Veröffentlichen auch Sie Ihre Arbeiten - es ist ganz einfach!
Mehr InfosDiplomarbeit, 2000, 89 Seiten
Diplomarbeit
Justus-Liebig-Universität Gießen (Sprache, Literatur, Kultur)
1,0
LIST OF ABBREVIATIONS
LIST OF TABLES AND FIGURES
1 INtroduction
2 Marketing Research
2.1 Primary Research: Surveys
2.2 Sampling
2.3 The Pretest
3 The Symbolic Interaction Theory for Social Research
4 Questioning Modes
4.1 Personal Interviews
4.1.1 Advantages of Personal Interviews
4.1.2 Disadvantages of Personal Interviews
4.2 Telephone Interviews
4.2.1 Advantages of Telephone Interviews
4.2.2 Disadvantages of Telephone Interviews
4.3 Self-administered Questionnaires
4.3.1 Advantages of Self-administered Questionnaires
4.3.2 Disadvantages of Self-administered Questionnaires
4.4 Frequency of Survey Techniques
5 Response Effects in Questionnaires
5.1 Response to Question Words
5.2 Response to Question Form
5.3 Response to Question Structure
5.4 Response to Question Topic
6 Macrostructural Level
6.1 The Structure of Questionnaires
6.1.1 VTC Atego/Eurocargo Survey
6.1.2 SunExpress General Customer Survey
6.1.3 Walt Disney's World on Ice Audience Survey
7 Microstructural Level
7.1 Investigation of Question Number
7.2 Investigation of Question Form
7.3 Syntactic Level
7.3.1 Sentence Types
7.3.1.1 Investigation of Sentence Types.
7.3.2 Sentence Forms
7.3.2.1 Investigation of Interrogative Sentences.
7.4 Lexical Level
7.4.1 Words per Interrogative Sentence
7.4.2 Wh -interrogatives
7.4.3 "You" attitude
7.4.3.1 "You" attitude in the VTC Atego/Eurocargo Survey.
7.4.3.2 "You" attitude in the SunExpress General Customer Survey
7.4.3.3 "You" attitude in Walt Disney's World on Ice Audience Survey
8 Comparison of the Results
8.1 Personal Questionnaires
8.2 Telephone Questionnaires
8.3 Self-administered Questionnaires
9 Conclusion
BIBLIOGRAPHY
ERKLÄRUNG
illustration not visible in this excerpt
Table 1: Survey techniques for the total period 1987-1996
Table 2a: Responses to positive vs. negative words in questions
Table 2b: Responses to positive vs. neutral words in questions
Table 3a: Responses to open and closed attitude questions
Table 3b: Responses to open and closed behavior questions
Table 4: Question structure
Table 5: Number of questions in questionnaire parts
Table 6: Number of questions per questionnaire
Table 7: Question forms per questionnaire
Table 8: Number of sentence types per questionnaire
Table 9: Structure of interrogative sentences
Table 10: Words per interrogative sentence
Table 11: Number of wh -interrogatives
Figure 1: Model of the symbolic interactionist view of question-answer behaviour
Marketing research is the fundamental basis of virtually all marketing activities. Therefore, it is of invaluable importance. Researchers have a variety of research instruments available to conduct a survey: face-to-face interviewing, telephone interviewing and self-administered questionnaires. Questionnaires are one of the most frequently used means in marketing research to investigate attitudes, beliefs and preferences of consumers. Thus, the objective of this thesis is to analyze and compare three different kinds of questionnaires used in marketing research. As questioning for information is the main theme of this paper, the focus will be placed on the linguistic description of questions.
"Although ours is a questionnaire intense society, with thousands of business forms, federal and state income tax forms, and other questionnaires, not everyone is experienced in completing formal questionnaires." (Fletcher/Bowers 1990: 115).
