Concept: Statistics for Large Databases

Introduction

This concept provides a list of statistical tools and approaches that have been used at MCHP. Many of them are particularly useful in working with large databases. Researchers and programmers might look here for an introduction to these tools and to identify valuable resources.

SAS Macros

• MCHP SAS Macros - a list of SAS macros developed at and/or used by MCHP staff for doing calculations and data manipulation. The list contains the name of each macro, a brief description of the macro, and a link to sample SAS programming code.

Descriptive Analysis

• Confidence Interval of the Median - a concept describing how to measure the confidence interval of the median.
  
  
• Rates of Utilization - a concept describing how to compute rates of events, typically by geographical regions.
  
  
• Variance - a concept describing the technique used in statistical analysis to calculate the degree of variance of collected data.

Regression Analysis

• Logistic Regression - concept on the relationship between a response variable and one or more explanatory variables where the response variable follows a binomial distribution.
  
  
  • Also see: Logistic Regression: Calibration of the Data Model (Hosmer-Lemeshow Goodness-of-Fit Statistic)
  
  
• Tobit Regression
  
  
• Topics in Regression Analysis for Discrete Data (See Topics in Regression Analysis for Discrete Data - (internal access only) - prepared by Lisa Lix and Arane Thavaneswaran, 2008) - includes topics such as:

• Logistic Regression and Conditional Logistic Regression models with examples and SAS syntax
  
• Poisson regression models with example and SAS syntax
  
• Model evaluation (goodness of fit, discrimination, validation)
  
• Risk ratio and odds ratio

Significance Testing

• Population Data & Significance Testing: A Brief Discussion - a concept providing responses to a question regarding the relevance of significance testing when using population-based data.
  
  
• Multiple Comparisons, including the Bonferroni Correction - concept describing a statistical method that adjusts the significance level when multiple comparisons are made.

Special Approaches

• Cochran-Armitage Test for Trend
  
  
• Constructing Income Quintiles
  
  
• Generalized Estimating Equations
  
  
• Gini Coefficient
  
  
• Intraclass Correlation Coefficient
  
  
• Lorenz Curve
  
  
• Prevalence and Incidence - a concept defining and describing how to calculate prevalence and incidence rates, as well as the issues involved in these calculations.
  
  
• Random Effects Models - concept for longitudinal data analysis (including statistical notation for defining these models).
  
  
• Receiver Operating Characteristic Curve - Sensitivity and Specificity
  
  
• Statistical Analysis of Wait Time
  
  
• Survival Analysis

MORE INFORMATION

• Swinscow TDV (Revised by Campbell MJ) (1997) Statistics at Square One Ninth Edition. BMJ Publishing Group [ website ]
  
  
• Statistical Help on the Web:

• The Internet Glossary of Statistical Terms [ website ]