Best Practices: Rituals and Rhetorical Strategies in the "Initial Telephone Contact"

1.

Introduction

2.

The Problem of Refusals

3.

Motives for Refusing

4.

Advice and Contradictions

5.

The Research and Method

6.

The Telephone Initial Contact: Situational Constraints

7.

Interviewers' Pragmatic Tasks, Rituals and Negotiations

7.1

The presentation of selves

7.2

Eligibility

7.3

The negotiation to obtain consent for participation in the later interview

7.4

Discourse regarding the reason and the way the respondents were selected

7.5

The discourse concerning the interview subject

7.6

The negotiation regarding the place and the time for the interview

8.

Re-framing Ethical Issues

9.

Conclusion

Acknowledgements

Notes

References

Author

Citation

Appendix: A Summary Table

Download:

139 kb

Discuss this Article:

Discussion Board

1.

Introduction

The problem of the refusal to be interviewed, or to answer particular questions during the interview (which gives rise to missing data), has long been debated in the survey literature and remains a concern in survey methods. However the traditional solutions to lower the rate of non-response that are suggested in the mainstream quantitative paradigm, e.g. weighting, are of only limited use. An alternative approach is currently emerging. Recently some survey researchers (for instance DE LEEUW 1999) have shown the value of certain interactional and rhetorical tactics which have been implemented by experienced interviewers. Here qualitative methods can actually contribute to identifying these "best practices" by analyzing the conversation between interviewer and respondent during the initial contact. From this perspective, reducing refusal rates could become a common ground for collaboration between survey and socio-linguistic researchers. Until now the need for integration between qualitative and quantitative approaches in the social sciences has long been acknowledged but research practice has rarely met this need. Consequently the two approaches still follow their separate ways and integration seems difficult to achieve. However discourse analysis (CICOUREL 1980; 1982; 1987; CORSARO 1981; GUMPERZ 1982; VAN DIJK 1983, 1985) and conversation analysis (SACKS, SCHEGLOFF & JEFFERSON 1974; SCHENKEIN 1978; ATKINSON & HERITAGE 1984; MAYNARD 1984) can offer a means to improve survey techniques, by identifying the best practices with which researchers may communicate with respondents. Early examples of such collaboration are just emerging (see MAYNARD, HOUTKOOP-STREENSTRA, SCHAFFER & VAN DER ZOUWEN 2001). [1]

2.

The Problem of Refusals

Sociological surveys or intensive face-to-face interviews cannot occur without the consent of respondents. This must usually be obtained over the telephone, by mail, or, now less frequently, by doorstep interaction (face-to-face contact). However, gaining consent is not without obstacles since many respondents are unsure if they want to participate and may resist giving their consent. Refusal is a phenomenon which generally occurs during the initial contact, but it can occur even after the previous consent to the interview: e.g., the respondent can miss the appointment or refuse to answer particular questions (giving rise to so-called missing data). The literature uses the term non-response. There are a variety of kinds of non-response: straightforward refusals, people who are difficult to find, who are never at home, who are absent for varying periods of time, people who dislike being interviewed, those who are ill, people identified for inclusion in the sample but who have died, people who cannot be located because of address error, and respondents who, during the interview, do not answer particular questions or select the "don't know" category. Leaving aside the problem of access failure (due to address errors, potential respondents' lifestyle characteristics and temporary circumstances affecting availability), FITZGERALD and FULLER (1987, p.4) propose the terms refusers for people who clearly refuse to be interviewed, and reluctant or difficult-to-reach for respondents who require an extensive number of call-backs before agreeing to be interviewed. [2]

Non-response rates have been increasing since the 1960s (GOYDER & LEIPER 1985; SCHLEIFER 1986). In academic research in the U.S.A., the contemporary percentage of non-response fluctuates between 20%-33% of sampled respondents (BREHM 1993, p.16); in the 1980s non-response rates in surveys conducted by the SURVEY RESEARCH CENTER (Ann Arbor) have been around 30%, while non-response in surveys by the NATIONAL OPINION RESEARCH CENTER (University of Chicago) has been around 25%. More alarming are the percentages in research by private institutes and commercial agencies in the U.S.A., which have fluctuated between 30%-50% (CRESPI 1988). In Great Britain non-response percentages are around 27-40% (COLLINS, SYKES, WILSON & BLACKSHAW 1988, p.217) and in Italy around 28-49%. To these percentages it is necessary to add a further 10% for telephone surveys in the U.S.A. (BREHM 1993, p.25) and 7-22% in Great Britain (COLLINS et al. 1988). It is known that answering machines have become a new tool for refusing the interview (TUCKEL & FEINBERG 1991; OLDENDICK & LINK 1994; BOSIO 1996, p.40). In fact the SURVEY RESEARCH CENTER (Ann Arbor) has collected percentages ranging between 6% and 42% of the sample (OKSENBERG, COLEMAN & CANNELL 1986, p.98). Unfortunately there is also a contribution to non-response as a result of some populations being "over-surveyed" and thus becoming indifferent to further surveys (STEEH 1981; GROVES & LYBERG 1988; BREHEM 1993, p.17), this method having become increasingly pervasive in both privately-funded and publicly-funded research; for example in the Detroit Area Studies, non-response rates have risen from 12.5% in 1952 to 32% in 1988¹⁾ (BREHEM 1993, p.17). [3]

The effects of non-response appear even at the sampling level: e.g., in the GSS (General Social Survey, University of Chicago) and in the NES surveys (National Election Studies, University of Michigan), young people (i.e. respondents under 30 years old) are underrepresented and the elderly (who are 65 or older) are over-represented (BREHEM 1993, pp.26-28). The percentage of black people in the academic samples may often be over-represented because, believing that blacks are more inclined to refuse the interview, this group is over-sampled but may then prove more responsive than was anticipated (o.c., p.29). Women may be over-represented because they work at home more than men, making women easier to contact for a survey (o.c., p.30). Again the NES and GSS samples over-represent less educated respondents, while the NES also over-represents respondents with college degrees²⁾; this error is very frequent in telephone surveys because it is easier to lie about having a degree (and other "socially desirable" attributes) over the telephone, and also less well-educated respondents are less likely to have access to a telephone (o.c., p.31). Finally the NES samples over-represent poor and rural people, while the GSS under-represents them (o.c., p.33 and p.36). [4]

FITZGERALD and FULLER (1987, pp.7-11), and MARRADI (1989, p.73) maintain that refusals are not randomly distributed but are correlated with precise demographic variables. Refusers are far more likely to live in city centres (where the fear of criminality exists), to be elderly, to have a low income, to be married or separated (while widows, the divorced and single people are more likely to agree to give an interview), and to be residents in high-rise apartments or in duplex, town, or row houses. But earlier research by DE MAIO (1980), who tried to reconstruct the demographic characteristics of 1,262 refusers in a national survey in the U.S.A. carried out in 1977, reaches different conclusions. The refusers were over thirty years old. Among this group, people over fifty were more available to give the interview after a second contact made one month later. People with low incomes were more available to be interviewed. There were no significant differences in regard to race and gender. Respondents living in rural areas were much more approachable and co-operative than people living in urban areas. Different characteristics are present in other countries, for example, among refusals in Poland. LUTYNSKA (1987, p.49), comparing 60 surveys conducted by Polish universities and academies between 1982-1985, found that refusals occurred more often among men than women; people with low education; the elderly; people holding specific political positions and, usually, those in high status professions; urban residents; "special categories" of citizens (intelligentsia, scientific workers, policemen, military, so-called entrepreneurs, and persons with an unsettled legal situation, e.g. those with housing or tax problems). However, it must be noted that ethnographic observations conducted by interviewers (see BOCCUZZI 1985) suggest that methodologists should be skeptical about statistical data on the effect of socio-demographic characteristics on refusals. This approach distrusts the apparent determinism of studies relating demographic characteristics to refusal, and suggests that more attention should be paid to the local context in which refusals occur. [5]

The phenomenon would not be so serious (and could relatively easily be managed) if the 70% who participate in surveys were identical to the 30% who do not. In this case the representativeness of samples would not be so seriously in question. But those who refuse to be interviewed are not a random sub-sample of the sample which has been drawn (MARRADI 1989, pp.73-76). As KISH (1965, p.558) stated, the substitution of non-respondents is often no improvement because substitutes tend to be more similar to respondents than non-respondents. The latter display distributions of attributes which are systematically different on all the main socio-demographic properties from those who participate in surveys (CASTELLAN & HERZEL 1971, p.302; STINCHCOMBE, JONES & SHEATSLEY 1981; GOYDER 1987; MARRADI 1989; BREHEM 1993, p.17). So non-response introduces a non-random element into samples which have been randomly drawn and causes serious bias in univariate statistics such as means, proportions and variance (MOSTELLER 1968, p.120; PLATECK 1977; 1980; KALTON 1983; BREHEM 1993, pp.93-100) and in bivariate and multivariate coefficients (TOBIN 1958; SCHWIRAN & BLAINE 1966; HECKMAN 1976; 1979; PITRONE 1984, p.150; BREHEM 1993, pp.100-106). Several statistical models of coping with non-response have been created. The most common technique is weighting the respondents' answers based upon the refusals' demographic characteristics (GOUDY 1976; PLATECK, SINGH & TREMBLAY 1978; O'NEIL 1979; FITZGERALD & FULLER 1987, pp.7-11; BREHEM 1993, chapters 5 and 6). But this procedure is artificial and often arbitrary (PITRONE 1984, pp.149-150; MARRADI 1989, pp.68-78) because researchers attribute opinions and attitudes to people who never responded, by weighting the answers given by respondents who belong to the same socio-demographic group of non-respondents. The equivalence of the two groups has never been proved. For this reason it is fundamentally important to adhere as much as possible to the randomness of samples by trying to persuade the maximum number of people in the sampling frame to participate (LISSOWSKI 1969; LUTYNSKA 1987, p.46). This directs survey methodology toward an identification and improvement of suitable rhetorical tactics to be used by interviewers to maximize survey response (DE LEEUW 1999). [6]

3.