Even though the term question is generally understood, a short definition will be provided. In this thesis the term question will be used in a very broad sense in accordance with Stenström (1984: 1): "Given that speakers A and B cooperate, a question (Q) is any utterance by A that may elicit a response (R) from B." (own emphasis). Therefore, questions need not be stated in the interrogative form in order to count as such. The interview situation itself makes it clear to respondents that the utterances of the interviewer have to be interpreted as questions. Interrogative sentences will then be only those sentences that are questions due to their syntactic form. This will be explained in more detail in chapter seven.
The research question, which is to be seen as the thread running through this work, can be formulated as follows: What are the differences between questionnaires used in personal face-to-face interviews, telephone interviews and self-administered questionnaires and how are these differences reflected in language? My claim is that self-administered questionnaires use different questioning strategies and can clearly be distinguished from questionnaires used in personal face-to-face interviews and in telephone interviews. I also assume that the former ones are easier to complete because they contain shorter and more simple sentences. This assumption is based on the fact that in self-administered questionnaires, as the name already indicates, respondents are left to themselves when filling out the questionnaire.
In this thesis it will be further investigated whether questionnaires used in face-to-face interviews can be distinguished from those used in telephone interviews. I suppose that telephone questionnaires take a position in between questionnaires used in face-to-face interviews and self-administered questionnaires. On the one hand, the interviewer on the telephone can be asked for clarification if the meaning of a question is unclear to the respondent. In this way telephone questionnaires resemble face-to-face questionnaires. But on the other hand, in telephone interviews no supplementary material e.g. show cards, can be used to facilitate understanding. Thus, in another way, telephone questionnaires resemble self-completion questionnaires. As a consequence, I expect to find questioning strategies from both other questionnaire types in telephone interviews.
Of these three different types of questionnaires mentioned, three from each category have been chosen to be analyzed, so that in total the corpus consists of nine questionnaires.[1] Whereas in the beginning of the linguistic analysis the entire corpus serves as basis for the investigations, later three questionnaires (P3, T2, S3), one from each category, have been selected for a more detailed analysis. They serve as examples of the different questioning modes. It is to be noted that because of the restricted number of questionnaires analyzed, all my investigations remain limited in scope and cannot be regarded as being representative. However, this is not the aim of this thesis. The goal is to point out general differences.
Not only private marketing research companies make use of questionnaires, but all sectors involved in the survey industry do, e.g. federal governments, the academic sector, the mass media, etc. Therefore, questionnaires from the social sciences are included in the analysis as well.
In the following paragraphs, it will briefly be explained how I intend to proceed. Chapters two to four lay the theoretical foundation for the linguistic analysis in chapters five to seven. Following this introduction (chapter one), chapter two is dedicated to marketing research. The purpose of surveys will be explained as well as the importance of the sampling procedure and of the pretest. Moreover, a distinction between primary and secondary research will be made. Nevertheless, in this thesis only primary research is of interest since the collection of information by questionnaires belongs to primary research.
Chapter three introduces the symbolic interaction theory for social research. This model of question-answer behavior provides a theoretical description of the interview situation which has influence on the completion of questionnaires.
In chapter four the advantages and the disadvantages of the three different questioning modes will be outlined, personal interviews, telephone interviews, and self-administered questionnaires. This is essential to the thesis because in the practical part it will be analyzed how the features of the questioning modes are reflected in questioning strategies. Following this detailed information, a table will indicate the frequency of the different survey techniques.
In the course of this paper, personal face-to-face questionnaires are simply called personal questionnaires, which, however, does not imply that questionnaires used in telephone interviews are impersonal. The terms personal questionnaires, telephone questionnaires and self-administered questionnaires serve well the purpose of distinguishing between the different questioning modes. Furthermore, the terms self-administered and self-completion questionnaires will be used interchangeably.
Chapter five deals with response effects in questionnaires and explains the influence of question words, question form, question structure and question topic on responses. This structure is based on the book The Practice of Questioning by J.T. Dillon (1990). Due to the unavailability of completed questionnaires, an own analysis of answers is impossible. Questions and answers, however, cannot be studied in isolation. Thus, regarding response effects, Dillon's findings will be taken as a given.
