Requests, Blocking Moves, and Rational (Inter)action in Survey Introductions

Abstract

We draw on conversation analytic methods and research to explicate the interactional phenomenon of requesting in general and the specific case of requesting participation in survey interviews. Recent work on survey participation has given much attention to leverage-saliency theory, but has not engaged how the key concepts of this theory are exhibited in the actual unfolding interaction of interviewers and potential respondents. We do so using digitally recorded and transcribed calls to recruit participation in the 2004 Wisconsin Longitudinal Study. We describe how potential respondents present interactional environments that are relatively discouraging or encouraging, and how, in response, interviewers may be relatively cautious or presumptive in their requesting actions. We consider how the ability of interviewers to tailor their behavior to their interactional environment can affect whether the introduction reaches the point at which a request to participate is made, the form that this request takes, and the sample person's response. Our analysis contributes to understanding how we might use insights from the analysis of interaction to increase cooperation with requests to participate in surveys.

Declining participation in survey interviews is a problem of urgent importance for social science (Groves, Singer, and Corning 2000; Roose, Lievens, and Waege 2007). Leverage-saliency theory (LST) provides a perspective on survey participation that is related to rational choice theory but also emphasizes that the content of survey requests matters for participation. Nonetheless, LST and the research on which it builds give little attention to the details of survey interactions themselves. Instead, LST treats a survey design as offering a relatively fixed set of attributes, each of which has a “leverage” (valence and weight) for the sample person that can be made salient in the interaction and thus exert influence that varies across sample persons. With a conversation analytic approach, however, it is possible to explore matters of leverage and salience, and by extension, rational choice, as dynamic aspects of the interaction between an interviewer and a potential respondent. In this paper, we draw on conversation analytic methods and research to explicate the interactional phenomenon of requesting in general and the specific case of requesting participation in the survey interview. By using this approach in combination with LST, we seek to better understand why requests for participation requests unfold as they do and how interviewers can improve the chances of achieving cooperation.

LEVERAGE-SALIENCY THEORY

Leverage-saliency theory (LST) is a theory of how potential respondents make decisions to participate in the survey interview. The theory can be considered an elaboration of a simple rational choice theory (RCT) with social and cognitive elements (Roose et al. 2007); the theory assumes that a potential respondent has an expected utility for participating in a survey and agrees to do so if this expected utility is greater than other uses of time and effort. “Leverage” refers to the potential respondent's assessments (including both valence and weight) of attributes of the survey that make participation more or less appealing. As examples, for many people a cash incentive might have a positive valence and a greater weight as the size of the incentive increases; and a long interview might have negative valence and a weight that increases with the length of the interview (Dijkstra and Smit 2002:126). Whether an attribute has a positive or negative leverage varies across sample persons. A specific survey topic may have positive leverage for sample persons who are more generally interested in talking about that topic and a negative leverage for those who are not (Groves et al. 2006; Groves et al. 2000).

“Saliency”—or salience—refers to the prominence of different attributes of survey participation for a sample person who is deciding whether or not to participate. While orthodox rational actors use all available information, LST calls attention to how survey organizations and interviewers provide information to sample persons and thereby influence their decision-making. A survey might both provide a financial incentive and appeal to “civic duty,” for example, but emphasize one of these, and thus make it more salient and potentially more influential to sample persons as they decide whether to participate. Consequently, requests for participation in a given survey might obtain different responses from the same person, depending on which attributes are made most salient. In LST, the decision to participate is based on the combination of the leverage and salience of attributes, with the leverage of a specific attribute mattering more or less according to how salient it is.

LST provides an explicit model of how survey practitioners could heighten the probability of acceptance by increasing the salience of attributes with positive leverage and neutralizing the salience of those with negative leverage. This accords with actual practice: interviewers emphasize the positive aspects of participating and either omit negative aspects or present them in a way that attempts to mitigate them, as when an interviewer acknowledges that an interview takes a long time but notes that it can be broken into parts and that the sample person can quit at any time (Dijkstra and Smit 2002:127). By emphasizing that the leverage a survey attribute has differs across sample persons, LST calls attention to the importance of the interviewers’ tailoring of requests to a sample person's cues. Interviewers can encourage participation by observing “idiosyncratic concerns of the householder and [customizing] their remarks to those concerns” (Couper and Groves 1992; Groves, Cialdini, and Couper 1992; Groves et al. 2000:299; Maynard and Schaeffer 2002b).

When the leverage a survey attribute has for a particular sample person is defined as a psychological attribute, it cannot be directly observed, but previous research has identified, for example, groups of sample persons for whom the leverage of an attribute, such as incentives, can be presumed to vary. Thus, Groves et al. (2000) examine whether financial incentives have the same effect in generating participation among sample persons with different levels of community involvement (see also Groves et al. 2006, Roose et al. 2007). Survey practitioners recognize that if interviewers can assess the leverage that a survey attribute has for a sample person, they can tailor their communications to increase the salience of survey attributes with positive leverage (Groves et al. 2000:307). We take a less psychological approach and examine the talk between interviewer and sample person in order to observe how the sample person may display the leverage of various attributes of the survey or the request itself and how, in response, the interviewer makes features of the survey design salient or suppresses their salience. That is, both leverage and salience are at least partly phenomena that interviewer and sample person observably develop as they interact with one another.

Research in Conversation Analysis (CA) has shown how, in “pursuing a response” (Pomerantz, 1984b) or revising initiations including requests (Davidson 1984), participants display their understanding of the leverages (to use the vocabulary of LST) that could prevent acceptance of such requests. Previous survey research has coded what interviewers do during persuasion into “broad categories of interviewer actions” and “strategies” in interviewers’ turns of talk (Groves and Couper 1998:260) but has done so apart from the embedding of these actions and strategies in interaction. Furthermore this research has sometimes used interviewers’ reports of their strategies, which can be unreliable (Campanelli, Sturgis, and Purdon 1997), rather than recordings (Groves and Couper 1996). In contrast, our approach uses recordings rather than recollections, and analyzes real-time, actual practices that may reduce or enhance participation rates.

We now turn to the study of interaction, first explicating the generic action of requesting. Then our attention goes to requests for participation in the survey, and we examine differing interactional environments in which interviewers produce these requests. Later, we show that interviewers may not be able to ask for participation because sample persons “block” the production of a request. In these interactional exchanges, we explore leverage and salience as matters displayed in the talk of co-participants.

ANALYZING REQUESTING ACTIONS

CA research on requesting, and especially the more generic practices or structures that requests involve, provide a resource for gaining new insights into survey interview requests. Although previous studies of survey introductions have used CA (Houtkoop-Steenstra and vanden Bergh 2002; Maynard and Schaeffer 1997, 2002a), none has fully explicated the more generic practices of ordinary requesting as a backdrop for understanding the dynamics of asking for participation.

