Third Millennium Analytics, Inc. can assist with the development of data management protocols, codebook construction, merging of data sets, and developing syntax/coding for index or scale construction. Each of these services is described in more detail below.
Data Management Protocols. To ensure reliable, high-quality data are collected, written guidelines should be developed for data entry procedures, to delineate data security protocols, data coding schemes, coding reliability assessment, and for the editing and cleaning of data. A well-designed data management system should help with the training of staff and the management of data quality audits.
Codebook Construction. Data codebooks are essential information sources for variable coding, as they provide a concise listing and descriptions of the variables in a dataset. They also often contain information describing the study, sampling information, technical information (number of observations, number of records per observation, etc.), structure of the file, details about the data (e.g., where specific variables can be found, etc.), and often contain text of the items/questions and response categories.
Importing/Exporting Data Files. Assistance can be provided with the safe and efficient conversion and transfer of data files. Files often need to be imported and converted (e.g., from excel files, text files, or complex files of mixed, grouped, or nested data) for use with various statistical programs. As some statistical programs are better suited for some tasks or techniques, we can assist with format conversions from one statistical program to another.
File Operations. We can merge multiple data files (i.e., files with same cases but different variables as well as files with the same variables but different cases), aggregating data, weighting data, and transposing cases and variables (i.e., changing file structure, which is often needed when creating longitudinal data sets).
Metadata Programming. With proper documentation, we can program variable properties (e.g., variable labels, value labels for discrete variables, setting missing values), as well as document file properties.
Data Transformations. Various common data transformation procedures can be performed, including the recoding of categorical variables, “binning” of scale variables, simple numeric transformations, arithmetic and statistical functions, manipulation and coding of string variables, setting date and time functions, etc.
Cleaning and Validating Data. Data validation reports can be prepared, which help identify invalid values and duplicate entries. Invalid values are checked to identify possible keystroke errors or errors in data entry. If no keystroke/coding errors are found, assistance can be provided for decisions about whether the invalid data should be excluded from the analysis. If IBM® SPSS® Data Collection is used, data entry and data validation are automated, which accurately and efficiently keep your data clean.
Special Data Processing. Our special data processing services include conditional processing, looping, and repeating functions.
Scale/Index Construction. Scales with well-established and documented psychometric properties can be identified and recommended. Scales can also be adapted or slightly revised, if needed, to meet particular needs. We can also help conceptualize and operationalize attitudinal and behavioral measures (i.e., scales and indices) as well as assess the reliability and validity of your measures.
Missing Data Analysis. Missing data is a common problem that, if not dealt with properly, can lead to statistical power issues, and/or biased and misleading results. We assess randomness of missing data, and understand the implications of applying “available case” methods, likelihood based ignorable analyses, and various multivariate techniques for imputing missing values or statistical estimates.