. Although students perceive a positive change in their competencies throughout the semester, they still find the course equally difficult. Effort has a special position among the six components as it describes the individual behavior in order to pass the course, contrary to the other components where the student evaluates the course in a more passive way. Additionally, not so much organizational but individual factors affect attitudes towards statistics and student achievement. These factors are self confidence, mathematics experience, gender and age. Therefore this thesis concludes with a number of recommendations for teachers to develop and organize their course and motivate their students. Lastly, future research should concentrate on ‘teacher quality’ a bit more, as this concept remained underexposed in this study. The most important message that I want to transmit is that highly motivated teachers can motivate their students and as a result student achievement will be high irrespective of the topic.

The Hunt for the Last Respondent has been inspired by concerns about the possibly detrimental effect of nonresponse on the accuracy of survey outcomes, as response rates are generally considered to be the most important criterion of survey quality, and the Netherlands is notorious for its low response rates. It addresses the following questions: Why are high nonresponse rates a reason for concern? Who are less likely to respond, either because they are more difficult to contact or because they are more reluctant to cooperate? How can response rates be enhanced? These questions are addressed in an extensive study of the international literature on nonresponse presented in part 1 of this thesis. Nonresponse is worrisome when those persons who are not contacted or who refuse to cooperate differ from respondents and these differences cannot be adjusted for by using auxiliary information, generally background variables. Part 1 gives an overview of background variables as correlates of contactability and cooperation. In addition, attention is paid to causal factors of survey participation, such as the attitude towards surveys, the topic of the survey, social isolation and social participation, and to the distinction between situational and survey-related reasons for refusal. Finally, it presents an overview of response enhancing strategies. The second, empirical part of the study elaborates on the literature presented in part 1 and addresses the questions: How to study nonresponse? Do enhanced response rates improve the accuracy of survey outcomes? How to combat nonresponse error and allocate funds effectively? To answer the first question, two instruments and two models are introduced. The instruments are standardized contact forms on which to register fieldwork details and a follow-up survey among persistent refusers. The first can provide information on the degree of contactability and cooperation of the respondents, the second on final nonrespondents. The models assume different relationships between difficult respondents and final nonrespondents. Nonresponse on two surveys is analysed, namely the Dutch Amenities and Services Utilisation Survey 1999 and its follow-up survey among persistent refusers, and the European Social Survey 2002/2003. In these surveys high response rates were obtained, even among adamant refusers. Based on the analyses the final two questions of this study were answered. It turns out, as others have found earlier, that enhancing response rates does not always improve the accuracy of survey outcomes. The study recommends to spend a part of the funds for data collection on obtaining information about final nonrespondents as this is more effective than raising the response rate by a few percent.