## Regression Models for Decision Making: A Cost-Benefit Analysis of Incentives in Telephone Surveys (2000)

Citations: | 1 - 1 self |

### BibTeX

@MISC{Stevens00regressionmodels,

author = {Andrew Gelman .Matt Stevens and Andrew Gelman and Matt Stevens and Valerie Chan and Marcia Meyers},

title = {Regression Models for Decision Making: A Cost-Benefit Analysis of Incentives in Telephone Surveys},

year = {2000}

}

### OpenURL

### Abstract

A standard method for reducing survey nonresponse is to offer incentives to survey participants. We reanalyze the data from a recent meta-analysis (Singer et al., 1999), using a hierarchical regression model, in order to estimate the effects of prepaid and postpaid incentives for face-to-face and telephone surveys. We then apply the results of our fitted model to the New York City Social Indicators Survey, a biennial telephone survey with a high nonresponse rate. We consider the balance of estimated costs, cost savings, and response rate for different choices of incentives. In general, we propose a methodology for evaluating some cost-benefit decisions in sample survey design. Keywords: decision analysis, hierarchical linear regression, informal Bayes, meta-analysis, survey nonresponse 1 Introduction What with over 90% of U.S. households having telephones, and the convenience of random digit dialing and computer assisted interviewing, telephone interviewing has become increa...

### Citations

1240 | Bayesian Data Analysis
- Gelman, Carlin, et al.
- 1995
(Show Context)
Citation Context ... The simplest task is to model ff, for which we use independent normal random effects: ff jsiid N(0;s2 ); wheresis assigned a uniform prior distribution (as in the hierarchical models of chapter 5 of =-=Gelman et al., 1995-=-). There is no loss of generality in assuming a zero mean for the ff j 's if a constant term is included in the set of predictors X . We choose the predictors X and model the coefficients fi using an ... |

331 |
Variable Selection via Gibbs Sampling
- George, McCulloch
- 1993
(Show Context)
Citation Context ...r, this strategy requires more research in setting up a reasonable class of prior distributions. A related approach is the formal selection of subsets of predictors using Bayesian methods (see, e.g., =-=George and McCulloch, 1993-=-, Madigan and Raftery, 1994, Draper, 1995), but we doubt these methods to be appropriate for our problems here for two reasons. First, formal Bayesian selection rules they tend to select relatively fe... |

265 | Models selection and accounting for model uncertainty in graphical models using occam’s window
- Madigan, Raftery
- 1994
(Show Context)
Citation Context ...re research in setting up a reasonable class of prior distributions. A related approach is the formal selection of subsets of predictors using Bayesian methods (see, e.g., George and McCulloch, 1993, =-=Madigan and Raftery, 1994-=-, Draper, 1995), but we doubt these methods to be appropriate for our problems here for two reasons. First, formal Bayesian selection rules they tend to select relatively few predictors, which in this... |

150 | Survey errors and survey costs - Groves - 1989 |

128 |
Sampling: Design and analysis
- Lohr
- 2010
(Show Context)
Citation Context ...e size, but more importantly it can cause bias in estimation and lack of credibility of final survey results. Known sources of bias can be corrected using weighting and poststratification (see, e.g., =-=Lohr, 1999-=-), but some bias will remain to the extent that nonrespondents and respondents differ systematically after adjustment. Credibility is harder to quantify, but we would all feel more comfortable reporti... |

111 |
Assessment and propagation of model uncertainty (with discussion
- Draper
- 1995
(Show Context)
Citation Context ... reasonable class of prior distributions. A related approach is the formal selection of subsets of predictors using Bayesian methods (see, e.g., George and McCulloch, 1993, Madigan and Raftery, 1994, =-=Draper, 1995-=-), but we doubt these methods to be appropriate for our problems here for two reasons. First, formal Bayesian selection rules they tend to select relatively few predictors, which in this case could le... |

17 |
Bayes methods for combining the results of cancer studies in humans and other species
- DuMouchel, Harris
- 1983
(Show Context)
Citation Context ...hen doing this, we also shifted over to a hierarchical structure to directly model the 101 response rates y i and thus handle the concerns at the end of the list in the previous section. (As shown by =-=DuMouchel and Harris, 1983-=-, a hierarchical model allows for the two levels of variation in a meta-analysis.) We start with a binomial model relating the number of respondents, n i , to the number of persons contacted, N i (thu... |

11 | The Effect of Incentives on Response Rates in Interviewer-Mediated Surveys - Singer, Hoewyk, et al. - 1999 |

9 |
Bayesian regression with parametric models for heteroscedasticity
- BOSCARDIN, GELMAN
- 1996
(Show Context)
Citation Context ... estimate the vector fi, so for mathematical simplicity we combine (1) and (2) to yield the (approximate) model, y isN((Xfi) i + ff j(i) ; oe 2 + V i ); (3) where V i = y i (1 \Gamma y i )=N i . (See =-=Boscardin and Gelman, 1996-=-, for similar heteroscedastic hierarchical linear regression models.) Finally, we must specify the predictors X and the model for the vectors fi and ff. The simplest task is to model ff, for which we ... |

5 |
Making Hard Decisions, second edition
- Clemen
- 1996
(Show Context)
Citation Context ...tly lay out the costs and benefits of proposed strategies such as incentives in the context of the particular survey. This is in line with the general recommendations of decision analysis (see, e.g., =-=Clemen, 1996-=-) that the key step is to enumerate decision alternatives and consider their expected consequences. We found this perspective to have implications in setting up regression models whose parameter estim... |

5 |
A new perspective on meta-analysis
- Rubin
- 1989
(Show Context)
Citation Context ...xperimental but generally not assigned randomly (value, timing, form). This is a common problem when a meta-analysis is used to estimate a "response surface" rather than simply an average ef=-=fect (see Rubin, 1989-=-). A third potential difficulty arises from the clustering in the data (between two and five observations for each experiment), but this was easily handled using a hierarchical model as discussed in S... |

3 | Methods for Obtaining High Response Rates in Telephone Surveys.” Paper presented at the Workshop on - Cantor, Cunningham - 1999 |

3 |
A Tale of Many Cities: The New
- Garfinkel, Meyers
- 1999
(Show Context)
Citation Context ... the estimated coefficients. We discuss modeling issues further in Section 4.1. 3 Application to the Social Indicators Survey We apply the fitted models to the New York City Social Indicators Survey (=-=Garfinkel and Meyers, 1999-=-), the survey that motivated this work. The survey was divided into two strata: an individual survey, which targeted all families (including single adults and couples), and a caregiver survey, which w... |