| ||||||||||||||||||||||||||||||||||||||||||||||||||||
Introduction
Nonresponse Bias in Household Surveys
ELEANOR SINGER is Research Professor Emerita in the Survey Methodology Program, Survey Research Center, Institute for Social Research, University of Michigan.
Address correspondence to the author; e-mail: esinger{at}isr.umich.edu.
The basis for unbiased inference from relatively small observed samples to largely unobserved populations is probability sampling. Such samples, in turn, provide the foundation for studies that inform national, regional, and local economic, social, and health policies, as well as marketing and political decisions (e.g., National Research Council 2006
). Underlying this inferential process, however, is the assumption that all elements designated for the sample are actually observed or measured. In recent years, increasing violation of this assumption because of an inability to contact sample members or because of their unwillingness to participate has led to increasing attention to survey nonresponse. "Nonrespondents" are those missing from a probability sample; the largest components of nonresponse, and therefore those of greatest concern to survey methodologists, are noncontact and refusal. These components have different causes and may, therefore, have different consequences for survey estimates (Groves and Couper 1998; Groves, this issue).
Concern about survey nonresponse is of course not new. Smith (2002, pp. 27–28) notes that "early research extends back to the emergence of polling in the 1930s and has been a regular feature in statistical and social science journals since the 1940s. An analysis of JSTOR statistical journals dates the first nonresponse article from 1945 and the Public Opinion Quarterly index’s earliest reference is from 1948. The index of Public Opinion Quarterly contains 125 articles on this topic; a full-text search of journals covered in JSTOR finds the following number of articles, by subject area, that included the word ‘nonresponse’: political science—62, economics—87, sociology—146, and statistics—431. . . . The vast majority of these articles focus on the main issues in the field: reducing nonresponse, measuring nonresponse bias, and compensating for nonresponse by imputation and/or weighting." Statisticians have been more concerned with ways of adjusting for the bias introduced by nonresponse; social scientists and survey methodologists have tended to focus on understanding and reducing nonresponse itself.
Neither is this the first special journal issue devoted to survey nonresponse. Within the last 10 years, there have been at least two others. The first, edited by Edith de Leeuw, appeared in 1999 as Volume 15, No. 2, of the Journal of Official Statistics. Growing out of a series of sessions organized at the Fourth International Social Science Methodology Conference in Essex in 1996, it included de Heer’s report of the results of an international survey on nonresponse trends, as well as numerous papers on incentives and interviewer strategies for reducing nonresponse, and several papers exploring its causes, concomitants, and effects. The second special issue, edited by Peter Lynn and growing out of a one-day conference organized by the Journal of the Royal Statistical Society in 2004, appeared in July 2006 as Volume 169, Issue 3, of the Journal of the Royal Statistical Society: Series A (Statistics and Society); it focused largely, though not entirely, on nonresponse in longitudinal studies and on methods of estimating and adjusting for bias in survey statistics resulting from unit nonresponse. By way of contrast, the present issue is devoted largely to the work of survey methodologists.1
Judging from papers presented at the International Workshop on Household Survey Nonresponse, which was organized in 1990 and held its seventeenth annual meeting in August 2006 (http://www.nonresponse.org), research interest in survey nonresponse since the mid 1980s, especially among survey methodologists, can be divided into roughly three periods, although of course there is considerable overlap among them, and all of them build on statistical and social science research going back many decades. One important question during the first period, extending roughly from the mid-1980s until the early 1990s, was whether response rates were actually declining and, if so, which components of response rates were responsible and how widespread the decline was.2 Thus, one of the workshop’s first initiatives was an international survey on nonresponse in official statistical surveys, begun in 1990 and repeated over time until 1997. Ultimately, this collaborative effort resulted in a article by Edith de Leeuw and Wim de Heer (2002), "Trends in Household Survey Nonresponse: A Longitudinal and International Comparison." De Leeuw and de Heer analyzed trends in overall response rates and their components on 10 different surveys (e.g., labor force, expenditures, health) carried out by 16 countries (mainly from Europe but including the United States, Canada, and Australia), though not all countries provided data for each survey. That analysis clearly established the ubiquity of declining response rates, though there was variation by country and by type of survey. Summarizing their findings, the authors state, "In sum: (1) countries differ in response rate; (2) the response rates have been declining over the years; (3) the trends differ by country; (4) there are no differences between countries in the rate [at] which noncontacts are increasing; and (5) the difference in response trends is caused by differences in the rate at which the refusals are increasing" (de Leeuw and de Heer 2002, p. 48; emphasis added).