Motives for Refusing

The creation of suitable and successful rhetorical strategies cannot occur without understanding the motives for refusing. What are the reasons for refusing an interview? It is not easy to have a complete picture because, as DE MAIO warns:³⁾ "several factors prevent adequate quantification of the results—the number of reasons given is large, the concentration in several categories is small, and qualitative differences pervade the reasons offered by refusers" (1980, pp.230f.). Respondents withdraw from the survey even after the interview has started. There are also linguistic features that may intervene in some countries. For instance VIGDERHOUS (1981), analyzing a survey carried out in Canada, found that some refusals were caused by respondents' inability to properly speak the interviewer's language. SMITH's research also seems discouraging in regard to dealing with refusals. After having used nine different techniques (1983, p.51) in order to study 315 refusals in a US national sample survey, he sums up:

we come close to the conclusion that nothing works in estimating non-response bias. Each of the methods we examined proved to be of limited usefulness. (...) In sum, our analysis of non-response on the 1980 GSS suggests that there is no simple, general, accurate way of measuring non-response bias (o.c., pp.65f.). [7]

By taking into account respondents' motivations we can see that reasons for refusals by respondents may include: language problems, lack of time or incompatibility of their schedule with responding to a survey (CONVERSE & SCHUMAN 1974, p.41), distrust of the interviewer, whom respondents may suspect to be a door-to-door sales person⁴⁾ or even a predator; lack of interest in the topic (CONVERSE & SCHUMAN 1974, p.41; BOCCUZZI 1985; SENF 1987; BREHEM 1993, p.53); aversion to sociological surveys, disbelief in their anonymity (SHARP & FRANKEL 1983, p.43; LUTYNSKA 1987), or a diffuse fear of crime (CONVERSE & SCHUMAN 1974, p.41; BOCCUZZI 1985, p.243; BREHEM 1993, p.52), feeling unable to successfully complete the task of responding to a survey; not being used to offering a personal contribution to collective progress, even less so to cultural progress (BOCCUZZI 1985, pp.246f.), previous negative interview experiences; concerns about privacy (DE MAIO 1980)⁵⁾; prohibition by relatives⁶⁾, and the interviewer's unprofessional performance. Refusals to participate in interviews, respondents' failure to show up for appointments, or their simulated absence when the interviewer arrives at their home, can all be caused by events that occurred during the initial contact. [8]

Even though the stage where refusals arise is one of the most important steps in the whole research process, few handbooks [CONVERSE & SCHUMAN 1974, pp.36-45; SURVEY RESEARCH CENTER (Ann Arbor) 1989, pp.7-9; SINGER & PRESSER 1989; SURVEY RESEARCH CENTER (Berkeley) 1990, pp.29-30; FOWLER & MANGIONE 1990, pp.56-58; MORTON-WILLIAMS 1993] and articles discuss in detail the "process of the initial contact". Further, the advice reported in survey handbooks in order to manage the initial contact is often unrealistic and contradictory (see section 4). Moreover, the existing literature does not examine the contribution that discourse and conversation analysis can make. This latter topic is the main focus of this article (see section 6). [9]

4.

Advice and Contradictions

The following reflections are connected with a general problem already pinpointed by ethnomethodological studies on social rules (GARFINKEL 1963, CICOUREL 1964; WIEDER 1974) about the gap between instructions or professional norms (in this case set down in interviewing handbooks) and local interactional constraints which make it impossible to consistently apply such norms. Paradoxically, if an interviewer abides strictly by the instructions in such manuals (e.g., that interviewers must have flexible schedules in order to make appointments at any time that is convenient to respondents, that they should explain the aim of the interview⁷⁾, wait for the respondent to set a date for the interview, be precise about how long the interview will take, on request describe the specific questions that will be asked, always tell respondents the whole truth regarding the procedures and reasons for their selection, etc.) s/he may often fail to obtain the interview. [10]

Different advice has been elaborated about how to manage the initial contact, particularly in order to deal with respondents claims and questions. Unfortunately these suggestions are often unrealistic, contradictory and confuse the interviewer. Besides, like the questionnaire itself, they are the result of a positivist conception of the interview, which considers the respondent as a data-bank (BOKSZANSKI & PIOTROWSKI 1980, p.46; BOCCUZZI 1985) and the interviewer as an automaton (CICOUREL 1964, p.90; DEUTSCHER 1972, p.325; LAVRAKAS 1987, p.112). Respondents' different cultures and cognitive schemata, like interviewers' attitudes, subjectivity and fears, have little place in the handbooks of survey method. [11]

There is not always consistency in the practical advice and norms offered by handbooks⁸⁾ about the interviewer's ideal behavior during the "initial contact". DILLMAN, GALLEGOS and FREY (1976) analyze the experimental results from a telephone survey, where one group of interviewers did a systematic introduction (using the respondent's first name, describing the survey, sampling techniques, the length of the interview, the organization carrying out the survey), and a second group where the interviewers introduced themselves spontaneously. DILLMAN and colleagues indicate that systematic introductions did not significantly reduce the number of refusals. O'NEIL (1979) reaches a similar conclusion. Let us now look at the list of the main advice we can find in handbooks of interviewing. [12]

Regarding respondents' clarification requests the SURVEY RESEARCH CENTER (Berkeley) advises "don't say too much because you may lose your respondent (...) suggesting areas of concern which had not occurred to the respondent" (1990, p.29). The SURVEY RESEARCH CENTER (Ann Arbor) advises "not to be too specific about the interview in introducing yourself and the survey (...) and use very general statements" (1976, p.7). In contrast FOWLER and MANGIONE suggest "give respondents an opportunity to ask the questions on their minds, so that respondents do not go into interviews with less information than they want" (1990, p.74). CANNELL, FOWLER and MARQUIS (1968) advise that, when interviewing well-educated respondents, interviewers should communicate more information about the research in order to enhance the quality of responses. [13]

Regarding questions about how the respondent has been selected the SURVEY RESEARCH CENTER (Berkeley) suggests a brief response based upon the topic of random selection. The SURVEY RESEARCH CENTER (Ann Arbor) recommends another explanation stating that a random or cross-section sample is necessary because it is not possible to interview everybody (1976, p.8; 1983, p.315). [14]

If the respondent claims to be busy DILLMAN suggests accepting without objection the respondent's answer and asking for another telephonic contact (1978, p.262)⁹⁾. The SURVEY RESEARCH CENTER (Ann Arbor) writes

you should assume the respondent is not busy and approach the meeting as though the interview were going to take place right then—at the time of contact. (...) Avoid questions such as 'Are you busy now?' or 'Could I take this interview now?' or 'Should I come back?' Questions which permit undesired responses can lead or even push a respondent into refusing to be interviewed (1976, p.7). [15]

The same opinion is shared by BAILEY who states that the interviewer should never allude to the possibility of adjourning the interview, giving up only if the respondent is really not in the condition to answer and asks the interviewer to come back later (1978, p.220). Few handbooks or other sources of guidance actually deal with handling both those respondents who claim to be busy and those who seem not to be busy when contacted. [16]

Regarding the duration of the interview, the SURVEY RESEARCH CENTER (Berkeley) suggests a vague answer, such as "That's hard to say because it varies depending on how much you have to tell us, but it usually takes less than half an hour" (for a telephone interview) (1990, p.30). However, the SURVEY RESEARCH CENTER (Ann Arbor) recommends: "Always be honest about the length of the interview" (1983, p.318). As CONVERSE and SCHUMAN wisely emphasize: "truth telling about the interview time is complex (...) There is a tradeoff between honesty and practicality" (1974, p.42). [17]

In relation to the usefulness of the surveys the SURVEY RESEARCH CENTER (Ann Arbor) suggests telling respondents that surveys are used to formulate more effective governmental policies and to improve the quality of life (1976, p.8). Regarding the respondent's concern about the privacy of answers, the SURVEY RESEARCH CENTER (Berkeley) suggests saying: "We take our promise of confidentiality very seriously. Your name will never be connected with any of the answers you give me" (1990, p.30). This kind of answer may be satisfying to those respondents who value expressions of trust but may be less satisfying to people who are more suspicious. If respondents appear to be concerned that the interview or questionnaire may compromise her/his privacy, the interviewer may say: "Look, we'll do it like this: I'll come to your place and we'll start the interview; if you feel that some questions are too personal, then you don't have to answer them. You answer only what you want to ..." [a similar tactic is suggested by SURVEY RESEARCH CENTER—Ann Arbor (1983, p.318) and SURVEY RESEARCH CENTER—Berkeley (1990, p.29)]. This suggestion contradicts the pressure that research project leaders usually apply to interviewers to reduce missing data in questionnaires. Consequently the interviewer needs to solve the problem with other rhetorical strategies. [18]

5.