Chapter six is dedicated to the analysis of the macrostructure, the overall structure, of questionnaires. As an example, one questionnaire from each of the three different questioning modes will be analyzed in depth. The purpose is to find out whether there is a common structure in all questionnaires.
Chapter seven focuses on the microstructure of questionnaires. After a comparison of the number of questions in the selected questionnaires, question form will be analyzed. A basic distinction will be drawn between open-ended and closed questions. Chapter 7.3 focuses on the syntactic level where sentence types and sentence forms will be studied. In chapter 7.4 the emphasis is on the lexical level. First, a look at the average number of words in interrogatives will be taken, then wh -interrogatives and the "you" attitude will be considered.
Analyzing all these aspects is necessary to provide a good picture of the different questioning strategies pursued by the individual questionnaires and to finally answer my research question. In chapter eight, I will summarize the results for each questionnaire type separately in order to demonstrate how language differs in the three questioning modes. In chapter nine, the conclusion, some final remarks will be made.
The proceedings in this paper will be from the more general to the detailed. This means that the introduction of the necessary terminology will be followed by the analysis of response effects. The macro-level, then, focuses on questionnaires as a whole, and the micro-level on single questions and question words.
It is to be noted, that "question-answer behaviour involves complex interrelationships between sociological, psychological and linguistic variables." (Foddy 1993: XI). Therefore, while a lot of research on questionnaires can be found, most of it is from a social sciences perspective or a marketing perspective. Authors like Converse and Presser or Sudman usually take a prescriptive stance and set out to give advice on how questionnaires have to be constructed. But, this is not the aim of descriptive linguistics. It seems that from a linguistic point of view, questioning strategies in questionnaires have not been analyzed much, since it is very hard to find linguistic literature. Thus, the interdisciplinary nature of this thesis is reflected in the bibliography. Books from the fields of marketing, sociology, psychology and, of course, linguistics have been consulted. According to Foddy (1993: X) Stanley L. Payne's book The Art of Asking Questions from 1951 is still the one in the field of question formulation to which most credit is attributed.
This chapter provides some general information on marketing research and questionnaires. The definition of primary and secondary research will be followed by an explanation of the steps involved in the sampling process. Furthermore, the importance of pretests will be emphasized. However, first a definition of marketing research will be provided. According to Giles (1994: 29) "marketing research may be defined as the objective and systematic collection, recording, analysis, interpretation and reporting of information about: existing or potential markets (i.e. market research [...]), marketing strategies and tactics, and the interaction between markets, marketing methods and current or potential products or services."
The origins of marketing research can be found in the United States at the beginning of the 20th century where the first formal marketing research organization was established in 1911. Whereas in the beginning the development proceeded slowly, marketing managers have now realized the great importance of up-to-date information on relevant markets (cf. Giles 1994: 29f.). However, the collection of information is essential not only in marketing research, but also in social sciences.
In this thesis no distinction will be made between social surveys and surveys used in marketing research since the questionnaires used in different kinds of surveys are very similar. This has been stated by Atteslander (1993: 126):
"Markt- und Meinungsforschung (Demoskopie) haben in allen industrialisierten Gesellschaften seit Mitte der 30er Jahre eine ungeheuere Entwicklung genommen, so daß Umfragen in der Öffentlichkeit oft mit empirischer Sozialforschung gleichgesetzt werden."
According to Kotler (1991: 102f.) the marketing research process usually involves the following five steps:
1) Definition of the problem and research objectives
2) Development of the research plan
3) Collection of the information
4) Analysis of the information
5) Presentation of the findings
This thesis, however, deals only with point number three, the collection of information by means of questionnaires.
Contrary to secondary research, which makes use of already existing sources, primary research refers to the collection of new data. This can be achieved through observation, experimentation, or surveys. Surveys are frequently employed in primary research and are the only method of data collection dealt with in this paper.