The last year for which data are included in this international analysis of official (that is, government) statistics was 1996; Curtin, Presser, and Singer (2005) report that response rates on the Survey of Consumer Attitudes (SCA), a nongovernment survey carried out monthly by the University of Michigan, declined even more steeply from 1997 to 2003, when caller-identification technology became widespread, than from 1979 to 1996. Until 1985, noncontacts were a minor source of nonresponse on the SCA; thereafter, their role increased dramatically. But unlike refusals, the rate of increase in noncontacts has slowed in recent years, so that refusals now contribute more to response rate declines than noncontacts.
A necessary precondition for comparative and longitudinal research on response rates is a standardized way of accounting for the disposition of sampled cases. Hence, this period of research on the nonresponse phenomenon also saw the development of standard definitions in the way response rates are computed and reported. Under the auspices of the American Association for Public Opinion Research (AAPOR) and the leadership of Tom W. Smith, a committee including Barbara Bailar, Mick Couper, Donald Dillman, Robert M. Groves, William D. Kalsbeek, Jack Ludwig, Peter V. Miller, Harry O’Neill, and Stanley Presser made recommendations that were issued by AAPOR in 1998 as Standard Definitions: Final Dispositions of Case Codes and Outcome Rates for Surveys.3 The definitions have been updated continually to reflect new developments in the field; the current (fourth) edition was issued in 2006.
The period also saw the publication of two book-length studies of nonrespondents: Goyder’s The Silent Minority (1987) and Brehm’s The Phantom Respondents (1993). The first comprehensive text on nonresponse, Groves and Couper’s Nonresponse in Household Interview Surveys (1998), was also conceived in this period. Aside from its empirical studies of nonresponse in various contexts, the book offered a theoretical framework for studying and understanding the phenomenon that has guided much subsequent research, including the idea that nonresponse is a stochastic, rather than a fixed, property of individual respondents that is influenced by aspects of the survey made salient by the interviewer at the moment of interaction with the householder.
As is often the case, by the time the phenomenon of increasing survey nonresponse had been officially established, renewed efforts were already underway to reduce it. The second period of recent nonresponse research, which extended roughly from 1992 (when the Office of Management and Budget [OMB] asked COPAFS [Council of Professional Associations on Federal Statistics] to convene a symposium on the use of monetary incentives in household surveys4) until 2002, was characterized primarily by investigations of field work procedures for reducing noncontact as well as refusals. Thus, the period saw the publication of important articles on list-assisted sampling (e.g., Brick et al. 1995; Tucker, Lepkowski, and Piekarski 2002) and how to deal with answering machines (e.g., Link and Oldendick 1999; Oldendick and Link 1994; Piazza 1993; Tuckel and Feinberg 1991). The period also witnessed a lively debate on the ethics, appropriateness, and likely consequences of using monetary incentives to encourage reluctant respondents to participate in surveys (e.g., Groves, Singer, Corning, and Bowers 1999; Singer, Groves, and Corning 1999). Despite a lack of agreement on these issues, the use of incentives in interviewer-mediated surveys increased dramatically, both in the initial approach to respondents and for refusal conversion purposes, as did research on the effectiveness of incentives’ timing, amount, and kind (for a summary, see Singer 2002). A small separate branch of nonresponse research focused on training interviewers to avert refusals by means of tailoring the content of their interaction to concerns raised by reluctant respondents (Cantor et al. 2004; Groves, Cialdini, and Couper 1992; Groves and Maher 2004; McConaghy and Carey 2004; O’Brien 2004). On the statistical side, nonresponse research saw an emerging focus on multiple imputation to adjust for unit, as well as item, nonresponse (e.g., Heeringa, Little, and Raghunathan 2002; Marker, Judkins, and Winglee 2002; Rubin and Zanutto 2002), in addition to continued concern with issues of weighting and variance estimation in the presence of nonresponse (e.g., Bethlehem 2002; Gelman and Carlin 2002).
The publication of Survey Nonresponse, edited by Groves, Dillman, Eltinge, and Little (2002), which grew out of the International Conference on Survey Nonresponse held in Portland, Oregon, in 1999, marked the close of the second period of recent nonresponse research and ushered in the third, current, period.
