The study was conducted by the Schneider Institute for Health Policy, Brandeis University. Westat, Inc., collected and prepared the data.
ADSS files underwent disclosure analysis by SAMHDA to ensure that the identities of facilities and clients were protected. This involved reviewing the data files for potential risks as well as examining any external threats to confidentiality, such as other data sources that could be linked to ADSS. Such external data sources were found. To address this problem while still creating a public use file of the greatest utility possible, micro-aggregation of certain variables was performed. This involved identifying the problematic variables, sorting records by the first problematic variable, organizing records into groups of three based on their value for this variable, averaging the values for each grouping, and applying the average to the records in each group. This was repeated for each of the problematic variables, which included client count and financial data. Geographic identifiers were also removed. The overall impact of these protection procedures was small and should not affect most analytic uses of the data.
The Phase I facility public use file includes 2,394 of the original 2,395 records. One facility’s record was deleted due to the presence of outlying data.
Client records can be matched between corresponding Phase II and Phase III abstract and follow-up data files using the CASEID variable. Facility data can be matched across the Phase I, II, and III data files and cost study data files using the FACID variable.
The unit of analysis for the Phase I facility (Part 1) and Phase II administrator (Part 2) data is the facility. Data from the Phase II and Phase III abstract and follow-up files (Parts 3–8) are analyzed at the client level. Analyses for the cost study (Part 12) are conducted at the modality or "type of care" level within facilities.
Please note that the unit of time for some variables in the facility file is specified in a separate variable, and these units are distinct from each other. For example, to analyze length of treatment, the researcher needs to examine two variables: QUANTITY VAR NAME and UNIT VAR NAME. QUANTITY specifies the "quantity" of treatment length, while UNIT specifies the unit of QUANTITY, such as days, weeks, months, years, or sessions.
The Finite Population Correction Factor and the two Stratified Jackknife Factor data files are provided for use with the WesVar and SUDAAN statistical software and are not intended for use with other statistical packages. WesVar was developed by Westat, Inc., and SUDAAN is a product of the Research Triangle Institute. These three files are being distributed as received from the principal investigator and have not been tested by SAMHDA.
The data from the follow-up Incentive Study in Phase III are not released as part of this public use collection.