Requests also have occupied investigators in the philosophical tradition of speech act theory, and in the pragmatics area of linguistics. Searle (1969:69; 1975:61), a speech act theorist, considers the “essential rule” of requesting to be that an utterance attempts to get a recipient to “do something,” according to a set of “felicity conditions” (Grice 1975) or shared rules and rational understandings. These understandings provide the opportunity for a hearer to infer that a question such as “Could you do this for me?” is requesting something rather than seeking information about the hearer's ability (Searle 1969:68). In pragmatics, the politeness theory of Brown & Levinson (1987) suggests that requests are done in “indirect” ways (like Searle's example) to avoid threatening the “face” of a recipient. From a CA standpoint, Curl and Drew (2008) review and critique these approaches, suggesting that speech act theory analytically overemphasizes participants’ cognition and use of inferencing rules to go beyond the literal meaning of utterances, while politeness theory—although interested in the actual design of utterances—either uses fabricated examples or abstracts examples of actual, spoken utterances from their contexts of production.

Instead, like Curl and Drew (2008), and following Heritage's (1984:19-22) critique of attempts to develop causal, abstract explanatory models of social action without analyses of the concrete conduct of actors, we study requests in their interactional contexts. To start, we explicate practices of requesting by using the CA concept of “preference.” Preference might seem to have a psychological referent—the desires of a speaker who emits a bit of talk to achieve those desires. In CA, however, preference refers to design features of the talk itself: the interactional accompaniments to turns of talk that initiate actions and to the subsequent or responsive turns. Preference structure, in other words, is exhibited in patterns of talk. For example, as first or initiating turns in two-part sequences, offers are preferred over requests,ⁱ and one way this is visible is that requests may be deferred through preliminary moves including a turn or turns that provide background to a projected request, as when a caller leads up to a request with “I have a big favor to ask you” and some background information about a broken “buttonholer” (Schegloff 1980). The preliminary turns and backgrounding foreshadow or project a request for help without yet explicitly requesting. Then the call recipient interjects, “Rita, I told ya when I made the blouse I'd do the buttonholes.” Thus, the potential recipient of a projected and dispreferred action—a request—preempts with a preferred action—an offer. Other features of requesting also attest to its dispreferred and delicate status: When a person telephones to make a request, co-participants may work through several more casual topics before the request is finally performed (Schegloff 2007:83-87).

Understanding that, interactionally, requests are dispreferred actions in ordinary conversation gives us some purchase on the structural difficulties associated with requesting participation in the survey interview. Interviewers are working against the pattern whereby a request may be postponed until some later point in the conversation, after other topics are pursued. In contrast to ordinary conversation, the requesting action in a survey is often the first order of business and is the sole purpose of the call. Furthermore, although preliminary identification and recognition activities may precede the request, it is extremely rare for sample persons to offer participation. That is, the interaction that leads to the request is highly unlikely to generate an offer to participate. In the subsample of 200 acceptance calls we studied, only two result from offers made by sample persons. At line 20 below is one of these (see ASR supplemental materials for transcribing conventions):

(1) HP059_107012g.wav

1 FR: Hello?
2 (0.3)
3 FI: ↑Hi: ↓could I speak to Brenda Caw please?
4 (0.4)
5 FR: ↑Speaking.
6 (.)
7 FI: ↓oHi:o I’m calling about the Wisconsin Longitudinal S:tudy, .hhhhh
8 u:m: (0.2) d- (0.2) didju receive a letter? (.) from us recently
9 regarding (this) stu[dy? ]
10 FR: [Yeah: ]I ↓did an’ I work nights and I’ve
11 gotten all kinds of hhhhh
12 FI: ↑Oh↓:: o[kay:.]
13 FR: [messa]ges an’ crap,
14 (0.2)
15 FI: [↑Oh(h) o↓k(hh)ay I’m s:(hh)orry.]
16 FR: [How long is this gonna take. ]
17 (0.3)
18 FI: .hhhhh U:m: well it's hard tuh say becuz it varies from person tuh
19 person, .hhhh on average, it['s-]
20 FR: [Wel]l let's just ↓do it.
21 (0.4)
22 FI: Oh: okay? (0.4) if: at any time you need tuh go ↑just let me ↓know,
23 ...

Preceding the offer, this female respondent (FR) initiates a sequence at line 16 (in talk that overlaps the turn at line 15) asking about the time involved for completing an interview. The female interviewer (FI) answers in a hedging way (lines 18-19), and then FR urges movement toward the interview in a way that is consistent with her displays that the messages she has been getting as well as the duration of the interview have negative leverage; that is, when FR says, “Well let's just do it” (line 20) it suggests getting the task done immediately to thereby end the calls and messages or otherwise to remove the bothersome nature of the interview. While her offer thereby may be rather begrudging, it does preempt the need for the interviewer to request participation.

So preemptive offers can happen in the survey introduction, but the overwhelming pattern is for interviewers to produce (or attempt to produce) a formal request. Regularly, sample persons withhold offering participation, even though interviewers, in the ways that they lead up to a request, present opportunities for sample persons to do so. In fact, usually the lead-up not only fails to elicit an offer, but often yields a pre-empting declination that blocks the interviewer from making of a request altogether.

FEATURES OF REQUESTING ACTIONS

One of the most prominent features of requesting actions is the degree to which they exhibit entitlement to make the request (Curl and Drew 2008:130). Researchers have studied requests in “institutional” settings such as home health care (Heinemann 2006; Lindström 2005) and a “copy” shop (Vinkhuyzen and Szymanski 2005), while Curl and Drew (2008) compared requests in ordinary conversation with calls to institutional settings such as doctors’ offices and other organizations, finding that requests with imperatives or interrogatives with modal verbs (e.g., “can you,” “could you,” “will you”) are high on entitlement and predominate in conversation, whereas requests to institution-based co-participants are regularly formed as declaratives prefaced with variants of “I wonder if,” and are thereby low on entitlement. A second dimension that requests can display is an assumption that granting or accepting the request does not face many contingencies—that the recipient of the request can fulfill it because there a few impediments. A third dimension on which requests vary is the use of mitigating and politeness terms such as “please” (Heinemann 2006). Their use or non-use may diminish or reinforce the displays of entitlement or contingencies just described. For example, the request (arrowed) at line 7 in extract (2), with the modal form “Could you,” is an entitled request in which there are no mitigating items and displays a “known contingency” (Curl and Drew 2008:143) when mentioning Leslie's upcoming trip (line 9).

(2) Field SO88:2:8:1

1 Les: Hello:?
2 (0.3)
3 Gor: It's Gordon.
4 Les: .hhhh Oh Gordon. Sh'l I ring you back darling,
5 Gor: Uh:: ↓no y- I don't think you can,
6 (0.3)
7 Gor: But uh: just to (0.3) say (.) Could you bring up a letter. ←
8 (.)
9 Gor: When you come up,
10 .
11 . ((Discussion re: which form))
12 .
13 Les: Okay 1

Compare this request with one that a caller makes to a doctor, and which uses an I-wonder preface (Curl and Drew 2008):

(3) 2:1:9

Doctor: .hhHello:
Caller: Hello I. I'm wonderin’ if a doctor could call and see Robert Smith please

With such prefaces, requests to persons in institutional settings are formatted to display low entitlement and an awareness of the contingencies surrounding granting of the request. These displays may be general enough to be accomplished by the I-wonder preface alone—what Gill (1998) has called a “speculative” one—or there may be a more concrete naming of contingencies (Curl and Drew 2008:141).