The current period of nonresponse research reflects the reluctant recognition that despite increasingly costly efforts to make contact with designated households and persuade respondents to participate, response rates have not only continued their decline but also have done so at an increasing rate (Curtin, Presser, and Singer 2005). The preface to Survey Nonresponse acknowledges as much: "Declining cooperation rates increase the cost of conducting surveys . . . [and] can also damage the ability of the survey statistics to reflect the corresponding characteristics of the target population. . . . One of the important scientific challenges facing survey methodology at the beginning of this century is determining the circumstances under which nonresponse damages inference to the target population. A second challenge is the identification of methods to alter the estimation process in the face of nonresponse to improve the quality of the sample statistics" (Groves, Dillman, Eltinge, and Little 2002, p. xiii). The attention of survey methodologists therefore turned, inevitably, to an examination of the premise underlying the inferential paradigm of probability sampling: the relationship between nonresponse rates and nonresponse bias.
Several empirical studies had already called into question the necessary relationship between the nonresponse rate and the size of the nonresponse bias postulated by the standard formula (Curtin, Presser, and Singer 2000; Keeter et al. 2000; Merkle and Edelman 2002). But for any given study, it was impossible to tell whether the response rate was a good indicator of bias. Lacking, clearly, was a theoretical framework that might allow such predictions to be made.
Other things, too, would be needed in an environment in which low response rates became the rule rather than the exception. Analysts would need auxiliary data for adjusting their estimates; how could such data be obtained, and what kinds of data were most useful? Could information about the behavior of respondents, for example at the initial contact, be used not only to tailor the content of interactions between interviewers and respondents but also to aid in nonresponse adjustment? What kinds of statistical models are appropriate for inference in situations where a large proportion of designated sample elements is not observed by the researcher? What is the relationship between nonresponse error and measurement error, and how do they jointly affect the total error of survey estimates?
The articles in this issue focus on questions relevant to the concerns of this latest, current period of nonresponse research. For the most part, they aim to increase understanding of the sources of nonresponse bias, in order to use them as a basis for nonresponse adjustment. The lead article, by Robert M. Groves, directly tackles the central question of the relationship between nonresponse rates and nonresponse bias, concluding that there is no necessary connection between them; there is no minimum response rate below which a survey estimate is necessarily biased and, conversely, no response rate above which it is never biased. Furthermore, the bias can vary across different statistics in the same survey. The article examines how nonresponse bias arises and proposes ways of assessing whether it exists for a particular statistic in a given survey. It concludes with an examination of the case for probability sampling in the presence of high nonresponse and offers some practical advice for survey practitioners.
Three of the articles examine various aspects of the relationship between survey nonresponse and nonresponse bias. Abraham, Maitland, and Bianchi examine the possibility of nonresponse bias in the American Time Use Survey (ATUS), whose response rate has been below 60 percent. Because the ATUS sample is drawn from the outgoing rotation groups of the Current Population Survey (CPS), the authors have a "gold standard" against which to compare both respondents and nonrespondents on characteristics relevant to ATUS estimates. That exercise evidences very small biases due to ATUS nonresponse. Nevertheless, the authors acknowledge that other variables, which they were unable to examine (e.g., volunteering, measured on a CPS supplement), might have revealed much greater differences between respondents and nonrespondents and therefore a greater potential for bias.
For a variety of reasons, response rates on random digit dial (RDD) telephone surveys tend to be lower than those in face-to-face surveys, and estimating the effect of nonresponse on crucial survey variables is especially difficult in such surveys because of the paucity of information available on the sampling frame. Johnson and his colleagues suggest a possible way of supplementing this information by linking telephone numbers to zip codes, and thence to census tract–level variables, demonstrating the method on an RDD survey of illicit drug use in Illinois.
In "Experiments in Producing Nonresponse Bias," Groves, Couper, Presser, Singer, Tourangeau, Acosta, and Nelson extend the work of Groves, Presser, and Dipko (2004) by experimentally varying topic salience and incentives in a series of mail experiments and examining not only response rates but also the resulting impact on nonresponse bias in key survey estimates. The failure of the first two experiments to produce the expected results led to a series of refinements that demonstrate the importance of mode, confirm the role of topic salience in response, nonresponse, and nonresponse bias, and support the role of incentives in counteracting biases potentially resulting from topic salience.