"Before manufacturers put a new product on the market, they want and need to know how consumers react to that product. Marketing research is conducted with the goal of predicting such consumer reactions." (Guy 1987: 224).
The aim of marketing research, however, is not only to predict consumer reactions, but also to gather information about the entire market. Rogge (1992: 19) states that it depends on who conducts the research.
"Der Gegenstand der Marktforschung ändert sich mit dem Träger der Marktforschungsaktivitäten. So sind Verbände oder andere Zusammenschlüsse von Unternehmen mehr an übergeordneten Branchen und gesamtmarktbezogenen Daten interessiert." (emphasis of the author).
According to Kotler (1991: 106) surveys are best suited for descriptive research. In order to gather information on people's knowledge, attitudes, preferences, etc., the questionnaire is the ideal means.
"Broadly speaking, a questionnaire consists of a set of questions presented to respondents for their answers. The questionnaire is very flexible in that there are any number of ways to ask questions. Questionnaires need to be carefully developed, tested, and debugged before they are administered on a large scale. One can usually spot several errors in a casually prepared questionnaire." (Kotler 1991: 107).
Before any specific decision on the format and content of a questionnaire can be made, it is essential to define the research question. Furthermore, the target population which is to be surveyed must be determined and a decision on the sample size has to be made. This will be explained in the next chapter.
The selection of respondents to be surveyed is called sampling. Sampling is very important because the respondents are the ones from whom conclusions will be drawn (cf. Chee/Harris 1993: 64). According to Rossi/Wright/Anderson (1983: 1)
"sample surveys consist of relatively systematic standardized approaches to the collection of information on individuals, households, or larger organized entities through the questioning of systematically identified samples of individuals."
In a census the total population is sampled, however, this is hardly ever possible in marketing research. Nevertheless, everybody must have an equal chance of being sampled. Apparently, large samples provide more reliable results, but they are also more costly. Therefore, probability sampling seems to be the best way to select respondents.
"The basic type of probability sample is the simple random sample in which every item in the relevant universe has an equal opportunity of being selected [...]. In contrast, a non-probability sample is arbitrary and does not permit use of standard statistical tests. Examples of non-probability samples are the convenience sample and the quota sample." (Chee/Harris 1993: 64f.).
Unfortunately, it is beyond the scope of this paper to explain all the different sampling techniques. However, broadly speaking the term sampling can be "defined as the selection of a limited but representative number of a larger population." (O'Sullivan et al. 1983: 190f.). The following chapter focuses on pretests, which have to be conducted with people from the actual sample.
After the specific questions are formulated (cf. chapter five), it is essential to pilot-test and revise these questions to make sure that the respondents will be able to provide the information desired. Obviously, this takes time, effort, money and patience (cf. Dillon 1990:113). According to Sudman/Bradburn (1982: 283) the survey should not be conducted if the resources to pilot-test the questionnaire are not available.
"Erst wenn der Fragebogen ein Höchstmaß an Klarheit und Übersichtlichkeit erreicht hat, wenn er für die Befragten in jeder Einzelheit unmißverständlich ist und von den Befragern leicht und sicher gehandhabt werden kann, ist eine Gewähr gegeben, daß die Interviews ohne Schwierigkeiten und unter den gleichen Bedingungen verlaufen." (von Kirschhofer-Bozenhardt/Kaplitza 1975: 126).
An effective way of conducting a pretest is to send questionnaires to people of the target group which is to be sampled. Afterwards, respondents will be contacted again. They have to explain their answers and tell whether they found that the skip patterns and questions were logical. The purpose of this process is to identify questions that cause confusion.[2]
In addition, data analysis and report preparation costs must be covered by the budget. If no computers are used in the data-gathering process, questionnaire responses have to be converted to a form for computer analysis after conducting the survey. Furthermore, researchers must be able to cover the expenses for preparing and distributing a written report of the survey.
