Statistical Modeling

The basic aim of statistical modeling is to derive a mathematical representation of the relationship between one or more response variables and a number of explanatory variables, including a measure of the inherent uncertainty of any such relationship. Whether your response variables are measured as continuous, categorical or count outcomes, we provide the optimal modeling solutions to extract the parameters of interest from your data. Our data analysts have years of experience in the following areas:

Multivariate Analysis Techniques

We offer full services and support in multivariate statistics.

Analysis of Data from Designed Experiments

In applied sciences, it is common to design experiments in order to obtain an unbiased estimation of parameters while controlling for confounding variables. We specialize in the following analysis methods for experimental data:

Analysis of Variance (ANOVA)
Analysis of Covariance (ANCOVA)
Repeated Measurements ANOVA
Linear Fixed-Effects Models for Continuous Outcomes
Mixed-effects Models for Replicated, Blocked Designs
Multi-level Models for Split-Plot Experiments

Categorical Data Analysis

Often the outcomes of interest are frequency counts of observations occurring in specified response categories rather than continuous variables but are expressed as frequency counts of observations occurring in the response categories. We apply modern statistical methods to analyze categorical data:

Logit Models - Logistic Regression
Loglinear Models for Contingency Tables
Models for Ordinal Variables
Multinomial Response Models
Exact Tests for Small Samples
Models for Matched Pairs

Analysis of Clustered or Longitudinal data

In applied sciences, one is often confronted with correlated, so-called clustered data, such as data from surveys with complex multi-stage sampling designs, data from cluster randomization trials, multivariate observations, repeated measurements, etc. Longitudinal data in which the outcome variable is repeatedly measured over time are a special kind of clustered data. In the analysis of clustered data, special statistical models are used to adjust for the stochastic dependence structure. We specialize in:

Linear Mixed Models for Repeated Continuous Outcomes
Nonparametric Models for Experimental Longitudinal Data
Marginal, GEE and robust longitudinal modelling approaches
Repeated Measurements Analysis of Variance
Statistical Methods for Cluster Randomization Trials
Multilevel Modelling of Complex Survey Data
Models for Overdispersed Count or Categorical Data
Models for Repeated Categorical and Count Outcomes

Survival Analysis

Survival data is a special kind of longitudinal data where the outcome of interest is the time to a specific event. One special feature of survival data, censoring, occurs when the event of interest is not observed for a study participant during the observational period. Analysis of censored data requires special statistical methodology. We have years of experience in the following analysis techniques:

Standard Survival Analysis Procedures

Nonparametric estimation of survival and hazard functions: Life tables, Kaplan Meier curves
Comparing survival distributions from different sample populations using log-rank and other global test procedures
Cox Proportional Hazards Regression
Parametric Modeling Approaches

Advanced techniques for multivariate survival data

Multi-state models for event history data
Competing risk models
Analysis of recurrent or multiple events