The remaining articles deal with a variety of other issues related to survey nonresponse. Olson tackles a question often raised in connection with nonresponse bias but not satisfactorily addressed in the literature: namely, whether efforts to increase participation increase measurement error bias as the price of reducing nonresponse bias, and with what effect on total survey error. With the advantage of a data set containing information on level of effort, participation, and actual responses, as well as information from administrative records, Olson is able to show that estimates of total bias based on all respondents are generally lower than estimates based only on respondents who are most likely to be contacted and to cooperate. But the relationship between nonresponse bias, measurement error bias, and response propensity is statistic-specific, as well as specific to the type of nonresponse (i.e., noncontact or refusal). Olson also finds that error properties of statistics (in her case, means) may differ from the error properties of the individual variables used to calculate the statistics. Finally, she shows that concerns about increasing measurement error bias as a result of incorporating low-propensity respondents in the sample pool are borne out for some, but not all, statistics of the survey.
One of the earliest examinations of the consequences of lower response rates for nonresponse bias was a study by Keeter and his colleagues (2000), which showed that despite a drop in response rate from roughly 60 percent to 30 percent between a "standard" five-day RDD attitude survey and a "rigorous" survey using the same questionnaire but carried out over a much longer field period, very few variables showed significant change. In this issue Keeter, Kennedy, Dimock, Best, and Craighill show that despite a further decline in response rates, a replication of the comparison between "standard" and "rigorous" surveys carried out in 2003 produced significant differences on only 7 of 84 variables, and that the sample composition aligns closely with estimates from the U.S. census and from other large government surveys. The authors extend their analysis of nonresponse bias by examining the responses of those who refused twice or required at least 21 calls before being interviewed, as well as those who broke off the interview without completing it.
The article by Brick, Dipko, Presser, Tucker, and Yuan grew out of a study designed to address the problem of coverage rather than nonresponse. Increasing use of cell phones raises the question of how best to cover the combined cell and landline telephone population. Brick and his colleagues evaluate the use of a dual-frame RDD sample, consisting of both cell and landline numbers, for this purpose. To their chagrin, they discover that nonresponse biases in the two samples outweigh the potential reduction in coverage error, leading them to conclude that unless nonresponse bias can be addressed, the reduction in coverage bias from including cell phones may be more than offset by an increase in bias resulting from nonresponse.
One hypothesized cause of the marked increase in telephone survey nonresponse in recent years is the proliferation of telephone calls by marketers and fund-raisers, in addition to those by survey researchers. Thus, some survey researchers have hypothesized that the reduction in such calls resulting from the Do Not Call Registry might over time have a positive effect on telephone survey response rates. The final article in this issue, by Link, Mokdad, Kulp, and Hyon, examines the effects of the registry, instituted in 2003, on response rates to Behavioral Risk Factor Surveillance System (BRFSS) surveys carried out between 2002 and 2005. Using ARIMA time series analysis, they find no indication that the Do Not Call Registry has either increased or reduced response rates to the BRFSS surveys in the three years since its inception.
We are at the beginning of a new period of research on nonresponse bias, one that will have large consequences for survey practitioners and survey methodologists. A number of the articles in this issue have implications for "best practices" in an era of low response rate surveys, such as the collection of auxiliary data for purposes of adjustment, the use of multiple approaches to assess nonresponse bias on key estimates, and the avoidance of mechanisms for increasing response rates that simply exacerbate existing nonresponse biases. However, the fruits of these practices are in their infancy, and it remains to be seen to what extent they can reassure consumers of survey statistics about the validity of estimates derived from low response rate surveys and, at the same time, convince survey practitioners chafing under increasing costs that the advantages of probability sampling, at a time when low response rates are increasingly common, are worth the required investment of time and money.
In closing I would like to express my appreciation to all those who reviewed submissions for this special issue, as well as the authors, for their extraordinary cooperation in helping to get the issue out in record time. I also want to thank the editor of the Quarterly for his role in initiating the special issue and facilitating its progress at every step of the way, and the managing editor, for her invariable and invaluable help.
 |
Footnotes |
|---|
1. The International Conference on Survey Nonresponse, organized in 1999, tried hard to bridge the "two cultures" of nonresponse research, and the published volume resulting from that conference (Groves, Dillman, Eltinge, and Little 2002
) includes papers by both survey methodologists and statisticians. 