Considering all these aspects one might come to the conclusion that the costs of getting the answers outweigh the benefits. If primary research seems not worth the effort, the researcher will have to rely on secondary resources.
This chapter is dedicated to the introduction of the symbolic interaction theory (cf. Foddy 1993: 19ff.). It provides a theoretical framework of question-answer behavior and explains the artificiality of the interviewing situation. The sociologist Herbert Blumer coined the term symbolic interactionism in the 1960s. He claims
"[...] that social actors in any social situation are constantly negotiating a shared definition of the situation; taking one another's viewpoints into account; and interpreting one another's behaviour as they imaginatively construct possible lines of interaction before selecting lines of action for implementation." (Foody 1993: 20).
These assumptions can be related to survey research where verbal behavior is of primary importance. When the research situation is unclear to respondents or they are unsure what kind of answers the researcher is looking for, they will try to find clues to reach a mutual understanding with the researcher.
"Jede Befragung stellt eine soziale Situation dar. Dazu gehören nicht nur die Menschen, die miteinander sprechen, sondern auch die jeweilige Umgebung. Von sozialer Situation ist selbst dann zu sprechen, wenn jemand für sich allein auf einen schriftlichen Fragebogen Antwort gibt, oder wenn er telefonisch befragt wird. Gegenseitige Erwartungen, Wahrnehmungen aller Art beeinflussen Verhalten und verbale Reaktion." (Atteslander 1993: 129).
The interaction between interviewers and respondents can be illustrated in the following figure.
Figure 1: Model of the symbolic interactionist view of question-answer behaviour
illustration not visible in this excerpt
Source: Foddy (1993: 22) (emphasis of the author).
This model involves a complex four-step communication cycle. When encoding questions or decoding answers, interviewers always take into account presumptions about themselves and about respondents. The same is true for respondents who include their own presumptions and those of the interviewer when they decode a question or encode an answer. As can be seen, successful communication is only possible if respondents understand questions the same way that the interviewer does, and the interviewer understands answers in terms intended by respondents.[3]
This insight has been formulated by Kromrey (1998: 342f.).
"Alle am Kommunikationsprozeß Beteiligten müssen die erforderliche Kommunikative Kompetenz besitzen, d.h. jeder Beteiligte muß die Bedeutung der Zeichen kennen, mit deren Hilfe kommuniziert wird (hier: sprachliche Zeichen, deren Grammatik und Semantik), und zwar müssen alle Beteiligten (Forscherin, Interviewer, Befragter) den Zeichen die gleiche Bedeutung beimessen." (emphasis of the author).
As in social and market research usually a large number of people is interviewed, it is a big challenge for researchers to word questions in such a way that they provide the same frame of reference to all respondents. This, however, is essential with regard to standardization and comparability of answers.
Interviews represent an artificial communication situation. Kromrey claims that interviews can never be a neutral survey instrument. According to Kromrey (1998: 339) interviews differ from normal conversations in three aspects. First, the interacting persons are strangers to each other. Second, interviews represent an asymmetrical social situation in which one person asks the questions and the other person gives answers. And third, because of assured anonymity, the interview has no social consequences. The fixed structure in interviews can be described as: question-answer-question-answer sequence. Although in the symbolic interaction theory respondents are not merely passive agents. "They should be seen as being engaged in joint 'sense-making' activities with the researcher." (Foddy 1993: 23).
Chapter five of this thesis deals with the implication of the preceding statements concerning response to question words, question form, question structure and question topic.
The following chapter concentrates on the different research instruments available for conducting surveys: personal interviews, telephone interviews and self-completion questionnaires. The advantages and disadvantages of each questioning method will be explained, and some figures about their frequency will be provided. However, this does not imply that any survey method is superior to any other. The utility always depends on what the researcher intends to find out. Therefore, the drawbacks of one method might be the benefits of another.
