- Corrections to standard errors (SE) in the regression programs (multiple, logit and probit) for surveys based on complex sample designs. Design-based corrections to SE in SDA were required in order to obtain improved significance testing and valid statistical inference.
- Precision-based (PB) specifications for suppression of cell percentages in the crosstabulation program and cell means in the means program. This tool is designed to alert researchers to estimates that have low precision (e.g., the analysis is based on too few cases with a large standard error).
- SDA output for crosstabs and frequencies in XML, which allows greater flexibility in graphical displays of data.
The Center for Behavioral Health and Statistical Quality (CBHSQ, formerly the Office of Applied Studies), the Substance Abuse and Mental Health Services Administration (SAMHSA) supported the development of SDA 3.4 and 3.5 through the Substance Abuse and Mental Health Data Archive (SAMHDA) contract at ICPSR.
For more details, including a list of key contributors, see the latest issue of the ICPSR Bulletin.
The Bulletin also includes a description of ICPSR's new guidance on data management plans and a summary of participation in our Virtual Data Fair earlier in November.