As we examine survey requests, we will see that modal verbs and I-wonder prefaces play a role in displaying entitlement but do so in relation to other aspects of the request, including those based on the scripted introduction. In addition, in a slight departure from Curl and Drew (2008), we distinguish “contingency” as a separate dimension of requests from entitlement. That is, we suggest that different formats for requesting, beyond displaying the “stance” of speakers toward the projected “grantability” of a request (Curl and Drew 2008:149), can respond to concrete features of the interactional environments in which the request is produced. Requests may be sequentially retrospective and take account of the interaction so far, in addition to projecting a type of next action. To adapt Heritage's (1984:242) felicitous words, requests in various ways are often “context-shaped” as well as “context renewing.”

To put matters in terms of LST, CA research about requesting actions could suggest that speakers take into account the valence and weighting that attributes of a request—their leverages—potentially have for the recipient of the request. That is, syntactic forms such as declaratives and positive or negative interrogatives, modal verbs, mitigating words or phrases, and utterance prefaces, can be used alone or together to increase or decrease the salience of features of a request that potentially have leverage with a particular sample person. As will be shown in the analysis, CA-informed research makes it possible to examine what happens in the interaction between interviewers and householders and how they assess in situ the leverage of some attribute that may be operative for a sample person.

DATA AND METHODS

We use recordings of interactions, data about sample persons, and materials prepared for interviewers from the 2004 round of the Wisconsin Longitudinal Study (WLS) survey. WLS began with a one-third sample of 1957 Wisconsin high school graduates and had follow-up waves in 1964 (mail to parents), 1975 (telephone), and 1992 (telephone and mail). WLS collects a wide range of economic, familial, health, and other information, and has been used in many studies (Hauser 2005). Because we have considerable information about those who refused to participate in 2004 as well as those who agreed, WLS provides an exceptional opportunity for our investigation.

This study is part of a larger project in which we analyze interaction between interviewers and sample persons using both CA and quantitative methods, in order to estimate the effects of interactional variables on survey participation. In 2004, calls to WLS sample persons were digitally recorded, and there were 8261 WLS sample persons for which field efforts were made that did not result in the sample person being classified as a noncontact or as incapable of participating. To ensure that the cases in our analysis had comparable histories of contacts with interviewing staff, we drew our project sample only from the 4,627 sample persons for whom the initial telephone contact with the sample person resulted in either a declination (some of which may have been converted to acceptances in a subsequent call) or an acceptance (some of which may have resulted only in partial interviews). The rate of acceptances versus declinations among these 4,627 sample persons was 88.7%.

To draw the project sample, we used logistic regression to estimate the propensity of sample persons to refuse to participate, based on their education, cognitive test scores from high school, and health status (Hauser 2005)). We then selected matching pairs of calls in which one sample person agreed to participate and the other declined. Pairs were selected to match exactly on sex and on past record of WLS participation, and then as closely as possible on the propensity score. We use matched pairs so that the successful and unsuccessful calls we analyze will be comparable with respect to important predictors of participation that precede the interaction. The most desirable sample using this method is one that is homogenous within pairs but heterogeneous across pairs with respect to the matching variables. To increase the contrast across pairs, we grouped available pairs into thirds according to their propensity to refuse and selected 100 pairs from the lowest third and 100 pairs from the highest third. The initial sample for the main project thus consists of 200 pairs (400 calls).

For the present investigation using conversation analytic methods, we analyze a subset of the calls. This collection of 57 acceptances and 51 declinations (a total of 108) was selected unsystematically based on the order of cases in sample lists and the availability of recordings. We made detailed transcriptions using conversation analytic conventions (see ASR supplemental materials). Transcribed acceptances and declinations are drawn about equally from the high- and low-propensity-to-decline strata of our sample. We use pseudonyms for all personal names, locations, and schools.

WLS has high continuing participation for a longitudinal study, but nonresponse bias remains a concern (Hauser 2005). Also, although the WLS is a longitudinal study that uses an advance letter, our research has implications for cold-called random-digit dialing samples and other surveys without previous contact. The matters of interaction we examine, such as small signs of encouragement or discouragement on the part of survey call recipients and devices for tailoring on the part of interviewers, are relevant in a wide variety of surveys as well as other contexts in which occupational callers solicit participation of various kinds (Weathersbee 2009).

Advance Letter

All calls in our CA collection are initial telephone contacts with the sample person for the 2004 survey. However, in the WLS, as in many surveys, the effort to secure participation begins with an advance letter (see ASR supplemental materials). If sample persons read the letter, various attributes of the survey discussed in the letter may affect decisions about participation. Some attributes described in the letter can be presumed to have positive leverage: the survey is confidential, it can be administered at a convenient time, sample persons may “enjoy” the interview and find it “rewarding,” the researchers “appreciate” continuing participation and will “be most grateful.” ⁱⁱ Other attributes may be either positive or negative, depending on the sample person's view of them: the University of Wisconsin is conducting the study, the study is sponsored by the National Institute on Aging, a report will be sent as soon as all interviews have been completed. Only one attribute can be presumed to be negative: sample members are asked to look over health insurance plans before the interview and to be familiar with the names of their plans. The letter does not mention another potentially negative leverage, the length of the interview, which was typically well over an hour. Overall, we speculate that an advance letter affects the context within which the initial contact takes place in several ways: It provides the interviewer with a resource to incorporate into her introduction, it may relieve the interviewer of needing to provide some details of identity or the purpose of the call, and some sample persons who have read the letter may become predisposed one way or another before getting the call.

Introductory Script

For the phone call to a household, the WLS provides the interviewer with an introductory script in the form of a series of screens on the interviewer's monitor. The first screen of interest contains the following:

Hello, my name is [SAY NAME]. I am calling from the University of Wisconsin Survey Center at the University of Wisconsin-Madison. May I please speak to [RESPONDENT'S NAME]? (IF NECESSARY: We're not advertising or selling anything.)

If the person who initially answers the phone appears to be the targeted sample person, the interviewer's script reads:ⁱⁱⁱ

Is this the [RESPONDENT'S NAME] that was enrolled in [NAME OF HIGH SCHOOL] High School in 1957? [IF YES:] As you probably recall from our recent letter, we are doing a follow-up study of our sample of people who were Wisconsin high school seniors in 1957. We'd like to interview you now for this important study.

At this point, interviewers have identified themselves by name and activated the salience of their institutional identity, the original survey in 1957, and possibly whatever other attributes (and leverages they evoke) that may be remembered “from our recent letter.” With its last sentence—“We'd like to interview you now for this important study,” the screen also poses the official or formal request. Consistent with previous research, we find that, in their practices, interviewers engage in considerable “analytic alternation” (Maynard and Schaeffer 2000): they use the script in their talk, but then embellish or improvise as the occasion calls for, only to return to a close reading of that script when possible or necessary. WLS interviewers were trained to consider the scripted introduction as “flexible,” meaning that they could use it as a guide rather than following it verbatim (Houtkoop-Steenstra and vanden Bergh 2002; Morton-Williams 1993).

