What's New at LI
LI Staff
Research Projects
Training Services
LI Bookstore
Drug Use Trends
Research Tools
Research Presentations & Posters
Staff Publications
Illinois Addiction Studies Archive
Links
CHS Home > Services > Research & Training > RESEARCH TOOLS
Lighthouse Analytical Training Series
The documents listed below were developed by Lighthouse Institute staff under a variety of grants (see individual acknowledgements) to help researchers over come specific problems associated with the collection and analysis of data generated during substance abuse treatment research. They are available free for downloading.
| Author: | Melissa McDermeit, Rodney Funk and Michael Dennis |
| Name: | Data Cleaning and Replacement of Missing Values |
| Description: | To provide methods of handling missing and inconsistent data. Consistency codes that tell the analyst the type of missing data, will be covered first. Then the replacement of those codes based on type of question and other cleaning issues will be discussed. The methods described herein are based on data cleaning done with the Global Appraisal of Individual Needs (GAIN; Dennis, 1998). |
| Date: | 6/24/1999 |
| Author: | Melissa McDermeit, Rodney Funk and Michael Dennis |
| Name: | Merging Files: Adding Variables or Cases |
| Description: | To show by example how to add new variables or new records to an existing data file and check the results |
| Date: | 6/24/1999 |
| Author: | Rodney Funk and Michael Dennis |
| Name: | Intra-class Correlation for Test-Retest Reliability and/or Stability |
| Description: | Standard texts on generalizability theory and psychometrics (e.g., Hayes, 1988; Lord & Novick, 1968; Nunnally & Bernstein, 1994) often recommend testing reliability with an intra-class correlation coefficient. This index represents the percentage of variance explained in the score by individual (or other unit of observation) and ranges from 0 to 1. It can be used to estimate the stability of measure over time (test-retest reliability if a short enough time) and/or to partial-out systematic variation from other sources (e.g., variations in raters, context) that would otherwise be included in the error in a Pearson Product Moment Correlation. Our aim here is to provide an example of how to use SPSS (Version 9.0, 8.0.1 or 7.5) to estimate the intra-class correlation coefficient and its 95% C.I. The first example is from data first appearing in Potthoff and Roy (1964) and is also used in an example in the SPSS Advanced Statistics 7.5 manual (SPSS, 1997). This example shows how to estimate the intra-class correlation coefficient with four time periods. The second example shows how when calculated on two groups (test-retest), the intra-class correlation coefficient is equal to the Person Moment correlation coefficient, while also showing the difference between reliability and stability. |
| Date: | 6/24/1999 |
| Author: | Melissa McDermeit, Rodney Funk and Michael Dennis |
| Name: | Means Tables |
| Description: | To show by example how to run simple means and tables of means and other descriptive statistics. |
| Date: | 2/18/2000 |
| Author: | Melissa McDermeit, Rodney Funk and Michael Dennis |
| Name: | Exploratory Factor Analysis with alpha method and varimax rotation |
| Description: | One of the uses of Factor Analysis is for the development and validation of scales within an inventory or test battery. Factor Analysis can be used to identify groups of similar items, create scales, and, thereby, reduce the number of variables used in further analysis. Factor Analysis can also be used to test the factor validity of test inventory or battery to make sure that items are more associated with the scales to which they are assigned that other scales or tests in the battery. |
| Date: | 1/05/2000 |
| Author: | Rodney Funk, Melissa Ives and Michael Dennis |
| Name: | Cluster Analysis |
| Description: | Cluster analysis is an exploratory technique used to develop a typology for describing the major subgroups in a population that is heterogeneous on more than one dimension of variables, settings, or over time. The aim of this memo is to explain the method the authors typically use to obtain a cluster analysis and its subsequent validation using SPSS (Version 7.5 to 14.0). |
| Date: | 3/21/2006 |
| Author: | Melissa Ives, Rodney Funk and Michael L. Dennis |
| Name: | Accessing Pooled GAIN data |
| Description: | To provide information about the kinds of data sharing that is possible among grantees or with those conducting secondary analyses and to provide information about HIPAA requirements prior to data sharing. This memo provides information about data sharing agreements (DSAs) and creating an abstract to obtain permission to use pooled GAIN (Dennis et al., 2003) data. |
| Date: | 4/19/2007 |
| Author: | Melissa Ives, Rodney Funk and Michael L. Dennis |
| Name: | Acknowledgement of funding and data sources for GAIN data |
| Description: | To provide instructions and information needed to create appropriate acknowledgements of those who provided data, wrote or reviewed the data or text, or provided procedural oversight for articles, books, presentations, etc. (Includes sample text and a table of CSAT and CHS studies where the GAIN (Dennis et al., 2003) was used. |
| Date: | 4/19/2007 |
| Author: | Rodney Funk , Melissa Ives and Michael L. Dennis |
| Name: | Imputation of Index Scores with Missing Data |
| Description: | To provide a preferred method for calculating values for summative index scores with missing data. |
| Date: | 4/23/2007 |
| Author: | Melissa Ives, Rodney Funk and Michael L. Dennis |
| Name: | Tukey Box Plots |
| Description: | Explains how Tukey box can create a graphic image of the several key measures of distribution, including the minimum, maximum, median, and 25th and 75th percentiles (the middle 50%). |
| Date: | 4/23/2007 |
| Author: | Melissa Ives, Rodney Funk and Michael L. Dennis |
| Name: | Discriminant Analysis |
| Description: | Discriminant function analysis is used to categorize records into mutually exclusive groups based on a series of predictor variables. The results from the classification function coefficients are then used to create a calculation rule to assign groups for new cases where the grouping value is unknown. |
| Date: | 4/23/2007 |
| Author: | Melissa Ives, Rodney Funk and Michael L. Dennis |
| Name: | Survival Analysis/Life Tables |
| Description: | Survival analysis, including Life Tables and Hazard analysis, is a way of examining the time to an event (e.g., discharge, admission/readmission, symptom onset, task completion, death) for one or more groups. |
| Date: | 4/23/2007 |
| Author: | Rodney Funk, Melissa Ives and Michael L. Dennis |
| Name: | Reliability: Calculating Cronbach's Alpha |
| Description: | While the SPSS Reliability command provides multiple options and models, this memo presents the typical format for obtaining and using Cronbach’s alpha and other statistics to assess a scale and its component items. |
| Date: | 4/23/2007 |