STATISTICAL ANALYSIS TECHNIQUES

Pan Atlantic SMS Group uses a wide range of standard and advanced statistical techniques to ensure accurate data analysis and the development of value added strategies and implementation plans for our clients.

Standard Analysis

Included as a deliverable of most research projects which we conduct are standard statistical analyses designed to simplify data. These include demographic breakdowns, frequencies of responses, mean responses and correlations, or relationships between variables.

Advanced Statistical Analysis

When appropriate, Pan Atlantic SMS Group also provides advanced statistical analysis to clarify data and support recommendations. These include:

Structural Equation Modeling (SEM)
SEM compares a set of data to a theoretical model. This is a statistical technique for building and testing theories of behavior, attitudes, etc. that employs a variety of specific statistical analyses including factor analysis, path analysis and multiple regression. In a way, SEM represents the overall analysis for which other statistical tests are employed. SEM is used to:

• Explain data (and the variables within that set) in terms of actionable results; and
• Use datasets or specific variables to predict future outcomes.

Path Analysis
Path analysis seeks to define relationships within and among variables to create predictive models. It commonly involves the use of diagrams of behavior, attitude formation, etc. to map the causes and effects that exist among a set of variables. It then attempts to account for the direction and significance of these relationships.

Multiple Regression
Often referred to as the simplest type of path analysis, regression analysis describes the relationships of each variable to other individual variables or sets of variables. By determining these relationships, regression analysis can be used as predictors of future data. For this reason, multiple regression is also known as analysis of key drivers; regression models can assess drivers of behavior, attitudes, etc.

CHAID Analysis
CHAID (Chi Squared Automatic Interaction Detection) is used to construct a predictive model based on classifying data into meaningful subgroups. CHAID divides the sample into groups that share similar characteristics and that have some predictive value with regard to an outcome variable.

Cluster Analysis
Cluster analysis is the process of identifying unique sub-groups or segments within a larger population. It searches for patterns in data to identify homogeneous groups or demographic segments and is often used in social and customer-related research.

Factor Analysis
Factor analysis involves analyzing data to simplify complex patterns into simpler or fewer factors. It can reduce a large set of variables down to a few key factors that can be used as a basis for future actions and recommendations. Factor analysis is useful for determining the underlying issues that may not be immediately obvious.

Perceptual Mapping
Perceptual mapping is a technique that provides a visual display of data to reflect the perceptions of respondents. For example, this analysis might use data that assesses dimensions such as product attributes and provides a meaningful representation of attributes with relation to competing products. This technique is useful in product positioning and consumer behavior studies.

Data Mining
Data mining involves analyzing existing data for meaningful relationships or patterns that may have been excluded from previous searches. There are two ways of doing this:

• Querying databases in order to satisfy or find data to support a hypothesis
• Querying databases in order to generate new hypotheses based on rigorous statistical correlations or requirements