FORMS OF REQUESTING IN WLS INTERVIEWS

In developing LST, Groves and Couper (1998:219-45) noted how experienced interviewers, in “maintaining” the interaction with sample persons, may observe the leverage some survey attributes have with sample persons. Interviewers can use cues provided by sample persons to tailor their own talk to increase the salience of attributes with positive leverages and to decrease the salience of those with negative leverages (Groves et al. 2000:308). Our examination of practices indicates that interviewers show sensitivity to the interactional environment in which the survey request occurs, including the various kinds of detailed vocalizations as well as silences from the sample member and sometimes from a spouse or other informant who initially answers the phone. Depending on these detailed cues, we came to classify the actions of sample persons or informants when answering an interviewer's call as discouraging, encouraging, or ambiguous, following how interviewers themselves seemed to orient to the cues.

Past CA research considers many instances of requests among non-professionals and requests where non-professionals appeal to individuals in particular institutional (e.g., health care) settings. In requests for survey participation, the usual roles are reversed in that institutional actors (interviewers) are making requests of individuals who are approached solely because they were 1957 Wisconsin high school graduates and not in their occupational or other institutional identities. While our study adapts the concepts of entitlement, contingency, and mitigation described in earlier studies of institutional and ordinary interactions, we add two other relevant dimensions of survey requesting—task partitioning and preemption—into the analysis, as follows.

1. Entitlement. Scripted requests in the WLS (“We'd like to interview you now for this important study”) exhibit entitlement in the sense of claiming a right to the interview by taking participation for granted through the syntax and verb forms being employed. Spoken interviewers’ requests nonetheless vary in the degree of entitlement. Those that are relatively high in entitlement, as compared with those that are not, employ modal verbs (“we would like to interview you now”) or turn-initial copular verbs (“is this a good time to start the interview?”). Requests are relatively higher in entitlement when they use a declarative syntactic form as compared with an interrogative syntactic form, except when a declarative request contains a “wondering” preface (“we were wondering if now is a good time to start the interview”) or other prefacing phrase of speculation (Gill 1998).

2. Contingency. WLS survey requests vary in when they suggest the interview could be done. Those that present only the option of doing the interview “now”—what we call “one option” requests—display a presumption that there are no obstacles or contingencies that stand in the way of current participation. Those that exhibit an orientation to a possible high level of contingencies pose multiple options, for example by saying that the interview could be done now or later (or that it could be postponed until a more convenient time, or after the survey center sends additional information about the survey).

3. Task partitioning. Interviewers sometimes offer to break the interview into parts. When interviewers simply ask, for example, “to interview” the sample person or “to do the study,” they imply that it would be completed in one sitting (low partitioning). In contrast, when an interviewer asks to “start” or “begin” the task, or offer to complete the interview “in parts,” the request can be heard as implying that the instrument could be administered incrementally (high partitioning).

4. Mitigators. WLS survey interviewers vary in their use of politeness markers and other hedges such as “please, “just,” “some,” “might,” “trying” that can weaken the force or boldness of a request (Brown and Levinson 1987; Watts 2003). For example, although one way a request can be high in entitlement is by use of modal verbs, such a request can also include mitigating terms that diminish it, as when an interviewer asks, “Would you be able to work on that some this morning,” and the “be able to” and “some” soften the request.

5. Preemption. Sometimes interviewers skip sections of the opening script prior to the request, including sections that verify whether the recipient is a 1957 graduate of a particular high school, ask about the advance letter, or state the purpose of the call (“doing a follow-up study”). Because they omit this material, such pre-emptive requests thereby get produced early in the call. In the contrasting category are requests that are preceded by most or all sections of the scripted opening and therefore occur later.

Using these five practices, and analyzing the interactional environments of 69 explicit requests in our subsample of 108 cases,^iv we classify requests as relatively cautious or presumptive, where these are ends of a continuum through which interviewers design their utterances. Table 1 depicts a continuum on which requests can be arrayed, with strongly cautious requests at one end and strongly presumptive requests at the other.

Table 1.

Continuum of Cautious and Presumptive Requests

Fully Cautious Requests	Requests That Are Ambiguous	Fully Presumptive Requests
Low entitlement (“I wonder if . . .” or comparable prefacing)	Low or high entitlement	High entitlement (“We would like” “is” prefacing)
High contingency: two timing options (“now” or another time)	High or low contingency	Low contingency: only 1 timing option (“now”)
Task partitioning (“start” or “begin”)	Possible task partitioning	No task partitioning
Use of mitigators	Some mitigators	No mitigators
All three preliminary sequences present: sample person verification, letter receipt, study description	One or two preliminary sequences preempted	Preemption of all three preliminary sequences sample person verification, letter reference, study description

Fully presumptive requests are those in which at least three of the following are present: entitlement is high through the use of modal verbs (would) or copular verbs (is); only one option is presented for the timing of the interview; there is no task partitioning; there is no mitigation; or there is preemption of at least one of the following scripted statements: sample person verification (“is this the [name of person] who graduated from [name of high school] in 1957”?), reference to the advance letter, or description of the study. Fully cautious requests are those in which there are at least three of the complementary practices—i.e., entitlement is low, more than one timing option is presented, task partitioning is present, there is mitigation (at least two forms), and there is no more than one preemption. Some requests are in between, in that, for example, a request can be “mostly” presumptive or cautious depending on which and how many of the practices we just described interviewers deploy.

SURVEY REQUESTS IN THEIR INTERACTIONAL ENVIRONMENTS

Discouraging Environments and Requesting

A sample person or informant can create a discouraging interactional environment with the content of their statements, by failing to respond when an interviewer's talk provides an opportunity, responding at such points in a terse fashion, and imbuing their talk with various prosodic cues (pacing, intonation, volume). For example, in one of our calls (HP005), a female informant replies to an interviewer's request to speak to a male sample person by asking, “Who's calling please?” This response presents at least a mild challenge to the interviewer because it is a dispreferred response to the request (the preferred response being “yes” or something comparable), and it inserts a repair sequence before the request is actually answered, thus suggesting that there is a trouble with the request (Schegloff 1979:38). The interviewer then identified himself as calling from the WLS and reported that they had called “a couple days ago.” The informant replied, “Yep, many times,” in what might be a mild rebuke. On some calls, it is possible to hear offline interaction between the informant and the sample person that exhibits a stance toward the interview, but in this case, the informant said she had to “go down and let him know” and nothing was audible for about 25 seconds. That she had to retrieve him from elsewhere in the household, apparently almost a half-minute away, might also be relevant to the interviewer.