First, a distinction has to be made between personal interviews that are structured and those that are unstructured. As the name already indicates, no formal questionnaires are used in unstructured interviews. Therefore, they will not be of further interest to this thesis. All information provided on personal interviews will relate to personal structured interviews where the interviewer asks the questions in a prescribed form and order.
In face-to-face interviews, respondents are contacted personally by the interviewer in order to answer survey questions which the interviewer reads out to them. The respondents also give their answers orally, but sometimes charts, cards or other visual aids are provided in order to facilitate the answering process (cf. Guy 1987: 244).
Personal interviews are employed when personal data is to be gathered, and when it is necessary to overcome resistance by the interviewee. Furthermore, personal interviews can lead to very accurate data since it is possible to supplement answers by observation of extra-linguistic features.
"This survey method typically is the best means of obtaining detailed information, since the interviewer has the opportunity to establish rapport with each respondent and explain confusing or vague questions." (Chee/Harris 1993: 63).
As already mentioned, one great advantage is that interviewers can use show cards, product displays and card games in order to make the interviewing process less monotonous, reduce ordering effects and motivate the interviewees to cooperate. As a consequence, much information can be obtained and information can be gathered in great depth. Moreover, the interview situation can be controlled and therefore influence by third persons is limited. This leads to highly reliable results (cf. Weis/Steinmetz 1998: 73ff.).
A recent development in personal interviewing is the use of laptop computers. This interviewing technique is called CAPI (computer assisted personal interview).
"The computer-assisted interview is becoming more and more frequent, in which data is gathered directly in machine-readable form, obviating the use of pencil and paper and the subsequent coding of responses." (Kaase 1999: 164).
Since the interview situation as such differs little from the traditional mode (cf. Kaase 1999: 241), the following analysis will be limited to traditional face-to-face and telephone interviews.
The previously cited advantages usually outweigh the disadvantages of very high costs and the long time span needed to collect primary data through personal interviews. Costs increase because often respondents cannot be reached at their homes or they refuse being interviewed. Furthermore, presently fewer and fewer people are willing to let strangers into their homes (cf. Porst/Ranft/Ruoff 1998: 4f.). This makes advance letters or calls, to announce the interview, obligatory. As an alternative, personal interviews may have to be conducted at public places.
Another disadvantage is the large potential for bias in personal interviews. Respondents might want to create a favorable image of themselves and distort their answers. Therefore, in order to reduce bias it is necessary to match the characteristics of respondents and interviewers.
"In a face-to-face survey, the interviewer's sex, race, and nonverbal cues can bias results. Most people prefer not to say things they think their audience will find unacceptable. For that reason, women will be more likely to agree that sexual harassment is a problem if the interviewer is a woman. Members of a minority group are more likely to admit that they suffer discrimination if the interviewer is a member of the same minority." (Locker 1997: 376).
Not only respondents but also interviewers are a source of bias (cf. Giles 1994: 42). Consequently, in order to ensure a standardization of the interviewing process it is extremely important to train interviewers well and to provide clear interviewer instructions. For example, no personal interpretation can be given to any question and skip patterns and filters have to be followed rigidly. Furthermore, interviewers are not allowed to show approval or disapproval of answers given. Another great danger is that interviewers may fill out the questionnaires by themselves. This must be prevented by all means.
Telephone interviews differ from personal interviews in that only verbal communication is possible.
"The telephone interview is a cross between the mail survey and the personal interview. It is a personal interview in the strict sense because it involves a conversation between two human beings. However, it also is similar to a mail survey because it employs a fairly uncomplicated questionnaire." (Fletcher/Bowers 1990: 124).
Today, the number of households without a telephone does not constitute a major source of bias since in Germany over 90 percent of the households have a telephone (cf. Kaase 1999: 168). Above all, people from lower socio-economic groups still do not have a telephone, but these echelons of the society are generally of less interest to marketing research because their purchasing power is low.
