Types of Driver Analysis
Types of Driver Analysis
Driver analysis is an analysis of the association of brand attributes with brand performance, from which we infer which of these drive brand choice.
Type of analysis includes
1. Stated Importance
To find out how important an attribute is in the decision making process, we could just ask the consumer directly
The question might ask the respondent to rank the attributes, or use a scale, or even to list just the x most important
2. Regression Analysis
Regression analysis is the most widely used of all statistical techniques
This technique establishes the relationship between the independent variables (attributes) and the dependent variable (usage/commitment etc.)
Weights (?) are given to each independent variable to indicate its importance in determining the dependent
In essence, the method involves the drawing of a line that minimizes the distance between the line and the observed data points
3. Partial Least Square (PLS)
PLS simultaneously extracts one set of latent variables for the set of manifest independent variables and another set of latent variables for the set of manifest dependent variables. PLS extracts the latent factors in order to combat the multicollinearity that is likely to exist within the independent variables.
The cross-product matrix for the independent and dependent variables is established. The X-scores of the independent latent variables are then used to predict the Y-scores or the dependent latents, and these predicted Y-scores are then used to predict the manifest dependent variables. From this process, PLS produces the weight matrix that reflects the covariance structure between the independent and dependent variables.
These weights are the values that are then used to determine the importance of the independent variables or attributes (and hence the numbers that appear in the ranking of the attributes).
4. Jaccard Coefficient
Jaccard is not strictly a statistical technique, but rather a simple measure of association
It makes use of an attribute association grid (brands by attributes), with binary or dichotomous data
5. CHAID Analysis
CHAID is a classification technique that evaluates complex interactions between variables and displays them in an easy-to-interpret diagram
It also gives an indication as to which variables have the biggest impact on the dependent variable
Chi-square tests decide how the