When the sample person (MR) comes to the phone, and in response to the interviewer (line 2), confirms his identity, it is with a terse, downward intoned “yes” at line 3.^v

(4) HP005

1 MR: .hhh Hello?
2 MI: tch .h ↑Hi Mister Martino? hh
3 MR: Yes.
4 MI: My name is Brandon Johnson. I'm calling from the Wisconsin
5 Longitudinal Study? .h Ah d- we sent you a letter ahu:::h probly
6 about th:ree months ago. I don't know if it- do you remember what
7 (0.4) th- ah Wisconsin Longi↑tudinal ↑Study is?
8 (0.3)
9 MR: No.
10 (.)
11 MI: #No? .hh Um (0.3) es↑sentially what it is is back in nineteen fifty
12 seven when you gradu↑ated from uh Stockdale ↑High School I think it
13 ↑says. .hh Um (0.4) we did a s- we began a st↑udy with you and
14 we've talked with you about ↑ev'ry:: twelve years since then?
5 (0.4)
6 MI: .hh Do you #re↑member ↑that at all?
7 (0.9)
18 MR: Ye:ah I remember o::ne.
19 (.)
20 MI: Okay. .hh well- (.) basically it's been a↑bout (0.2) n:: eleven
21 years, and so we're ↑doing another wa:ve of this study right now.
22 .hh um .h I was wonder↑ing if- do you have some t↑ime to maybe
23 begin it ↑now or would you like us to send you another letter to
24 remind you about what it is?
25 (0.2)
26 MR: ↑I':::m not gonna ↓be innerested sir. hh

And, after the interviewer (MI) identifies himself by his name and the name of the study (lines 4-5), the possible complete turn, its questioning intonation, and the subsequent inbreath (line 5) occasion an opportunity for acknowledgment by MR, but he bypasses this opportunity. The interviewer's claim of having “sent you a letter” (lines 5-6) also meets with no response, and when MI asks whether MR remembers the WLS, MR delays (line 8) and then answers in the negative (line 9), again tersely and with downward intonation. As MI begins to describe the study, he embeds an implicit confirmation request about the MR's high school (lines 11-12), but receives no response (notice the inbreath and other hesitations at line 13). After MI completes the study description and ends this utterance with rising intonation (lines 13-14), MR still withholds response. Subsequently MI asks again whether MR remembers the study (line 16). MR delays in answering and then only does so with a weak confirmation token and with a vague reference that minimizes his previous involvement (line 18). In a number of ways, then, this sample person, like the informant who answered the phone, shows a discouraging stance toward the possibility of participation in the interview.

In this discouraging context, after confirming MR's apparent remembrance by suggesting a time span since the last interview and identifying the study as “another wave” (lines 20-21), MI produces a request with a number of cautious features. At line 22, there is hesitation before the request, a preface that is low in entitlement (“I was wondering if”), a re-started utterance (“do you have some time to maybe begin it now . . .”) that is mitigated with the “maybe,” a suggestion of partitioning (with “begin”), and a version of two options for timing (“now” or later after re-sending the letter). None of the preliminary sequences are omitted: There is a reference to the advance letter at lines 5-6, MI's proposal to the recipient that he graduated from Stockdale High School in 1957 (which serves as sample person verification) at lines 11-13, and a description of the study accomplished over lines 13-18. The interviewer, operating in a discouraging interactional environment produces an earnestly cautious request that is, nevertheless, turned down (line 26).

In some cases—in similarly discouraging environments— interviewers’ cautiously formed requests succeed in gaining acceptance. Our point here, however, is to observe that interviewers may embellish their requests in a variety of ways that are tailored to unfolding discouragement from sample persons in the interaction.

Encouraging Environments and Requesting

When interviewers obtain early cues from sample persons that can be interpreted as encouraging, they are regularly more presumptive in their requesting practices. Encouraging environments include those in which the sample persons produce relatively immediate and explicitly agreeing responses (“right” or “correct” instead of “yes/no”), employ expansive (rather than terse or one-worded) confirmations and acknowledgments,^vi modulate the pitch within these utterances substantially (rather than using monotone), or offer unprompted displays that they recognize the study or the purpose of the call.

In extract (5), after MI introduces himself and asks to speak to the sample person (lines 3-5), MR relatively quickly acknowledges being that person (line 7), in an utterance with an intonational contour that rises and then falls slightly toward the utterance's end (sounding “affiliative,” per endnote 5). MR also acknowledges his high school graduation (line 11) at an early juncture in overlap with MI's inquiry (lines 8-10). And, when MI mentions the letter that had been sent (lines 13-14), MR not only interrupts to acknowledge receipt but also offers a report about where the letter is (lines 15-16).

(5) HP058

1 MR: tch Hello:?
2 (1.2)
3 MI: Hullo: my name is: (.) Marcus Beale an’ I'm calling from the
4 University of Wisconsin Survey Center: at the UW Ma:dison:? May
5 speak to Nathan?
6 (0.2)
7 MR: This ↑is Na:than.=
8 MI: =.hh (.) Hullo: Nathan? u:m:: (0.3) tch (0.2) is this the Nathan
9 Getz who wuz enrolled at Shellfish High School in nineteen
10 fif[ty seven?]
11 MR: [Yeah:. ]
12 (.)
13 MI: .hh An’ as you probally recall from uh recent letter
14 [we're (ogoin’ thruo)]
15 MR: [Yeah I got it ] leh- (.) layin’ on my ↑de:sk ↓in thuh
16 bedroom.
17 (0.4)
18 MI: Al↓right well is now a good time ↑for ya? sir? ←
19 (0.2)
20 MR: Hah?
21 (.)
22 MI: Is now uh good time to do the study?
23 (0.3)
24 MR: ↑Oh yah.

MI, at line 18, then preempts the next scripted item—the study description (“we are doing a follow-up study of our sample of people ...”)—and produces a request that, with an initial copula, is high on entitlement. Additionally, by posing only the option of “now” as “a good time,” the request is low in contingency, and MI does not offer to partition the task or use any mitigation. The repair that MR initiates at line 20 indicates that he had not heard the line 18 request, and after it is reproduced, he readily agrees to do the interview. The oh-prefaced form he uses proposes that his doing the interview can indeed be presumed (Heritage 1998). So even as this interviewer fashions a strongly presumptive request in line with an encouraging environment, with his acceptance the sample person then exhibits an orientation suggesting that such presumptiveness was warranted.

Ambiguous Environments and Requesting

The opening of each of these phone calls is comparatively brief. Extract (5) takes only 25 seconds from MR's answering “hello” to his acceptance, “Oh yah,” at line 24, while Extract (4) is about 47 seconds long. In these moments, nonetheless, interviewers may confront stances toward the interview that are strongly encouraging or discouraging. In reviewing the 69 requests in our subsample of 108 cases,^vii we identified 34 interview openings as predominantly encouraging for those requests and 10 as predominantly discouraging. Twenty-five environments, however, show either a mixture of encouraging or discouraging forms of responsiveness or consistently neutral displays. To us, they seem ambiguous, and we assume that if they are that way for us as analysts upon repeated inspection, it is because they are ambiguous for interviewers in the first place. That is, in classifying ambiguous environments, we are suggesting that interviewers’ own orientations exhibit this analysis.