Like personal interviews, telephone interviews can be facilitated by the use of computers. Researchers can read the survey questions from the computer screen and directly key in the answers of the respondents. In the 1980s Rossi/Wright/Anderson (1983: 9) already pointed out that computer assisted telephone interviews (CATI) will become more and more the standard, since the data tape they provide is readily available for further analysis.
The greatest advantage of telephone surveys is the rapidity with which data can be collected (cf. Salant/Dillman 1994: 38). It is also much cheaper to make telephone calls than to send interviewers to various locations. Therefore, more attempts to contact people can be made by fewer interviewers. Another advantage is the reduction of interviewer bias and the availability of data under better controlled conditions. If interviews are conducted from telephone studios, the supervision of interviewers is very easy. According to Giles (1994: 41) telephone interviews are "useful for radio and television surveys when checking that programmes are actually being heard or viewed."
Kaase (1999: 157) points out two important drawbacks of telephone interviews:
"first, the restricted interview duration of about thirty minutes, an insuperable limit imposed by experience with respondent resistance (but which need not apply when respondents are highly interested in the survey topic); and, second, the need to forgo tried and trusted complex, often visual survey tools developed with great effort, because their use on the phone is impossible."
As no visual aids can be used, e.g. show cards or lists, the number of alternatives offered to respondents must be limited. It is likely that respondents cannot memorize a large number of options and therefore simply select the first or last ones.[4]
Whereas impersonal information is quite easily obtained, personal information is hard to collect by telephone (cf. Chee/Harris 1993: 63). Furthermore, most answers will be spontaneous because respondents have very little time to consider their answers. This has to be taken into account when comparing data from different questioning modes.
A huge problem in telephone interviews is that interviewers cannot prove their identity. Frequently, people regard telephone interviews as an intrusion to privacy and are not willing to provide any information at all to a caller whose identity is unproved (cf. Litter 1984: 123f.). It is easier to refuse being interviewed on the telephone by simply hanging up on an anonymous voice than sending away an interviewer standing at the door (cf. Bradburn/Sudman 1988: 102). Moreover, bias occurs because of the frequent use of answering machines to screen callers. However, since interviewers only have available their voice to create rapport with the interviewee, interviewer influence is limited, too.
Last but not least the sample selection procedure from telephone books can lead to distortion and unrepresentative findings. Using this method, only people listed in telephone books have a chance of being selected. But presently many unlisted numbers exist (about 20%), and telephone books are quickly outdated (cf. Kaase 1999: 168).
Self-completion questionnaires are usually sent by mail to the respondents, but they can also be distributed personally. This occurs, for example, after a performance when the audience is asked for feedback (e.g. S3) or after a course when teachers are to be evaluated. Nevertheless, the absence of interviewers is the prominent feature of this questioning mode. "The interviewer is present only symbolically, in the form of written questions." (Bradburn/Sudman 1988: 95).
For mailed questionnaires cover letters are mandatory since people must be persuaded to participate in the survey. Cover letters usually explain who conducts the research, the purpose of the questionnaire, how people were selected and how the data will be used. Furthermore, an assurance of anonymity and a stamped and self-addressed return envelope are essential. Sometimes, a due date is indicated or some kind of incentive for participating is offered to respondents, e.g. to share the results of the survey (cf. Wilk 1975: 190f.).
Since no interviewer is present to ask questions orally, detailed instructions on how to fill out the questionnaire will be provided on the questionnaire itself. These instructions must be clearly set apart from the actual questions.
One of the greatest advantages of self-completion questionnaires is that they are relatively easy to distribute and to collect. Due to the fact that a large number of people can participate in mail surveys, it is possible to gather more information by mail surveys than by personal interviews. Another advantage of mail questionnaires is that respondents can answer the questions whenever they have the time to do so. Furthermore, mail questionnaires can lead to more correct results because no bias and distortion by interviewers occurs (cf. Wilk 1975: 187).
