Proposal Track
Research Project
Session Format
Presentation
Abstract
Researchers often use publicly available complex surveys as a data source. These data can provide useful insights into the description of educational settings and outcomes, but can present certain problems that should not be ignored. A complex survey is complex because of the sampling method employed by the survey designer, the analysis of such a survey must take the sampling method into consideration.
A common problem occurs when a researcher wishes to study a subset of the surveyed population. Often the predictors will have a standard error that is inflated, leading to (a) large interval estimates or (b) insignificant tests.
This presentation will example some of the causes of inflated variance and will examine the differences in estimation that are produced when using two software packages, R and SAS. The methods that are available for both programs when sub-setting will be reviewed.
Keywords
Statistical Methods, Complex Survey
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Daniel, Patricia, "Issues in Complex Survey Analysis: Why is Subsetting a Problem?" (2018). Georgia Educational Research Association Conference. 11.
https://digitalcommons.georgiasouthern.edu/gera/2018/2018/11
Included in
Issues in Complex Survey Analysis: Why is Subsetting a Problem?
Researchers often use publicly available complex surveys as a data source. These data can provide useful insights into the description of educational settings and outcomes, but can present certain problems that should not be ignored. A complex survey is complex because of the sampling method employed by the survey designer, the analysis of such a survey must take the sampling method into consideration.
A common problem occurs when a researcher wishes to study a subset of the surveyed population. Often the predictors will have a standard error that is inflated, leading to (a) large interval estimates or (b) insignificant tests.
This presentation will example some of the causes of inflated variance and will examine the differences in estimation that are produced when using two software packages, R and SAS. The methods that are available for both programs when sub-setting will be reviewed.