2. Groves and Couper (1998, pp. 156ff.) review early evidence bearing on this question.
3. Again, Smith (2002, p. 28) recaps earlier efforts to establish such standards, including two conferences organized in 1973 by the American Statistical Association with support from the National Science Foundation to "‘discuss the problems of present-day surveys . . . [and] explore whether or not these problems may now have reached a level or are growing at a rate that poses a threat to the continued use of surveys as a basic tool of social science research’" (quoting from the 1974 American Statistical Association report on the conferences). In the late 1970s the Council of American Survey Research Organizations (CASRO) and the Marketing Sciences Institute (MSI) launched a series of nonresponse studies that led to establishment of a CASRO task force on response rates and the issuing of a report, "On the Definition of Response Rates" (Smith 2002, pp. 28–30). But these early efforts did not come to fruition.
4. See http://members.aol.com/copafs/incentives/htm (accessed October 27, 2006).
|
References |
|---|
 TOP References |
|---|
Bethlehem Jelke G. (2002) Weighting Nonresponse Adjustments Based on Auxiliary Information. In Groves Robert M., Dillman Don A., Eltinge John L., Little Roderick J. A. (Eds.). Survey Nonresponse(Wiley, New York) pp. 275–88.
Brehm John. (1993) The Phantom Respondents(University of Michigan Press, Ann Arbor).
Brick J., Michael J., Joseph Waksberg, Dale Kulp, Amy Starer. (1995) Bias in List-Assisted Telephone Samples. Public Opinion Quarterly 59:218–35.[Abstract]
Cantor David, Allen Bruce, Schneider Sid J., Hagerty-Heller Tracey, Yuan Angela. (2004) Testing an Automated Refusal Avoidance Training Methodology. Paper presented at the annual meeting of the American Association for Public Opinion Research, Phoenix, AZ.
Curtin Richard, Presser Stanley, Singer Eleanor. (2000) The Effects of Response Rate Changes on the Index of Consumer Sentiment. Public Opinion Quarterly 64:413–28.[Abstract]
Curtin Richard. (2005) Changes in Telephone Survey Nonresponse over the Past Quarter Century. Public Opinion Quarterly 69:87–98.
de Leeuw Edith and de Heer Wim. (2002) Trends in Household Survey Nonresponse: A Longitudinal and International Comparison. In Groves Robert M., Dillman Don A., Eltinge John L., Little Roderick J. A. (Eds.). Survey Nonresponse(Wiley, New York) pp. 41–54.
Gelman Andrew and Carlin John B. (2002) Poststratification and Weighting Adjustments. In Groves Robert M., Dillman Don A., Eltinge John L., Little Roderick J. A. (Eds.). Survey Nonresponse(Wiley, New York) pp. 288–302.
Goyder John C. (1987) The Silent Minority(Westview Press, Boulder, CO).
Groves Robert M., Cialdini Robert B., Couper Mick P. (1992) Understanding the Decision to Participate in a Survey. Public Opinion Quarterly 56:475–95.[Abstract]
Groves Robert M. and Couper Mick P. (1998) Nonresponse in Household Interview Surveys(Wiley, New York).
Groves Robert M., Dillman Don A., Eltinge John L., Little Roderick J. A. (2002) Survey Nonresponse(Wiley, New York).
Groves Robert M. and Maher Patty. (2004) Implementation of Refusal Aversion Training in the National Survey of Family Growth. Paper presented at the annual meeting of the American Association for Public Opinion Research, Phoenix, AZ.
Groves Robert M., Presser Stanley, Dipko Sarah. (2004) The Role of Topic Interest in Survey Participation Decisions. Public Opinion Quarterly 68:2–31.
Groves Robert M., Singer Eleanor, Corning Amy D., Bowers Ashley. (1999) A Laboratory Approach to Measuring the Joint Effects of Interview Length, Incentives, Differential Incentives, and Refusal Conversion on Survey Participation. Journal of Official Statistics 15:251–68.
Heeringa Steven G., Little Roderick J. A., Raghunathan Trivellore E. (2002) Multivariate Imputation of Coarsened Survey Data on Household Wealth. In Groves Robert M., Dillman Don A., Eltinge John L., Little Roderick J. A. (Eds.). Survey Nonresponse(Wiley, New York) pp. 357–72.
