Study Design & Sampling

CSR provides technical expertise in research design and probability sampling procedures. We typically use true probability sampling procedures for studies. This can involve area probability sampling of Census tracts or blocks for in-person studies in the New England area, random-digit-dialing procedures for national telephone studies (typically a list-assisted approach using the GENESYS system), methods for randomly sampling from lists, or address-based samples for national mail surveys. CSR also has experience with dual-frame and multi-mode designs. Stratification and clustering are used within sample designs if it is necessary for the aims or for the cost structure of a study. If such complex sampling methods are used, all data files clearly identify the strata and/or clusters so they can be properly accounted for in analyses. CSR also produces estimated design effects and sample weights due to the complex sampling.