The Research and Method

The present qualitative study has been included within a larger survey with young respondents about the attitudes of juveniles towards drug consumption and issues relating to criminality. A sub-set of the sample was constituted by teenagers who had criminal records. Social workers of the Italian Department of Juvenile Justice provided the researchers with a list of these respondents. The list was given on the condition that the interviewers were not to reveal to the respondents how they had been selected. The social workers were concerned that these young people should not feel that their lives were being watched beyond the normal two year period after their release from prison. The author has analyzed ten initial telephone contacts made with this sub-sample of juvenile respondents (resident in a northeastern Italian town) which were made by one experienced, male interviewer. All respondents were contacted in order to arrange a later face-to-face interview. [19]

The initial telephone contacts have been tape-recorded and the transcripts analyzed following the criteria of "discourse analysis", collecting ethnographic and contextual descriptions of each initial contact, in accordance with the procedures commended by CORSARO (1985) and CICOUREL (1987). Three contextual features of the data have particularly been observed:

• resources = elements used by the interviewer and respondents to sustain the initial telephone contact

• constraints = elements which could have limited the range of responses at speakers' disposal

• effects = some consequences that resources and constraints could have had on actions [20]

The main aim of the research is to study the telephone interaction in the initial contact. This interaction, which is usually quite brief, assumes importance because it is most often in this step that refusals happen. Notwithstanding the importance of this interactional step, no socio-linguistic studies (as far as I know) have been devoted to the initial telephone contact (although aspects of this process are handled from a conversation-analytic perspective in SUCHMAN & JORDAN 1990 and from a practical perspective in SMIT & DIJKSTRA 1991). [21]

The findings of the present study can be generalized under some conditions and with some caveats. As with most qualitative research, the sample is not statistically representative. Nevertheless such work can make a considerable contribution to our knowledge of social processes, including those at work in various kinds of data collection. For instance GARFINKEL (1962; 1964) did two seminal discourse studies with small samples. In the former he analyzed protocols relating to verbal exchanges between an experimenter and ten undergraduates. In the latter study he analyzed twenty-five verbal exchanges collected by twenty-three of his undergraduates during the famous "breaching studies". CICOUREL (1982) studied the discourses of three medical visits between a male doctor and a female patient. VAN DIJK (1983) analyzed the conversation between an interviewer and two respondents (husband and wife) showing cognitive and discursive strategies in reasoning related to prejudice on the basis of ethnicity. The same sampling procedure has been followed by Erving GOFFMAN and by conversational analysts. In certain other disciplines, such as cognitive science, ethology, archeology, geology, and psychoanalysis, statistically-representative samples are quite rare. [22]

The aim of the socio-linguistic studies in the ethnomethodological discourse analysis and conversation analysis tradition is not to estimate a characteristic from the sample to the population or quantify the percentage of a shared attitude in the population, but to observe a recursive behavior or phenomenon, and find relations among variables. As PERÄKYLÄ, concluding his research on the relationship between a psychotherapist and a patient suffering from AIDS, says,

The results were not generalizable as descriptions of what other counselors or other professionals do with their clients; but they were generalizable as descriptions of what any counselor or other professional, with his or her clients, can do, given that he or she has the same array of interactional competencies as the participants of the AIDS counseling session have (1997, p.216). [23]

It should also be noted that, in evaluating research, the representativeness of the sample is only one of the criteria generally adopted. Other criteria are: reliability of method, validity of findings, comprehensiveness of data treatment, the accuracy of researchers, thick description, completeness (MILES & HUBERMAN 1994, p.279), saturation of categories (GLASER & STRAUSS, 1967), authenticity of description, consistency of statements (HAMMERSLEY 1992, p.67), credibility (WILSON 1989, p.27; HAMMERSLEY 1990, p.61), and plausibility of theories (HAMMERSLEY 1990, p.62). In that the evaluation of findings is grounded on multiple criteria, representativeness on its own is, of course, no automatic guarantee of generalizability. These are methodologically separate issues: representativeness is concerned with sampling considerations and generalizability is concerned with findings (GOBO, 2000). It is quite possible for a researcher to conduct a study on a representative sample but produce findings which cannot be generalized. This may happen for a number of practical reasons: the different reliability of different methods (e.g., interview versus focus group versus ethnography); sloppy data collection; researcher's biases in data analysis; unsuccessful access and relations in the field, as well as the ecological validity of collected data. [24]

The conclusions of the qualitative research described in this article are consistent with a recent quantitative study on twenty-two very experienced female CAPI interviewers conducted by DE LEEUW (1999), who shows that the most successful strategy to combat non-response is "tailoring the introduction" that is "grasping the doorstep situation, modify the introduction according to [respondent's] social and cultural class, and do not follow a set of fixed rules, adapt to the situation" (p.33). The present study shares DE LEEUW's emphasis on introductory strategies as the main way to reduce refusals but uses a different source of information about these interactions and strategies. DE LEEUW's account is based upon interviewers' accounts expressed in a debriefing. These accounts are very important but not always valid in reconstructing the basic rules (CICOUREL 1970) of interviewers' behavior, which are often unconscious to them. Discourse and conversation analysis are useful methods for examining this aspect. [25]

6.

The Telephone Initial Contact: Situational Constraints

Telephone calls are sui generis social situations. They are different from face-to-face interactions, even if they seem to reproduce some of the same moves. The "face" (GOFFMAN 1955) is a resource that interviewer and respondent will show on the day of the interview. So it has to be presented by the interviewer in the best possible way over the phone. Let us look at some characteristics:

• Speakers do not know each other. While the interviewer sometimes knows something about the respondent, the latter doesn't know anything about the interviewer except what s/he can infer through the voice and conversation (age, gender, etc.).

• Speakers do not see each other's face¹⁰⁾. They do not take into account their facial expressions or their clothes. It is possible that young respondents have to modify her/his discourse because parents can hear the conversation.

• Much more than in the face-to face situation, the speaker has to be aware of tone of voice, verbal expression, type of respiration and hesitations. Using the telephone requires the interviewer to rely only upon her/his voice and words to persuade the respondent. Respondents can evaluate only what the interviewer says and how s/he speaks. In contrast to other types of interaction, conversation plays a primary role because participants do not have information from media other than voice and pauses. [26]

7.

Interviewer's Pragmatic Tasks, Rituals and Negotiations

Before explaining to the respondent the aims of the interview, motivating him/her to participate and to be precise about the accounts that the interviewer seeks to elicit (the commitment procedure of MILLER & CANNELL 1982, p.308)¹¹⁾, and giving instructions about how to answer appropriately, or giving feedback to the answers, the interviewer should pay attention to some basic aims which structure the teleological context of rhetorical strategies. Before each telephone call the interviewer sets out to do the following:

1. control the eligibility of the respondent contacted;

2. obtain consent for participation in the later interview (if possible with a precise date and time), enlisting a kind of co-operation (FOWLER & MANGIONE 1990, p.55);

3. introduce her/himself, giving a good impression and generating trust and respect (FOWLER & MANGIONE 1990, p.64) so s/he can prepare the ground for the face-to-face encounter;

4. reassure the respondent about the ease of the interview;

5. assure the respondent that their privacy will be maintained;

6. in some cases, avoid revealing the source that furnished the respondent's address¹²⁾. [27]

These requirements on the interviewer give rise to a set of stages, each of which can be regarded as being conducted through particular ritual forms. In other words, features such as the interviewer's and respondents' presentations of self, the control of eligibility of the person contacted, the negotiation to obtain consent for participation in the later interview, discourse regarding the reasons why and how respondents were selected, discussion about the interview topic, and the negotiation regarding the place and time for the interview encounter, each presents particular discourse analytic characteristics. During the display of rituals the interviewer employs a sequence of rhetorical strategies in order to persuade people to become respondents. Examining these rituals identifies ways that discourse and conversation analysis of initial contact can help survey researchers to improve this important step and thus to deal with refusals. [28]

7.1

The presentation of selves

The first ritual aims to introduce the interaction between the two participants. Comparing the results of the sample of ten initial telephone contacts, the ritual of presentation generally involves the following moves:

1. The interviewer waits until the respondent gives an indication that thecommunication can start (usually with the signal "Hello!");

2. the interviewer repeats the respondent's utterance ("Hello");

3. then the interviewer pronounces his first name

4. then the interviewer says "I'm looking for ...";

5. naming the respondent's first name (that person whom the interviewer wants to interview). Often the interviewer does not immediately get through to the respondent but to a parent or brother or sister, so the interviewer must wait a few seconds¹³⁾; 

6. sometimes the interviewer says "thank you" to the person who calls the respondent;

7. while the interviewer waits or before starting the telephone call, the interviewer clears his throat. [29]

When the respondent arrives at the phone the interviewer continues the ritual in this way:

1. After the respondent gives a signal of presence ("yes", "hello") the interviewer speaks the respondent's first name, confirming that the selected respondent is really at the telephone;

2. after the respondent's confirmation, the interviewer says "hi";

3. then the interviewer repeats move 3 pronouncing his first name:

4. sometimes the respondent does not wait for the interviewer's explanation for the call. So s/he asks "Who?" or "Who are you?". Then the interviewer answers "You don't know me (attempting a smile), now I'll tell you ...";

5. then the interviewer starts an introduction about his profession "... I'm working for the University of Trento ...",

6. and the interviewer introduces the reason for the telephone call "... and I'm doing research on how young people view the drugs phenomenon". [30]

The friend format

Looking at the four initial contacts which began with a conversation with respondents' parents or relatives, we find that through moves 3, 4 and 5, in order to get through as quickly as possible to the designated respondent, the interviewer seeks to pass himself off as the respondent's friend. Looking at the remaining six initial contacts, when the respondent answers straight away, the interviewer tries to create a friendly contact, thus reducing the role distance and, eventually, status distances, while not upsetting the respondent with too much formality. [31]

These particulars are not trivial. We must keep in mind that one of the interviewer's goals is to reach the respondent as quickly as possible, trying to avoid the parents or anyone else who might impede the respondent's interview. Parents' reluctance about the participation of their son or daughter in an interview can be higher with surveys on sensitive topics, e.g. drugs: parents may think that the interviewer could be a threat to their children. Therefore, the interviewer must devise a strategy for obtaining the consent to talk to the respondents without entering into overly complicated negotiations with parents or relatives. If conducted properly, this strategy (which we can define as "simulation of a phone call from a friend") succeeds in its aim to deceive the parent (for ethical issues related to this strategy, see below). The interviewer generally follows a strategy intended to give the respondent's parent or relative some proof of his knowledge of the customs of politeness as distributed in northern Italy. For this reason, if the parents answer the phone, the interviewer introduces a shy formal greeting, e.g., "good morning", between moves 2 and 3. The tone of voice, the greeting, and eventually the conditional tense "Could I ...", would have been enough (in the interviewer's reasoning) to overcome the "parent obstacle". When the parent tries to resist (without insisting) the interviewer responds as follows:

1 Par.¹⁴⁾ Hello
2 INT       Hello, Good morning I'm Peter
3              I was looking for Sabrina
4 Par.      Who are you (tu)?
5 INT        It's Peter
6 Par.      (after a two second pause) Wait a moment [32]

The interviewer repeats his first name as if to say "I'm your daughter's friend. You've never met me". The interviewer is aided by the parent, since he received the familiar "tu" (you) rather than the third person formal expression "lei" which is usually used for adults in the Italian language. [33]

Thus the interviewer acts as if two types of appropriate rituals exist in this situation: the friendly one (reserved for respondents and their sisters and brothers) and the asymmetrical one (reserved for respondent's parents). The asymmetrical ritual with the respondent is considered inappropriate by the interviewer (in order to achieve interlocutive effects, such as to seem nice, to obtain consent, etc., that the interviewer wants to achieve). In fact, when the interviewer used the asymmetrical ritual with the respondent Sabrina and she answered "speaking", the interviewer became a little embarrassed and said "Ah ciao" in a subdued tone of voice as if he made a mistake. In this case the interviewer realized that he had mistaken the identity of the speaker. [34]

With move 11, the interviewer tries to anticipate the respondent's embarrassment (GOFFMAN 1956) about speaking with a person that he does not know. When the interviewer speaks, he uses a reassuring tone of voice to calm the respondent. However, it sometimes happens that the interviewer is embarrassed about the respondent's excessive concern expressed when, instead of waiting for the interviewer's explanation after hearing the interviewer's name, the respondent immediately asks, "Who are you?" In such cases the mild embarrassment occurs because the respondent seems to be unaware of the convention that a stranger (the interviewer) should introduce her/himself while the respondent politely waits for the interviewer's presentation. [35]

7.2

Eligibility

Connected with the previous ritual, the eligibility ritual aims to check whether the person who answers the phone corresponds to the selected respondent. The ritual of eligibility follows the requirements of research methods and sampling. There are two alternative techniques to select respondents. In academic surveys the researcher usually draws a precise respondent from lists; in commercial and market research the research leader usually issues the interviewers with a set of socio-demographic characteristics which must be achieved in a pre-determined proportion. In the approach called random digit dialing, the computer randomly selects telephone numbers and interviewers must ensure that those answering when these numbers are called have characteristics which fit those required to achieve the sampling proportions which have been set. In the present research the first alternative has been followed, where the interviewer looks for a precise respondent. [36]

7.3

The negotiation to obtain consent for participation in the later interview

Many respondents state that the main reason they avoid being interviewed (or express regret that they cannot participate) is that they have no time because of school (homework), jobs, or hobbies. For the interviewer, who is trying to obtain consent through persuasive techniques, the problem has at least three aspects:

1. Obtaining consent for the interview (telling respondent that it will not take a lot of time);

2. alleviating any embarrassment experienced by the respondent in trying to justify his or her reason for not responding at that time or which the respondent may feel when trying to find a free day on the spur of the moment (thinking about this over the phone can be quite embarrassing);

3. preventing themselves from feeling embarrassed about seeming insistent and pedantic. [37]

If the amount of time required for the interview seems to be the main problem, the interviewer will offer to help the respondent to find the time:

... it's not urgent, ... we can easily plan for it to take place in 5 or 6 days' time ... when it's fine for you, we can meet maybe next week and find a day that is better for you, when you're not busy. [38]

The interviewer repeats such comments to all reluctant respondents. [39]

After easing the respondent's embarrassment, the interviewer suddenly asks the respondent "Do you want me to call you back in a few days?", "I'll call you in ...". The strategy used by the interviewer, perhaps unintentionally, is to let the respondent seem to decide when to meet. Actually, the respondent only decides within the time limits that the interviewer has proposed (e.g. "next week ..."). So the respondent is only partially free to decide, because if s/he says "next month," the interviewer will remove the respondent's name from the list. [40]

Persuading reluctant respondents

Another reason sometimes offered by respondents to avoid the interview is a resolute "I don't care!" There is no possibility of changing a respondent's mind when faced with such a determined reply. However, in the present research, during the course of the telephone calls, only one respondent refused to be interviewed. This may mean that the tactics and strategies were successful or that the interviewer was lucky in finding available respondents. It is hard to say without additional research. Nevertheless, analyzing an unsuccessful strategy can be useful in discovering an interviewer's mistakes or proposing alternative tactics. Here is the transcript:

  1 

R

If I'm not available?

  2 

INT

Bah

breathing while forming

  3 

a half smile smiling

  4 

R

if you're not available we wonŽt do anything ... ha

  5 

INT

So then ... find someone else

waiting for R taking the floor

  6 

R

because I don't care ...

  7 

Also I discussed a lot in the past ...

  8 

INT

So if you've already discussed it a lot

  9 

you probably have many things to say, don't you?

smiling

10 

R

No, look, I'm fed up ... to the teeth

11 

about this kind of thing

12 

INT

I see ... so you don't want ...

13 

Absolutely not?

14 

R

There are many things to say ...

15 

INT

To say? What do you mean?

16 

R

No so to say ... because everybody has his own opinion, hasn't he?

17 

INT

Hum

18 

R

And then so ... (unclear) yet

19 

INT

20 

R

21 

INT

22 

is an interview ... only

23 

we won't discuss you and me ...

24 

R

25 

INT

OK!

26 

Thank you anyway

smiling upset

27 

R

You're welcome

a little bothered

28 

INT

Ciao

29 

R

Ciao [41]

The interviewer's strategy¹⁵⁾ can be divided into seven steps:

1. Trying to be polite with the respondent without seeming to be insistent. The interactional resource used here is smiling (line 3);

2. having the respondent explain the reason for his refusal (line 5). During his explanation, the interviewer may be able to find clues to counter the respondent's argument;

3. refuting the respondent's reasons. In line 8 the interviewer inverts the respondent's objection, offering a good reason to do the interview;

4. asking the respondent to consciously confirm his refusal (line 12). The interviewer hopes that, by accepting this move, the respondent will change his initial refusal;

5. emphasizing the respondent's value. The interviewer underlines that the respondent's contribution would be important because on other occasions he has already discussed the topic of the interview (line 9);

6. repeating respondent's words ["available" (line 3), "a lot" (line 8), "to say" (line 15)] in order to tune in to his language. This strategy has also been pointed out by MAYNARD (1992) in analyzing talk between doctor and patient.