Keeter Scott, Miller Carolyn, Kohut Andrew, Groves Robert M., Presser Stanley. (2000) Consequences of Reducing Nonresponse in a National Telephone Survey. Public Opinion Quarterly 64:125–48.[Abstract]
Link Michael W. and Oldendick Robert W. (1999) Call Screening: Is It Really a Problem for Survey Research? Public Opinion Quarterly 63:577–89.[CrossRef][Web of Science]
Marker David A., Judkins David A., Winglee Marianne. (2002) Large-Scale Imputation for Complex Surveys. In Groves Robert M., Dillman Don A., Eltinge John L., Little Roderick J. A. (Eds.). Survey Nonresponse(Wiley, New York) pp. 329–42.
McConaghy Mark W. and Carey Siobhan. (2004) The ART of Persuasion: A Controlled Experiment to Evaluate the Impact of Avoiding Refusal Training (ART) on Response in Britain. Paper presented at the annual meeting of the American Association for Public Opinion Research, Phoenix, AZ.
Merkle Daniel and Edelman Murray. (2002) Nonresponse in Exit Polls: A Comprehensive Analysis. In Groves Robert M., Dillman Don A., Eltinge John L., Little Roderick J. A. (Eds.). Survey Nonresponse(Wiley, New York) pp. 343–58.
National Research Council. (2006) Expanding Access to Research Data: Reconciling Risks and Opportunities(National Academies Press, Washington, DC).
O’Brien Eileen M. (2004) From Controlled Experiment to Production Environment: Refusal Aversion Training Adoption and Implications for Future Use and Research. Paper presented at the annual meeting of the American Association for Public Opinion Research, Phoenix, AZ.
Oldendick Robert W. and Link Michael W. (1994) The Answering Machine Generation: Who Are They and What Problem Do They Pose for Survey Research? Public Opinion Quarterly 58:264–73.[Abstract]
Piazza Thomas. (1993) Meeting the Challenge of Answering Machines. Public Opinion Quarterly 57:219–31.[Abstract]
Rubin Donald B. and Zanutto Elaine. (2002) Using Matched Substitutes to Adjust for Nonignorable Nonresponse through Multiple Imputations. In Groves Robert M., Dillman Don A., Eltinge John L., Little Roderick J. A. (Eds.). Survey Nonresponse(Wiley, New York) pp. 389–402.
Singer Eleanor. (2002) The Use of Incentives to Reduce Nonresponse in Household Surveys. In Groves Robert M., Dillman Don A., Eltinge John L., Little Roderick J. A. (Eds.). Survey Nonresponse(Wiley, New York) pp. 163–78.
Singer Eleanor, Groves Robert M., Corning Amy D. (1999) Differential Incentives: Beliefs about Practices, Perceptions of Equity, and Effects on Survey Participation. Public Opinion Quarterly 63:251–60.[CrossRef][Web of Science]
Smith Tom W. (2002) Developing Nonresponse Standards. In Groves Robert M., Dillman Don A., Eltinge John L., Little Roderick J. A. (Eds.). Survey Nonresponse(Wiley, New York) pp. 27–40.
Tuckel Peter S. and Feinberg Barry. (1991) The Answering Machine Poses Many Questions for Survey Researchers. 55: pp. 200–217.
Tucker Clyde, Lepkowski James, Piekarski Linda. (2002) List-Assisted Telephone Sampling Design Efficiency. Public Opinion Quarterly 66:321–38.[Abstract]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. C. Johnson and G. A. Kercher Identifying Predictors of Negative Psychological Reactions to Stalking Victimization J Interpers Violence, May 1, 2009; 24(5): 866 - 882. [Abstract] [PDF]
|
|
|
|
 |
|
 B. J. McMorris, R. S. Petrie, R. F. Catalano, C. B. Fleming, K. P. Haggerty, and R. D. Abbott Use of Web and In-Person Survey Modes to Gather Data From Young Adults on Sex and Drug Use: An Evaluation of Cost, Time, and Survey Error Based on a Randomized Mixed-Mode Design Eval Rev, April 1, 2009; 33(2): 138 - 158. [Abstract] [PDF]
|
|
|
|
 |
|
 R. Bautista, M. Callegaro, J. A. Vera, and F. Abundis Studying Nonresponse in Mexican Exit Polls Int. J. Public Opin. Res., December 1, 2007; 19(4): 492 - 503. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||