Extract (6) is an example of an ambiguous environment. Ultimately the interview is completed. When, at line 16, MR says his time is “real flexible” it constitutes acceptance and the interview progresses from there. But up to the point at which he agrees to begin, MR is expansive in some ways but only tersely responsive in others,

(6) HP001

1 MR: .h Hello. h
2 (0.2)
3 FI: tch tch ↑Hi: can I speak tuh Evan ↓Royal please?
4 MR: oYeaho thissiz Evan speakin.
5 FI: tch Hi:: u:h my name is Linda I'm calling from thuh University of
6 Wisconsin ↑Sur:vey Center? .hh Um:: is this the Evan Royal who
7 wuz enro:lled at Belmont High School in nineteen fifty ↑seven?
8 MR: Yuh: ↑huh
9 (.)
10 FI: .h ↑Great um: (.) well as you probably re↑call from our re↑cent
11 letter .h we're doing a followup study of our sample of people who
12 ↓were uh Wisconsin high school seniors in nineteen fif↑ty ↑seven .h
13 An we'd just like to interview you now for this im↑portant ↑study
14 if you've got some ti:me?
15 (1.5)
16 MR: How ↑long does this take.

After FI asks for the sample person (line 3), the call recipient identifies himself in an expansive way, but does so with uniformly falling intonation. Then, after FI's personal and institutional identification (lines 5-6), there is no acknowledgment even though FI ends with rising intonation and takes an inbreath. In answering the sample person verification question (lines 6-7), however, MR produces an acknowledgement with notable pitch movement within the utterance. But MR does not acknowledge either the letter reference (lines 10-11) or study description (lines 11-12), and FI goes on to produce the request for participation. The environment is ambiguous in having both encouraging indicators (expansive self-identification, upward intoned confirmation) and discouraging indicators (downward intonation on the self-identification, withheld acknowledgment at turn transitions).

FI's request appears oriented to this ambiguity. While it is high on entitlement (“we'd just like to interview you”), presents just one timing option (“now”), and does not offer to partition the task, there is no preemption (sample person verification, letter reference, and study description are all present) and there are two mitigating terms (“just,” “some”). Three features associated with presumptive requesting and two characteristic features of caution are present. This is consistent with a broader pattern: In our collection of 25 requests occurring in ambiguous environments, 18 have more presumptive than cautious features. That is, requesting practices in ambiguous environments are more often like those in encouraging environments, and the smaller number of requests in ambiguous environments that are cautious are not strongly so. For example, in one call identified as having an ambiguous environment, the interviewer's request was, “And I was just wondering if now is a good time for you to start that study.” This displays facets of presumptiveness—no display of contingencies and only one mitigating “just” in the preface—but otherwise it is marked by low entitlement, task partitioning, and no preemption, all practices associated with caution.

We attribute these patterns of interaction in ambiguous environments to a phenomenon documented in previous research on survey call openings, an interactional-structural tendency toward optimism in dealing with initial queries and other responses from the sample person (Maynard and Schaeffer 2002a). Presumption optimistically treats ambiguous signals from sample persons as foreshadowing acceptance of the request.

TAILORING AND NOT TAILORING THE REQUEST

Interviewers regularly design requests in ways that are sensitive to facets of the sample person's vocal and non-vocal feedback during the opening moments of the call. Interviewers cannot know exactly what may prompt encouraging or discouraging signs on the part of sample persons—whether it is their regard (or lack thereof) for the study, the University of Wisconsin, other matters discussed in the advance letter, events unrelated to the survey request, or a general propensity to cooperate with surveys or to refuse. For the WLS, the amount of time an interview will take is certainly an attribute that occasions the most frequent displays of the leverage of survey attributes by sample persons. But even if they are not able to discern the exact leverages (either valence or weight) attributes have for sample persons, interviewers can and do design their requests to reflect interactional signs that are interpretable as taking a positive or negative stance toward the task.

“Tailoring” has been used to refer to different types of responsiveness or strategic changes by interviewers, either within an encounter or across encounters, in pursuit of cooperation by a sample person. In some of their work, Couper and Groves (2002:169-74) use a “narrow” definition of tailoring to refer to the respondent's providing information in a turn and the interviewer's using that information appropriately in the next turn, and find that there may be a small positive effect of tailoring on participation. Dijkstra and Smit (2002:130) also show that tailoring is associated with increased participation, although the relationship is similar to that for other forms of persuasion. Groves and Couper (1996:67) note that some types of tailoring can take place across contacts with a single household, and Campanelli, Sturgis, and Purdon (1997:4-54 to 4-55) systematized such an approach in their study of face-to-face interviews, after they found there were few individual contacts with any opportunity for tailoring.

The type of tailoring that we identify here, tailoring to the interactional environment, has not been previously examined. Preliminary evidence indicates that small nuances make a difference.^viii Requesting in well-tailored ways to interactional environments means acting presumptively in encouraging environments, cautiously in discouraging environments, and perhaps slightly presumptively (or less cautious) in ambiguous ones. However, in our data a small number of requests are ill fitted in the sense that an interviewer may act presumptively when a sample person shows discouragement, or very cautiously when there are signs of encouragement. We found that eight of 34 requests in encouraging environments, and three of 10 requests in discouraging contexts, seemed ill fitted.

For each of these ill-fitted requests, we attempted to examine other cases in our subsample involving the same interviewer. This was possible for nine of the 11 ill-fitted requests, and we found that in seven of the nine instances, other requests for that interviewer were consistent with the ill-fitting ones. This consistency in an interviewer's formulating of requests suggests that some interviewers may have a particular style of requesting, either idiosyncratic to the person or based upon a mechanical reading of the script. For instance, one interviewer whom we call “Tom” used a presumptive style in four different interviews in our subsample. In an environment that was encouraging (HP057), his request was “we'd like to interview you now for this important study, is that all right?” Other than the tag question, this request follows the script appearing on his computer screen and is high on entitlement, low on contingency, contains only one timing option and does not offer to partition the interview. Another interview (LP052) also has an encouraging context, and Tom's request again follows the script. In yet another of Tom's interviews that have a more ambiguous environment (LP062), he performs no preemptions but still formats his request in a predominately presumptive way: “Ah is now a good time to start that?” And, in an interview that tended toward discouraging (LP008), Tom is even more presumptive than he was in encouraging or ambiguous environment. He pre-empts several matters to say simply, “Is now a good time er:::-”, at which point the sample person declines. Thus, across three different environments, Tom is presumptive in his requesting practices, exhibiting a style that seems impervious to the cues of his individual sample persons.

Accordingly, Tom appears similar to interviewers who appear to engage habitually in less tailoring and are less successful than others (Morton-Williams and Young 1987:51). As it turns out, of the 66 WLS interviewers who had 50 or more completed or refused cases, Tom has the highest refusal rate (45%, compared to 12% on average), which is strong evidence that being unresponsive to the interactional environment—i.e., being stylistic rather than responsive in one's requesting practices—is counterproductive. Our analysis provides insight into an interactional dynamic that could underlie findings that interviewers who follow a script have lower response rates than interviewers who use a less formal agenda (Houtkoop-Steenstra and van-den Bergh 2002; Morton-Williams 1993; Morton-Williams and Young 1987). From the very inception of the phone call, such interviewers may refrain from tailoring their talk to the discernable encouraging, discouraging, or even ambiguous cues from sample persons regarding their stance toward being interviewed. They engage in uniform if not mechanical requesting practices, whether by following the script or otherwise having a relatively rigid personal style.