7. explaining an interviewer's professional norm that states, "Don't judge respondents' remarks" (line 23). [42]

To be successful in gaining an interview perhaps the interviewer should have persevered (line 3) and, taking leave of the respondent, say that he will call him again after a week, hoping that during this period his non-cooperative attitude would change. Although the strategy the interviewer used was unsuccessful on this occasion, it reveals some interesting moves that can be adopted with less stubborn respondents. [43]

7.4

Discourse regarding the reason and the way the respondents were selected

Respondents sometimes ask why or how they were selected. In trying to avoid making the respondents suspicious, the interviewer explains the aims of the research and the sampling procedures in ways which are understandable by respondents. In doing so the interviewer may hide some background information or even make something up. That is, the interviewer "constructs lies". In trying to avoid upsetting respondents with past criminal records or making them suspicious, while maintaining the access agreement with the social workers who provided the respondent lists, the interviewer resolves the dilemma by lying to the respondents. The following excerpt is from the initial contact with one respondent. At the question about how he was selected, the interviewer answers:

  1 they are lists
  2 that were given to me ...
  3 look ... honestly I don't know
  4 probably I think
  5 or they are lists from schools or
  6 lists from job centers
  7 and they gather these lists throughout Italy
  8 and then they choose ...
  9 choose names at random
10 and they give them to the interviewers
11 they gave me a list of people
12 and now I'm checking
13 because someone says no
14 someone, instead, says yes
15 and so on ... [44]

The interviewer uses at least eight types of resources to construct such lies:

1. the word "list/s" (lines 1, 5 and 11) is used to intimate that the respondent is not the only person chosen but belongs to a larger sample;

2. by the use of the third person plural (line 2) the interviewer diverts from himself the responsibility for choosing the respondent and shifts it to unknown people;

3. the use of the adverb "honestly" (line 3) acts as a substitute for "Look I don't want to think up lies about how I got it, and it's much easier for me to tell you I don't know. If I knew, I would tell you". So the interviewer appeals to the respondent's tact, asking him to trust that the interviewer is really telling the truth;

4. the interviewer says "probably" and "I think" (line 4) which suggests he wants to be as honest as possible, as if to say: "Look, I will make an effort to give you an explanation, but I'm not positive about it. So take it with certain reservations because I don't want to mislead you";

5. to show the transparency of the procedure and to prove there are no ulterior motives, the interviewer mentions public institutions, such as the university (lines 5-6), instead of private institutions, or people who know the respondent, or, more threatening still for the respondent, particular institutions (police, social work department, etc.)¹⁶⁾;

6. the interviewer mentions the name of the country (line 7) to suggest that the respondent is part of a wide-ranging national sample, indicating that the research does not only focus on him;

7. the chance nature of the call ("choose names at random", line 9) is mentioned to strengthen the move described in f). The randomness involved is a highly familiar and reassuring factor because it reminds respondents of well-known prize shows and lotteries. In other telephone calls, not belonging to the sample of this study, the interviewer also used the word "computer". This word seems particularly effective in persuading respondents because of its relation to the idea (however mythical) of the impartiality of computers. Hence, it appears that the machine, not a person, is choosing the respondents;

8. the interviewer mentions an action ("they gave them to the interviewers", line 10) usually reserved for inanimate objects. Strengthening the two prior moves, this suggests that the respondent has been chosen as a potentially worthwhile research subject, not as a particular person. [45]

To sum up, paradoxically the lie succeeds by virtue of the request for trust. In other words just when the interviewer says to the respondent, "Believe me, I'm telling you the truth," the interviewer is telling him a lie (for relevant ethical issues see below). [46]

7.5

The discourse concerning the interview subject

Some respondents are not interested in knowing the subject of the interview, so they do not ask any questions. The interviewer usually talks briefly about the subject. The descriptions he uses include: "... how you see the drug phenomenon, what you think and so on" or even "... what your friends think, what they know, if they are informed, your opinion, your beliefs ...". The interviewer utters these words in close sequence, without hesitation, like something learned by heart. Perhaps this is because the interviewer is experienced and has already made a lot of telephone calls. However, if so it is hard to understand why other parts of the ritual are not performed as fluently. In this segment the interviewer repeats words like "know", "think", "are informed", "newspapers", "opinion", "believe". The effect that the interviewer wants to obtain with this stream of words is to impress the respondent in a positive way so the respondent believes s/he is speaking with a well-prepared person. Such language also simplifies the meaning of the interview. The interviewer often uses the word "problem" instead of the more neutral and correct "phenomenon", not because he is thinking that drugs could really be a problem, but because by using a word so familiar to the respondent, the respondent will have a rough idea about the topic of the conversation. The use of these journalistic and common-sense terms may have a double edge: these terms may be successful with some respondents while putting off others and making them refuse the interview. [47]

Sometimes the interviewer seems embarrassed to explain the subject of the interview or the topic of research to respondents¹⁷⁾. For example, he seems somewhat embarrassed to utter the word "drug". While talking with one respondent, the interviewer says: "... a study on ... (hesitation) on how the young people see the drug phenomenon" and then "an interview about how you see ... this ... problem of drugs ... what ... what you think about ...". [48]

Here the interviewer's hesitations could have a double meaning¹⁸⁾:

1. He could be worried that the respondent could become upset, i.e. the respondent could be uncomfortable because the interviewer apparently connects her/him to this topic ("Why me? What does he want to know? Will my parents find out ..., etc.);

2. the interviewer could feel a little ashamed to be working on this topic because it is a topic which is in vogue, taken for granted, and even too journalistic. Thus, self-critical doubt seems to betray the interviewer just when he should believe in what he is doing in order to persuade the respondent. These are, however, only hypotheses because it has been not possible to interview the interviewer. [49]

7.6

The negotiation regarding the place and the time for the interview

When the interviewer brings up the question of where the interview should be held, he proposes two alternatives: "Do you want me to come to your home or would you prefer somewhere else ... some other location, at a bar ... as you like ... wherever you feel most comfortable ..." Sometimes an answer is not given right away. The respondents seemed a little surprised by this proposal, maybe because it raises issues for them such as "what kind of questions will he ask me if he's now asking me this ... maybe my parents shouldn't hear? Will he embarrass me?". The "place question" seems to produce the opposite effect from that which was intended: instead of making the respondent feel at ease, it makes her/him suspicious. Most respondents probably choose their home for comfort and security. Home becomes a means of defending oneself (especially for female respondents) from an unknown person. For example, one female respondent said that, some days after the initial contact, she felt some regret about having too quickly consented to the interview. Friends of hers had stated that they would never give an interview to a stranger. [50]

8.

Re-Framing of Ethical Issues

An aspect of the initial contact which is often neglected because censure is attached to it is the case where the interviewer, in order to get consent to the interview, cannot be entirely truthful with the respondent or strictly ethical in the way proposed by the ethical codes of professional associations. This eventuality plays a role in structuring the discourse of the initial contact in a way which deviates from codes of ethics. For instance, concerning the source of respondent's names and addresses (see section 7.4) the professional norm would have been to "explain how the respondent has been selected [...] that he has been reached through an impersonal way, simply because it is necessary to sample a cross-national section of the population" [SURVEY RESEARCH CENTER—Ann Arbor 1969, pp.2f.]. [51]

However, sometimes interviewers are compelled to hide information from respondents. Particular examples include:

• In some studies (e.g. surveys of alcoholics, drug addicts, people with mental diseases, handicapped people, people who have been in prison, those who have received traffic tickets, etc.) there is non-random sampling and it is necessary to conceal how respondents are identified in order not to upset them;

• in many field studies concerning the institutional reasoning of lawyers, judges, politicians, policemen, etc., it is not always possible to reveal in detail the purpose of the research because it may challenge the respondent's attitudes¹⁹⁾;

• in field studies using covert observational methods;

• in initial contact with respondents, interviewers must often avoid detailed discussion about sensitive topics and questions to be asked in the interview. [52]

In this research, as already mentioned, to gain access to the names of these juvenile respondents the researcher made a promise to the social workers. Consequently the interviewers had to face a dilemma: either to tell the respondents a lie or to break the agreement with the social workers. In either case they would be out of step with the professional ethical codes laid down for interviewers. The ethical codes of the American Sociological Association and American Association for Public Opinion Research explicitly condemn such a behavior²⁰⁾. But it is easy to be an armchair critic, when we know that researchers and interviewers are sometimes compelled not to tell the whole truth in order to avoid upsetting the respondent who may, for this reason, refuse to participate in the survey. Besides, as HOLDAWAY stated, codes adopted by professional associations "deal with predictable and planned research, conditions which are not present in fieldwork" (1982, p.66). Consequently a balance must be reached in each case (ERIKSON 1967). [53]

