Consulting

Survey Analysis Consultation

We offer consultation on many aspects of survey analysis. Common topics of consultation include:

Weighting and Estimation

  • Constructing weights for unequal probability samples that appropriate account for unequal probability sample selection, nonresponse and frame coverage errors
  • Determining the appropriate variance estimation and point estimation techniques
  • Imputing missing data or otherwise compensating for missing data in statistical analyses
Consultants: Chris Wiesen

Analysis Software

  • Applying SAS/SUDAAN programming techniques to produce publishable tables and graphs quickly and efficiently
  • Outputting data into a variety of formats, including Microsoft Word, Excel and WordPerfect
  • Using statistical software packages, including: Stata, Lisrel, Mplus, Amos, MLWin, LEM and SPSS
Consultants: Chris Wiesen

Nonsampling Error

  • Analysizing nonsampling error in surveys including compensating for nonresponse, measurement error modeling and the evaluation of survey error
Consultants: Paul Biemer, Chris Wiesen

Compensating for Nonresponse

  • Increasing response rates through better survey and experimental design, improved interviewer training and supervision
  • Using incentives and more private data collection modes. However, some nonresponse is inevitable in any survey.
  • Reducing nonresponse bias in analysis through weighting adjustments, logistic regression imputation and response propensity modeling
Consultants: Paul Biemer, Chris Wiesen

Measurement Error Modeling

  • Models for studying measurement bias under complex sampling schemes
  • Multilevel models for describing the effects of interviewers, coders, and other survey personnel on survey response
  • Probability models for assessing the validity of self-reported drug use
  • Models for describing the effects of measurement and nonresponse error on categorical data
Consultants: Paul Biemer, Chris Wiesen

Evaluation of Study Error

Odum researchers have developed innovative experimental designs and methods for evaluating response error. We design small-scale cognitive laboratory studies that provide insight into the thought processes affecting the quality of answers to questionnaire items. We also design field studies to test proposed survey and study procedures under realistic conditions, as well as to evaluate the components of the mean square error, including:
  • Re-interview survey design for estimating reliability and response bias
  • Experimental designs for cognitive laboratory research
  • Designs and statistical methods for comparing alternative modes of data collection
  • Statistical models for nonsampling error structures
Consultants: Paul Biemer, Chris Wiesen