Environments, Requests, and Participation

The interactional sequences that eventually result in participation or non-participation begin in the opening few seconds, continue through introductory sequences, and continue beyond the request until acceptance or declination is determined. We do not yet have definitive quantitative information about how interactional environments and requests are associated with acceptances and declinations—precisely how sample persons’ cues and interviewers’ requesting practices influence response rates—but one matter for investigation is clear. In our 34 instances of encouraging environments, nearly all have presumptive requests and every case results in an acceptance, whereas in our 10 interviews with discouraging introductory environments there are 4 acceptances and 6 declinations, and in our 25 ambiguous cases are 18 acceptances and 7 declinations. In encouraging environments, both the interviewer's request and the outcome are relatively constant, and these are all likely to be determined in such close coordination that any role the interviewer might play in producing an acceptance, rather than simply allowing its expression, would be difficult to observe. In negative and ambiguous environments, however, there is more variability in both interviewer behavior and the ultimate outcome, making these environments a potentially fruitful site for quantitative analysis.

BLOCKING MOVES

So far, drawing on previous research about requesting as a social action, we have described and analyzed facets of the pivotal act of asking for participation in the survey interview. In our collection of calls, however, we found that interviewers often are not able to produce a request because a sample person refuses relatively early in the call. We call this a “blocking” move, for reasons related to the organization of sequences in conversation. Request sequences (request plus reply) are one kind of base sequence in conversation (Schegloff 2007:Chapter 4). Pre-sequences foreshadow such base sequences and enable participants to project whether a base first pair part (such as an invitation or request) will obtain its preferred second pair part (for example, acceptance). In this sense, and because they circumvent such responsive actions as rejections, pre-sequences deal with the delicacy of the initiated action (Schegloff 2007:34,90). The second part of a pre-sequence is a “go-ahead” move, with which the recipient of a pre-invitation or pre-request moves the talk toward the base sequence. Of course, as we showed earlier in extract (1), when the base sequence involves an action such as requesting, the pre-sequence initiation can provide an opportunity for a pre-emptive offer. But a pre-sequence also provides an opportunity for a blocking move (Schegloff 2007:33-4). For example, Party A's “What're you doing?” (a pre-sequence inquiry that could foreshadow an invitation as a base first pair part) can get a response from Party B such as, “Well, we're going out,” a second part to the pre-sequence that could preclude Party A's issuing an intended invitation.

In the survey interview, during the very first turns of the call, the sample person can understand the interviewer's talk as initiating a pre-sequence and project that a formal request for survey participation is forthcoming. The advance letter sent to sample persons probably facilitates such a projection for sample persons who have seen it. Consequently, although call recipients can preempt the request with an offer (extract 1) or generate go-ahead moves (Maynard and Schaeffer 2002a), they also can produce blocking moves and avoid hearing a formal request. In the excerpt below, the sample person fashions a blocking declination at line 9 and reaffirms it with a “thank you” (line 12) that invites—or initiates—closing the call (Maynard and Schaeffer 1997:57, 76-77, n22):

(7) HP018

1 FR: Hello Smi:th.
2 (0.7)
3 FI: tch .h Hi my name's Sharon I'm calling from the University of
4 Wisconsin. .h Can I speak to Michelle [Smith please?
5 FR: [This is Michelle.
6 (0.3)
7 FI: .h ((click)) Hi I'm calling for the Wisconsin Longitudin↓al Study.
8 ,.h Have you received our letter recent[ly? ]
9 FR: → [Yea:h] but I guess I don't
10 wanna par↓ticipate in it.
11 (.)
12 FR: So thank you.
13 (.)
14 FI: Oh I'm ↑sorry could I ask why?
15 (.)
16 FR: tch .h I just don't.

After the declination, the interviewer asks “why” (line 14) but to no avail (line 16). We have observed many similar instances of this in our data; perhaps interviewers ask “why” in an attempt to maintain interaction or elicit objections that they might rebut. However, in no case in the set of cases for CA examination were such questions successful in converting a (projected) declination into an acceptance.

Blocking declinations are both firm and common. In our subsample of 200 calls that resulted in declination, 126 (63%) are done using blocking moves—that is, the sample person declines before the request is made (see Table 2).

Table 2.

Location of Sample Person's Declination or Acceptance Relative to the Interviewer's Request for Survey Participation

Outcome – Sample Groups	Location		N
	Pre-Request	Post-Request
Acceptance	2 (1%)	198 (99%)	200
Declination	126 (63%)	74 (37%)	200
N	128 (32%)	272 (68%)	400

Because blocking moves are preemptive declinations, avoiding such moves so that a request can be made is an important matter.

Beyond getting to the request is another interactional issue that bears on acceptance rates. In a two-part “adjacency pair” sequence, an acceptance is a preferred response to a request.^ix Conversation analysts have established that when recipients of requests in ordinary conversation accept, they do so with a minimal gap, with items that occupy the entire turn and that state the acceptance semantically, and with little or no accounting or explanation (Heritage 1998: chapter 8; Schegloff 2007: chapter 5). By contrast, declinations are structurally “dispreferred” responses, and recipients usually delay them, use mitigating prefaces and, if they use a semantic rejection form, precede or follow it with other components in the turn, especially accountings or explanations. However, recipients often omit a semantic rejection and let the accounting stand in its stead. In the survey interview, when sample persons decline a request, they hesitate and embellish their utterance in various ways, but when they accept the request it is often a simple “Oh yah,” “sure,” “oh sure,” “yah,” or “okay” that appears relatively close to the request. In other words, the way in which speakers display the preferred or dispreferred character of a response results in acceptances, like good news, being “exposed” and declinations, like bad news, being “shrouded” (Maynard 2003). In some senses, preferred responses are interactionally facilitated and dispreferred ones are inhibited by these practices.

As acceptances are (structurally) preferred over rejections or declinations, the organization of the interaction order (Goffman 1983) may promote the granting of a survey request once that request is made. In other words, independently of a sample person's psychological attitude toward a survey, granting the request of an interviewer may have a positive leverage in its own right, one that is made salient by the articulation of the request itself. Preference structure can provide an interactional nudge towards acceptance.

Avoidance of Blocking Moves

Our point is that we can better understand the interactional practices that may alter the chances of an interviewer being able to avoid blocking moves and to produce the request if we investigate the dynamics operating in the opening moments of such calls. For example, is there anything an interviewer can do to affect whether a sample person will be encouraging or discouraging as the two parties progress through the interview introduction? Although we cannot now answer this question definitively, we can at least point to one line of direction for future research.