If we consider ethical issues only in the standardized way embraced by the codes of professional ethics, there are a number of researcher behaviors which could be considered, in a strict sense, unethical. Among them, covert observation is the most well-known. To the standardized and rigid conception of research ethics has been opposed the concept of "situation ethics" (FLETCHER 1966; DUSTER et al. 1979). The latter asserts that, in deciding if a course of research action is morally right or wrong, we need to evaluate several contextual features, such as the aims of the study, the type of social actors observed, the consequences of the researcher's actions, and so on. Some authors argue that covert research can be regarded as ethically justifiable when it is conducted on so-called "powerful groups", but unethical when conducted on powerless ones, particularly because "the poor, powerless and disreputable seldom complain about the studies published about them ... because they are seldom organized enough to do so" (BECKER 1964, quoted in FIELDING 1982, p.94). However the same group, depending on the perspective adopted, can be regarded either as powerful or as powerless. For instance, in his study of the National Front, Nigel FIELDING states that members of such a party "are both underdogs and repressors" (1982, p.92). In the context of organization studies another perspective maintains that covert research is ethical when the social actor observed plays a public/civil function or service for users, customers and clients. Policemen, civil servants, doctors, nurses, and so on play a public role and are expected to adopt a client-oriented or customer-oriented approach. From this perspective ROSENHAN's (1973) well-known study in psychiatric clinics has some justification. Another reason for ethical dubious practice, which fits also with David ROSENHAN's study, is the gain to scientific knowledge as a result of the research being carried out. Unfortunately it is not always clear who evaluates the importance of the findings. As a matter of fact researchers, agencies, the social science community, and those social actors who have been observed, may have different views and interests. [54]

In the present study the main criterion followed by the author was the obligation not to cause undeserved harm to social actors (see WALLIS 1977). For example, while an element of deception was employed in seeking to circumvent parents, this would be no greater than in other circumstances such as where a friend-of-a-friend obtains the respondent's telephone number but presents themselves as an actual acquaintance when asking the parent to hand the phone to their child. In fact, similar deception is sometimes used in market research. In both cases the respondent retains the right to terminate the call if she or he is concerned at the evasion of parental control or is for any other reason unwilling to continue the call. Another additional criterion is that of avoiding the publication of sensitive material. As BECKER maintains, "one should refrain from publishing items of fact or conclusions that are not necessary to one's argument or that would cause suffering out of proportion to the scientific gain of making them public" (1964, quoted in FIELDING 1982, p.91). The justification for not asking respondents for their permission to tape-record the telephone conversation and for not being entirely truthful about how their names were obtained, and the interviewer's justification for supposedly unethical tactics such as eluding the control of respondents' parents, lying about the source of respondents' names, and imputing to reluctant respondents a responsibility for the loss of their opinion from the research findings, relies on the fact that the consequences of all these actions for the respondents were minimal and their privacy was preserved. [55]

9.

Conclusion

This study has attempted to show how discourse analysis and conversation analysis can be used to identify interviewer strategies and thus contribute to the improvement of survey methods. Through a sociolinguistic analysis of the telephone initial contact and, in those cases where it occurred, respondents' actual opposition to consenting to an interview, it is possible to identify suitable rhetorical strategies in order to improve the initial contact and consequently reduce refusals. The study of dialogue between interviewer and respondent also represents an area of co-operative work that links qualitative and quantitative methodologists. Further, it is important to understand that refusals to participate in interviews, respondents' failure to show up for appointments, their simulated absence when the interviewer arrives at their home, can all be caused not only by respondents negative attitude towards the survey but also by tacit fears, distrust and culturally-based misunderstandings. Nonresponse should not be labeled only as psychological resistance, symptoms of rudeness, bad faith or fears, but as cultural responses to the researchers' requests and to the intrusive elements of contemporary sociological research methods. A socio-linguistic analysis can reveal these sociological patterns and offer the basis for designing new rhetorical strategies which take into account respondents' stereotypes and prejudices. Even the SURVEY RESEARCH CENTER (Ann Arbor) emphasizes "the answers to these questions should be modified depending on the specific concerns of particular respondents. Listen carefully to the informant/respondent. The answers to some of these questions change from study to study" (1983, p.316) and "don't read someone else's introduction; use your own words" (p.311) or "you should answer the respondent in your own words" (p.314). Such advice shows that even a temple of the positivist approach is beginning to abandon the dogma of standardization, and accepts the need for interviewers to adapt themselves to the respondents' cultural codes and psychological states. GROVES (1989, p.220) proposes that we create a task force of refusal converters. Will it be the emergence of a new profession? [56]

Acknowledgments

I want to thank Aaron V. CICOUREL and Alessandro DURANTI, who both have addressed the subject of research practices and have given me advice on methodology. I am especially grateful to William A. CORSARO, Nigel FIELDING, Barry SAFERSTEIN and Margrit SCHREIER who reviewed an early draft and made helpful comments and suggestions.

Notes

1) The present 30% of nonresponse is clearly beyond what is acceptable if we consider that twenty years ago CONVERSE and SCHUMAN wrote that it was "professionally acceptable to lose [as nonrespondents] the 10% of sample (with another 10% of not-at-home people" (1974, p.40). In their research conducted in Detroit the authors pointed out with concern the following nonresponse rates: 16% in 1969, 15% in 1970 and 12% in 1971. <back>

2) In the sixties surveys over-represented adult males and people with medium-high education (see PITRONE 1984, p.149). <back>

3) DE MAIO instructed the interviewers to collect details on the interactional situation in which the refusal occurred and to write down the refuser's exact words. But it is not methodologically correct to interpret literally the verbal text of the refusal. E.g. "I am busy", "I don't have time", may only be idioms with which to dismiss the interviewer instead of factual observations. It is not clear if DE MAIO problematized the distinction between text and meaning. <back>

4) In the sixties 60% of respondents declared that they had been contacted at least once on the false pretence of an interview that ended up with a commercial offer (BIEL 1969). <back>

5) SINGER, HIPLER and SCHWARZ (1992) note that paradoxically the interviewer's assurances about the confidentiality of respondents' information sometimes produce the opposite effect. <back>

6) It is not unusual to encounter refusals motivated by statements such as "my husband doesn't want ...". Also interviews can be interrupted by the arrival of husbands or parents who are against interviews about their family. <back>

7) The handbooks I have reviewed do not analyze problems about explaining the aims of the interview, or the problems related to gaining access to the respondent. They undervalue the social filter created by suspicious parents, wives and husbands, who could ask about reasons for the call and for information about the interview. <back>

8) The 1983 edition of the SURVEY RESEARCH CENTER—Ann Arbor—handbook dedicates eleven pages to the first telephone and face-to-face contact, with many examples of actual responses to the respondent's objections and resistances). <back>

9) ROGERS emphasizes that "offering the option of a later contact by either method provided one more opportunity to try to convince the respondent of the need for his cooperation ..." (1976, pp.201f.). CONVERSE and SCHUMAN (1974, p.41) suggest setting a second or third appointment. FITZGERALD and FULLER (1982, pp.11f.) maintain that various numbers of callbacks produce a significant reduction of reluctant or difficult-to-reach respondents and refusals. <back>

10) Interviewers mention that to show the face at the initial contact helps in obtaining consent for the interview. A larger number of refusals occur where initial contact is by telephone or at the respondent's entry-phone in a block of flats. <back>

11) Differently from these authors, BAILEY writes: "It is better that the respondent offers answers wider than necessary, instead of insisting on the fact that they have to be concise, appropriate, and give only the information required" (1978, p.221). <back>

12) The manual of the SURVEY RESEARCH CENTER (Ann Arbor) states that the interviewer must convince the respondent of four things: that s/he a) is a professional interviewer; b) is calling for a legitimate and reputable organization; c) is engaged in important and worthwhile research; d) the respondent's participation is vital to the success of the research (1983, p.311). <back>

13) Among the ten initial contacts four began with a conversation with parents and six got straight through to the intended respondent. <back>

14) Notation: [ interruption; Par. parent; INT. interviewer; R respondent. <back>

15) A somewhat similar strategy has been proposed by SURVEY RESEARCH CENTER (Ann Arbor) (1983, p.317) which classifies it as active listening. It consists in sympathizing with the respondent's objections, listening closely to what s/he is saying, rephrasing what s/he has said, and reflecting it back to him/her along with an explanation of why s/he need not be concerned. <back>

16) The affiliation seems to play an important role:

Brunner and Carroll (1969) find rather dramatic increases in the positive effect of the letter in the first interview when university affiliation versus a market research organization affiliation is made. Ferber and Sudman (1974) report an experiment in the city of Chicago where the University of Illinois letter increased cooperation from 75 to 89 percent versus a Census Bureau letter from 64 to 81 percent (GROVES 1989, p.211). <back>

17) Besides the interviewer can be embarrassed by the wording, the content and the form of the questions (LUTYNSKA 1980, pp.50-52; FIDELI & MARRADI 1996, pp.25f.). <back>

18) The SURVEY RESEARCH CENTER—Ann Arbor (1983, p.312, p.320) advises interviewers to avoid pauses. The rationale is that a pause gives the opportunity to refuse or avoid later appointments, and it shows the interviewer as not being self-confident. <back>

19) The aim of the research is another problem. As ROTH (1962) and HOLDAWAY (1982, p.65) pointed out, the final objectives of research are rarely known at an early stage. <back>

20) However is it not a lie communicating to the respondent that the researcher is interested in him/her as an individual person (CONVERSE & SCHUMAN 1974, p.45; FOWLER 1984, p.52; GROVES 1989, p.211) when the real motive of our interest is only to maintain the randomness of the sample? Is it not a lie, in order to arouse respondents' interest, to say that s/he will enjoy the interview (FREEDMAN & FRASER 1966; CONVERSE & SCHUMAN 1974), suggesting it will be a nice moment where s/he will think about things they have never thought about before, without knowing the respondent's real taste? <back>

References

Atkinson, Maxwell J. & Heritage, John (Eds.) (1984). Structures of social action: Studies in conversational analysis. Cambridge: Cambridge University Press.