Brusaw/Alred/Oliu (1987: 529) point out that mail questionnaires enable the researcher to obtain responses from people spread over a wide geographical area. Thus, individuals who live in remote areas or individuals who would not give personal interviews are easily reached. Another advantage is that the costs of mail questionnaires are usually lower than the costs of personal or telephone interviews since no large staff of trained interviewers is required. However, the costs of follow-up letters, due to low response rates, must be included in the calculations.
After having mentioned the advantages of self-completion questionnaires, the disadvantages cannot be neglected. Compared to an interview, mailing questionnaires and waiting for replies takes considerably longer. It has already been indicated that one of the largest problems with mail questionnaires is their response rate. Nevertheless, it is hard to specify an adequate response rate for a survey to be valid (cf. Fletcher/Bowers 1990: 123). When the response rate is low, people have to be reminded to fill out the questionnaire. In order to overcome non-response bias, research agencies can send follow-up letters or make telephone calls. Follow-up letters must be sent to all of the people who take part in the survey if anonymity is to be guaranteed. If an identifying number is put on questionnaires or return envelopes, to screen respondents who already answered, subjects might become suspicious.
Contrary to telephone interviews, mail questionnaires are definitely the wrong research instrument if spontaneity on a subject is required. At home people generally read through the entire questionnaire before answering it and frequently consult records. When they find a question too hard to answer, respondents might simply skip to the next question. This will result in incomplete survey data.
"Unless additional information is obtained from non-respondents, the results of postal interviews are likely to be biased, since there may be important differences in the characteristics of respondents and non-respondents."(Chee/Harris 1993: 63).
Furthermore, bias can occur because there is no guarantee that the addressee himself/herself fills out the questionnaire, it might well be another household member. According to Brusaw/Alred/Oliu (1987: 529) "people who have strong opinions on a subject are more likely to respond to a questionnaire than those who do not. This can slant the results."
In a mail survey the questionnaire must stand on its own. It must be totally self-explanatory because respondents do not have the opportunity to clarify any terms and thereby avoid misunderstandings. "Respondents who incorrectly think they understand a question may supply an invalid response." (Fletcher/Bowers 1990: 123). It is to be expected that above all less-educated persons may have trouble following the instructions and completing the questionnaire. Therefore, it is difficult to obtain complex information in mail questionnaires. A questionnaire asking for such information would have to be lengthy and thus the response rate would decrease. As a consequence, people claim that mail questionnaires are subject to so many errors and biases that the validity of the research is destroyed.
"It follows that self-administered questionnaires should generally be kept as simple, short, and self-explanatory as possible. Instructions should be brief and clear, answer categories unambiguous, and the line of questioning should avoid complicated skip patterns." (Sheatsley 1983: 199).
These general recommendations can be found in virtually all books about survey research.
The purpose of the following table is to show the absolute number and percentages of the different survey methods used in social research in Germany in the period under study.
Table 1: Survey techniques for the total period 1987-1996, number of instances in absolute figures and as a percentage
illustration not visible in this excerpt
Source: FORIS Informationszentrum Sozialwissenschaften,
Bonn/Germany; adapted from Kaase (1999: 284).[5]
[...]
[1] I received these questionnaires by contacting numerous marketing research agencies which sent the questionnaires to me after I had signed an agreement of confidentiality. The classification of the questionnaires into one of the three questioning modes is based on information provided by the research agencies.
[2] Converse and Presser (1986: 55) mention some amusing examples of misinterpretation: profits is taken for prophets; heavy traffic in the neighborhood meaning trucks to investigators and drugs to respondents; family planning signifies birth control to investigators and saving money for vacation to other people.
[3] It is to be noted that a similar communication model has been applied by Shannon &Weaver and can be found in Friedrichs (1990: 193).
[4] This aspect will be further explained in chapter 5.3.
[5] Because many projects contain a mixture of survey methods, multiple mentions were possible. Therefore, the percentages add up to over 100%.
Kommentare