Interviewers’ opening turns regularly have at least four components: a greeting term, interviewers’ personal identification (name), their institutional affiliation, and a request for the sample person (Hollander 2008). Sometimes these components are produced in one turn of talk, but more often they are spread across two or more turns of talk and are produced in various orders. As we have seen, for this study, interviewers can use three additional components from their script—verification of the sample person's identity by referring to the high school they attended, mention of the WLS advance letter, and a brief description of the WLS study. Variation in the presence and order of these components and its relation to survey response is a topic of our larger project.

Here we focus on just one component: identification-recognition sequences. When they have reached a household by telephone, some interviewers follow the opening script by producing a personal (name) identification along with their institutional affiliation before asking to speak to the sample person, as in extract (5). In other cases, Hollander (2008) observes, the interviewer provides no personal identification or institutional affiliation (although the institutional affiliation can be implied by naming the study). For example, in extract 1, after the household phone is answered, the interviewer says simply, “Hi, could I speak to Brenda Caw please?” (see also extract 6). And in still other calls, personal identification and institutional affiliation are offered only after the interviewer asks for the sample person and the call recipient identifies herself as that person:

(8) HP008

FR: ↑He↓llo.
(0.5)
MI: tch ↑Hi there. ↑He↓llo. May I speak to Cindy Masterson.
(0.7)
FR: You ↑got er. Whadaya want?
(.)
MI: ↑Hi Cindy my name's Lou↑is Palmer. .h ↑I'm calling
back from University of Wisconsin (.) Survey Center
(.) about the Wisconsin Longitudinal Study?

In these instances (extracts 1, 6, and 8), the interviewer's relationship with the sample person is at least initially more anonymous (Heritage 2002:331) than if she had offered her own name in her first turn. In (8), the sample person takes a challenging stance (“Whadaya want?”) right after confirming her identity (and subsequently produces a blocking declination), and it is after this challenge that the interviewer identifies herself and the survey center.^x If interviewers identify themselves before asking for the sample person, as in extract (5), it may make salient a positive leverage intrinsic to the interaction order—namely, the reciprocity attendant upon offering one's name as a way of, or before, requesting a name from a recipient (Sacks 1989). If interviewers also reveal the name of the organization and under whose auspices they are calling, and it is one that informants or sample persons recognize and respect, it may activate the attribute of trust, which research has shown to promote survey participation (Roose et al. 2007). However, when interviewers immediately ask to speak to sample persons, then informants or the sample persons who answer the phone have no knowledge about who is calling. In these cases, the interviewer may appear less trustworthy—or less like the agent of a trustworthy enterprise—to the sample person. And if call recipients do identify themselves or indicate that the interviewer has reached the household of the sample person, they are not reciprocating with such identifying information but rather providing it to a so far nameless and amorphous other. That circumstance alone can affect the call recipient's setting of an encouraging or discouraging environment for requesting participation. Thus, we propose that how well interviewers work to establish reciprocity and trust in concrete, self-identifying ways may affect the likelihood of blocking moves, being able to produce requests (including tailored ones), and, ultimately, obtaining acceptance of the request.

CONCLUSION

Leverage-saliency theory posits that participating in surveys is a decision based on the leverages that different attributes of participating have and the salience of those attributes when the decision is made. As Groves and Couper (1998) have noted, tailoring on the part of an interviewer involves using sparse interactional cues first to infer leverages and then to avoid or mitigate (make less salient) those that are negative while highlighting (making more salient) those that are positive. Consistent with this view, we propose LST and Rational Choice Theory more broadly will benefit from research that examines requests and other collaborative activities not just as being affected by the sociocognitive states of persons but as actual, concrete actions. From our detailed examination of actual calls for participation in the survey interview, complicated dynamics emerge. When and whether requests are produced, for example, is a contingent matter. If cues from sample persons are encouraging, effective interviewers may move efficiently toward asking for participation and may produce requests that are appropriately presumptive. When cues are discouraging, they may postpone asking for participation while working through preliminary, scripted utterances, and then form their requests also in a cautious way.

Sample persons, for their part, may seize interactional opportunities to issue refusals before the request has been made. Skilled tailoring, therefore, does not just facilitate having a request to participate accepted, it also may enhance the likelihood of making a request in the first place. Once a request is generated, the preference for agreement may lend a positive leverage to the interview request that derives from the interaction order rather than attributes of the survey. But once a refusal occurs, matters are different. There may be an interactionally generated negative leverage if individuals have any aversion to reversing stated commitments (Dijkstra and Smit 2002). Additionally, although interviewers are provided with a detailed list of reservations that sample persons might have about participating and retorts to those reservations, sample persons do not necessarily articulate reservations in the form shown in survey materials. Worse, when interviewers in our data subsample used the most tempting structural place in the interaction to ask why sample persons declined to participate—after a refusal was produced—they never obtained an interview.

To what extent variation in interviewer practices, the interactional moves of sample persons, and the interrelation between these practices and moves have readily measurable effects on response rates awaits further, quantitative investigation. This study nonetheless highlights two challenges for such research. First, if practices are effective because of their deployment in particular contexts, then their effectiveness can only be assessed by experimental designs in which that context is considered. One cannot simply assign some interviewers to do presumptive requests and others to do cautious ones; what may instead be optimal is properly varying the presumptiveness and cautiousness of requests depending on the circumstances, something interviewers would need to be trained to recognize—and to do so very quickly. Second, observational studies of practices need to be careful not to confuse the influence of an interviewer's practices on a sample person with the influence of a sample person's behavior on an interviewer. A naive study of our data would conclude that presumptive requests work better, when our examination of how the interactions unfold suggests that presumptive requests are disproportionately produced in already-favorable contexts. The most (inappropriately) consistent and presumptive interviewer in our data was also the least successful one overall.

More generally, we have sought to demonstrate the importance of examining the actual details of interactions, especially of seeking to provide an interpretive understanding of the particulars of any given interaction using generic insights gained through basic research on the sequential mechanisms of conversation. Before counting and modeling possible effects on outcomes in our own study, we seek a better understanding of requesting as a social action in its own right. We do so by specifying a request in concrete interactional terms: it is, relative to offers, a dispreferred and delicate social action to perform. The delicacy of requesting is reflected in the range and variability of practices that enter into the design of this ubiquitous social action. And whereas previous CA research has focused on the design of requests relative to displayed inferences about entitlement and contingency according to the ordinary or institutional settings in which they are produced (Curl and Drew 2008:148), we have suggested that designs also can be intimately related to their preceding sequential and interactional contexts. Studies of requests and solicitations with different types of surveys (cross-sectional as opposed to longitudinal) and in other settings—calls for tissue donation (Weathersbee 2009), for example—can build upon our own and other CA inquiry.

Our research emphasizes the conversation analytic proposal that, if effective tailoring happens in telephone requests, then the information interviewers take into account in their practices cannot exist simply in sample persons’ heads but must be observable in the interaction. For interviewers, the work of tailoring is responding to displays of leverage and salience in real-time. Understanding the details of this work demands close attention to what happens, turn by turn, during these interactions. The payoffs include both a better understanding of requesting as a social action and ways that requests in specific settings are to be configured to achieve their preferred outcomes.