Bailey, Kenneth D. (1978). Methods in social research. New York: Free Press.

Becker, Howard S. (1964). Problems in the publication of field studies. In Arthur J. Vidich, Joseph Bensman & Maurice R. Stein (Eds.), Reflections on community studies (pp.267-284). New York: Wiley.

Biel, Alexander L. (1969). Abuses of survey research techniques: the phony interview. Public Opinion Quarterly, 31, 213-223.

Boccuzzi, Ermelinda (1985). Parcellizzazioni e reificazioni nelle ricerche sociologiche: il punto di vista di un'intervistatrice. Rassegna Italiana di Sociologia, XXVI(2), 239-26.

Bokszanski, Zbigniew & Piotrowski, Andrzej (1980). Sociolinguistic aspects of questionnaire interview application. The Polish Sociological Bulletin, LII(4), 45-60.

Bosio, Albino C. (1996). "Grazie no!": il fenomeno dei non rispondenti. Quaderni di Sociologia, XL(10), 31-44.

Brehm, John (1993). The phantom respondents. Opinion surveys and political representation. Ann Arbor: University of Michigan Press.

Brunner, G.A. & Carroll, S.J, Jr. (1969). The effect of prior notification on the refusal rate in fixed address survey. Journal of Advertising, 9, 42-44.

Cannell, Charles F.; Fowler, Floyd J. & Marquis, Kent H. (1968). The influence of interviewer and respondent psychological and behavioral variables on reporting in household interviews. Vital and Health Statistics, Series 2(26). Washington, DC: US Government Printing Office.

Castellano V. & Henzel A. (1971). Elementi di teoria dei campioni. Roma: Ilardi.

Cicourel, Aaron V. (1964). Method and measurement in sociology. New York: The Free Press.

Cicourel, Aaron V. (1970). Basic and normative rules in the negotiation of status and role. In Hans Peter Dreitzel (Ed.), Recent sociology (no. 2: Patterns of communication behavior, pp.4-45). London: Macmillan.

Cicourel, Aaron V. (1980). Three models of discourse analysis: the role of social structure. Discourse Processes, 3, 101-132.

Cicourel, Aaron V. (1982). Language and belief in a medical setting. In Heidi Byrnes (Ed.), Contemporary perceptions of language: interdisciplinary dimensions (pp.79­100). Georgetown: Georgetown University Press.

Cicourel, Aaron V. (1987). The interpenetration of communicative context: examples from medical encounters. Social Psychology Quarterly, 50(2), 217-226.

Collins, M.; Sykes, Wendy; Wilson, P. & Blackshaw, N. (1988). Nonresponse: the UK experience. In Robert M. Groves, Paul P. Biemer, Lars E. Lyberg, James T. Massey, William L. Nicholas II & Joseph Waksberg (Eds.), Telephone survey methodology (pp.213-232). New York: Wiley.

Converse, Jean Marie & Schuman, Howard (1974). Conversations at random: Survey research as interviewers see it. New York: Wiley.

Corsaro, William A. (1981). Communicative processes in the social organization study: sociological approaches to discourse analysis. Text, 1(1), 5-63.

Corsaro, William A. (1985). Friendship and peer culture in the early years. Noorwood, NJ: Ablex.

Crespi, Irving (1988). Pre-election polling: source of accuracy and error. New York: Russell Sage Foundation.

de Leeuw, Edith D. (1999). How do successful and less successful interviewers differ in tactics for combating survey nonresponse? Bullettin de Methodologie Sociologique, 62, 29-42.

DeMaio, Teresa J. (1980). Refusals: who, where, and why? Public Opinion Quarterly, 44(2), 223-233.

Deutscher, Irwin (1972). Public and private opinions: social situations and multiple realities. In S.Z. Nagy & Ronald G. Corwin (Eds.), The social context of research (pp.323-349). New York: Wiley.

Dillman, Donald A. (1978). Mail and telephone surveys. New York: Wiley.

Dillman, Donald A.; Gallegos, Jean Gorton & Frey, James H. (1976). Reducing refusals rates for telephone interviews. Public Opinion Quarterly, 1, 66-78.

Erikson, Kai T. (1967). A comment on disguised observation in sociology. Social Problems, 14, 366-73.

Ferber, Robert & Sudman, Seymour (1974). Effects of compensation in consumer expenditure studies. Annals of Economic and Social Measurement, 3(2), 319-331.

Fideli, Roberto & Marradi, Alberto (1996). Intervista. In Enciclopedia delle scienze sociali. Treccani: Roma.

Fielding, Nigel (1982). Observational research on the National Front. In Martin Bulmer (Ed.), Social research ethics. An examination of the merits of covert participant observation (pp.80-104). London: Macmillan Press.

Fitzgerald, Robert & Fuller, Linda (1982). I hear you knocking but you can't come in: the effects of reluctant respondents and refusers on sample survey estimates. Sociological Methods and Research, 11(1), 3-32.

Fletcher, Joseph (1966). Situation ethics. London: S.C.M. Press.

Fowler, Floyd J. Jr. (1984). Survey research methods. Beverly Hills: Sage.

Fowler, Floyd J. & Mangione, Thomas W. (1990). Standardized survey interviewing. Minimizing interviewer-related error. Newbury Park: Sage.

Freedman, J. L., & Fraser, S.C. (1966). Compliance without pressure: The foot-in-the-door technique. Journal of Personality and Social Psychology, 4, 195-202.

Garfinkel, Harold (1962). Common sense knowledge of social structures: the documentary method of interpretation in lay and professional fact-finding. In Jordan M. Scher (Ed.), Theories of the mind (pp.pp.689-712). New York: The Free Press.

Garfinkel, Harold (1964). Studies of the routine grounds of everyday activities. Social Problems, XI(3), 225-50.

Glaser, Barney G. & Strauss, Anselm L. (1967). The discovery of grounded theory: strategies for qualitative research. Aldine: Chicago.

Gobo, Giampietro (2000). Representativeness and generalizability in qualitative research. Paper presented at the Fifth International Conference on Social Science Methodology, ISA, Koeln, October, 3-6.

Goffman, Erving (1959). On face-work: An analysis of ritual elements in social interaction. Psychiatry, XVIII(3), 213-231.

Goffman, Erving (1956). Embarrassment and social organization. American Journal of Sociology, LXIII(3), 213-231.

Goudy, Willis J. (1976). Nonresponses effects on the relationship between variables. Public Opinion Quarterly, 40, 360-369.

Goyder, John (1987). The silent minority. Nonrespondents on sample surveys. Cambridge: Polity Press.

Goyder, John & Leiper, Jean McKenzie (1985). The decline in survey response: a social values interpretation. Sociology, 19, 55-71.

Groves, Robert M. (1989). Survey errors and survey costs. New York: Wiley.

Groves, Robert M. & Lyberg, Lars E. (1988). An overview of nonresponse issues in telephone survey. In Robert M. Groves, Paul P. Biemer, Lars E. Lyberg, James T. Massey, William L. Nicholas II & Joseph Waksberg (Eds.), Telephone survey methodology (pp.191-211). New York: Wiley.

Gumperz, John (1982). Discourse strategies. Cambridge: Cambridge University Press.

Hammersley, Martyn (1990). Reading ethnographic research: a critical guide. Longmans: London.

Hammersley, Martin (1992). What's wrong with ethnography? Methodological explorations. Routledge: London.

Heckman, James J. (1976). The common structure of statistical models of truncation, sample selection, and limited dependent variables and simple estimator for such models. Annals of Economic and Social Measurement, 5/4, 475-92.

Heckman, James J. (1979). Sample selection bias as a specification error. Econometrica, 47, 153-61.

Holdaway, Simon (1982). "An inside job": a case study of covert research on the police. In Martin Bulmer (Ed.), Social research ethics. An examination of the merits of covert participant observation (pp.59-79). London: Macmillan Press.

Kalton, Graham (1983). Compensating for missing survey data. Ann Arbor: Survey Research Center, Institute for Social Research, University of Michigan.

Kish, Leslie (1965). Survey sampling. New York: Wiley.

Lavrakas, Paul J. (1987). Telephone survey methods. Sampling, selection and supervision. Newbury Park, CA: Sage.

Lissowski, Grzegorz (1969). Z zagadnien doboru proby. In Klemens Szaniawski (Ed.), Metody statystyczne w socjologii'. Varsavia: PWN.

Lutynska, Krystyna (1987). Questionanire studies in Poland in the 1980s (Analysis of refusals to give an interview). The Polish Sociological Bulletin, 3, 43-53.

Lutynska, Krystyna (1980). Some problems of methodological research into the interviewer influence. The Polish Sociological Bulletin, 2, 45-55.

Marradi, Alberto (1989). Casualità e rappresentatività di un campione: contributo a una sociologia del linguaggio scientifico. In Renato Mannheimer (Ed.), I sondaggi elettorali e le scienze politiche (pp.51-133). Milano: Angeli.

Maynard, Douglas (1984). Inside plea bargaining: the language of negotiation. New York, Plenum.

Maynard, Douglas (1992). On clinicians co-implicating recipients' perspective in the delivery of diagnostic news. In Paul Drew & John Heritage (Eds.), Talk at work. Interaction in institutional settings (pp.331-358). Cambridge: Cambridge University Press.

Maynard, Douglas; Houtkoop-Steenstra, Hanneke; Schaffer, Nora Cate & van der Zouwen, Hans (2001). Standardization and tacit knowledge: interaction and practice in standardized interview. New York: Wiley.

Miles, Matthew B. & Huberman, Michael A. (1994). Qualitative data analysis: an expanded sourcebook. Sage: Thousand Oaks.

Miller, Peter V. & Cannell, Charles F. (1982). A study of experimental techniques for telephone interviewing. Public Opinion Quarterly, 46(2), XX

Morton-Williams, Jean (1993). Interviewer approaches. Aldershot: Dartmouth.

Mosteller, Frederick (1968). Association and estimation in contingency tables. Journal of American Statistical Association, LXIII(321), 1-28.

O'Neil, Michael J. (1979). Estimating the nonresponse bias due to refusals in telephone surveys. Public Opinion Quarterly, 43, 218-232.

Oksenberg, Lois; Coleman, Lerita & Cannell, Charles F. (1986). Interviewers' voices and refusal rates in telephone surveys. Public Opinion Quarterly, 50, 97-111.

Oldendick, Robert W. & Link, Michael W. (1994). The answering machine generation. Public Opinion Quarterly, 58, 264-73.

Peräkylä, Ansii (1997). Reliability and validity in research based upon transcripts. In David Silverman (Ed.), Qualitative research (pp.201-19). Sage: London,.

Pitrone, Maria Concetta (1984). Il sondaggio. Milano: Angeli.

Platek, Richard (1977). Some factors affecting nonresponse. Survey Methodology, 3, 191-214.

Platek, Richard (1980). Causes of incomplete data, adjustments and effects. Survey Methodology, 6, 93-132.

Platek, Richard; Singh M.P. & Tremblay V. (1978). Adjustment for nonresponse in surveys. In N. Krishnan Namboodiri (Ed.), Survey sampling and measurement (pp.157­174). New York: Academic Press.

Rogers, F. Theresa (1976). Interviews by telephone and person: quality of responses and field performance. Public Opinion Quarterly, 40(1), XX

Rosenhan, David (1973). On being sane in insane places. Science, 179, 250-8.

Sacks, Harvey; Schegloff, Emmanuel A. & Jefferson, Gail (1974). A simplest systematics for the organization of turn-taking in conversation. Language, 50(4), 696-735.

Schleifer, Stephen (1986). Trends in attitudes toward and participation in survey research. Public Opinion Quarterly, 50, 17-26.

Schenkein, Jim (Ed.) (1978). Studies in the organization of conversational interaction. New York: Academic Press.

Schwiran, Kent & Blaine, Harry (1966). Questionnaire-return bias in the study of blue-collar workers. Public Opinion Quarterly, XXX(4), 656-663.

Senf, Janet H. (1987). The option to refuse: a tool in understanding nonresponse in mailed surveys. Evaluation Review, 11(6), 775-781.

Sharp, Laure M. & Frankel, Joanne (1983). Respondent burden: a test of some common assumptions. Public Opinion Quarterly, 43(1), 36-53.

Singer, Eleanor & Presser, Stanley (Eds.) (1989). Survey research methods. A reader. Chicago: University of Chicago Press.

Singer, Eleanor; Hipler, Hans-J. & Schwarz, Norbert (1992). Confidentiality assurances in survey: reassurance or threat? International Journal of Public Opinion Research, 4, 256-66.

Smit, J. & Dijkstra, Will (1991). Persuasion strategies for reducing refusal rates in telephone surveys. Bulletin de Methodologie Sociologique, 33, 3-19

Smith, Tom W. (1983). The hidden 25 percent: an analysis of nonresponses on the 1980 General Social Survey. Public Opinion Quarterly, 47(3), XX

Steeh, Carlotte (1981). Trends in nonresponse rates. Public Opinion Quarterly, 45, 40-57.

Stinchcombe, Arthur L.; Jones, Calvin & Sheatsley, Paul (1981). Nonresponse bias for attitude questions. Public Opinion Quarterly, 45, 359-75.

Suchman, Lucy & Jordan, Brigitte (1990). Interactional troubles in face to face survey interviews. Journal of the American Statistical Association, 85(409), 232-53

Survey Research Center (Ann Arbor) (1969). Interviewer manual. Ann Arbor: Institute for Social Research, University of Michigan.

Survey Research Center (Ann Arbor) (1976). Interviewer manual. Ann Arbor: Institute for Social Research, University of Michigan.

Survey Research Center (Ann Arbor) (1989). Interviewer manual. Ann Arbor: Institute for Social Research, University of Michigan.

Survey Research Center (Berkeley) (1990). Interviewer's basic training manual. Berkeley: mimeo.

Tobin, James (1958). Estimation of relationships for limited dependent variables. Econometrica, 26, 24-36.

Tuckel, P. & Feinberg, B. (1991). The answering machine poses many questions for telephone survey researchers. Public Opinion Quarterly, 55, 200-17.

Van Dijk, Teun A. (1983). Cognitive and conversational strategies in ethnic prejudice. Text, 3(4), 375-404.

Van Dijk, Teun A. (Ed.) (1985). Handbook of discourse analysis. London: Academic Press

Wallis, Roy (1977). The moral career of a research project. In Colin Bell & Howard Newby (Eds.), Doing sociological research (pp.149-69). London: Allen & Unwin.

Wieder, Lawrence D. (1974). Language and social reality: the case of telling the convict code. The Hague: Mouton.

Wilson, Thomas P. (1989). Metodi qualitativi "contro" metodi quantitativi nella ricerca sociale. Sociologia e Ricerca Sociale, X(29), 5-33.

Author

Giampietro GOBO, senior lecture at the Dept. of Social and Political Studies, Faculty of Political Science, University of Milan, has specialized in qualitative and quantitative methods. He published two methodological handbooks. The former on survey interview (Le risposte e il loro contesto. Processi cognitivi e comunicativi nelle interviste standardizzate, Milano, Angeli, 1997); the latter on the ethnographic methodology (Descrivere il mondo. Teoria e pratica del metodo etnografico, Roma, Carocci, 2001). His current research interests concern in organizational studies, mainly on cooperative interactions in workplace and management. He is co-chair of the Research Network on Qualitative Methods of ESA (European Sociological Association).

English articles are:

1. Class: stories of concepts. From ordinary language to scientific language. Social Science Information, 32(3), 1993, 467-89.

2. Class as metaphor. On the unreflexive transformation of a concept in a object. Philosophy of the Social Sciences, 25(4), 1995, 442-67.

3. Review of David Silverman, Doing Qualitative Research. A Practical Handbook, London, Sage, 2000. International Journal of Social Research Methodology, 3(6), 170-2.

German articles are:

1. Schlecht informiert, aber nicht gefahrdet AIDS: Metaphern einer Krankheit in den Augen der Jugend. In Carlo Buzzi and Pierangelo Peri (Eds.), Drogen im Alltag der Jugend (pp.149-169). Bozen: Provincia Autonoma di Bozen-Alto Adige, 1990.

Contact:

Dipartimento di Studi Sociali e Politici
Facoltà di Scienze Politiche, Università degli Studi di Milano
via Vivaio 7 - 20122 Milano, Italy

Tel. ++39 / 02 / 5835 - 8811 (office); 498 5793 (home)
Fax: ++39 / 02 / 5835 - 8840
E-mail: Giampietro.Gobo@unimi.it

Appendix: A Summary Table