Statistical services: an explanation of techniques
Key Driver Analysis (KDA)
One of the main objectives of market research is the measurement of customer satisfaction. The technique most often employed is regression analysis (often referred to as a Key Driver Analysis). The objective of the analysis is to determine which components of overall satisfaction are the most important.
Digitab offers a wide range of methodologies to suit your needs. These include the more familiar analysis methods - linear/non-linear or logistic regression and
more recently developed techniques including:
- Shapley Value Regression
This is a relatively new technique, which Digitab has developed into an analytical tool for examining how independent variables (components of overall satisfaction) overlap with each other and how this affects the dependent variable. This tool overcomes the problem of multi-collinearity in regression analysis. It also has the advantage of producing easily understood results, i.e. the contribution of each independent variable as a percentage of the whole (out of 100%).
- Latent Class Regression
Latent class regression combines the two analysis objectives: KDA and segmentation (see below), into one step. It fits regression equations to segments exhibiting similar response patterns.
- TABOO
Targeted Bootstrapping (TABOO) is a simulation technique that measures the effect of changes in one or more variables on a dependent variable. Unlike ‘conventional’ statistical methods, TABOO does not define a theoretical model but works only with the data, by selectively re-sampling the data many times.
- Structural Equation Modelling (SEM)
SEM is a complex model, merging regression and factor analysis. It’s a dedicated method to confirm or build a statistical model based on a theoretical model.
Segmentation
As a consequence of the increasing diversity of customer needs and behaviour and the variety of products on the market, segmentation has become one of the most popular statistical analyses.
By dividing the relevant market into smaller parts that are internally more homogeneous and externally heterogeneous it is easier to develop products which more closely meet individual consumer needs and to create appropriate promotional campaigns to reach each target group. Furthermore it is also possible to identify groups of similar products which give companies information about the competitive products closest to their own product.
Depending on the researcher’s objectives, the following statistical methods can be applied by Digitab:
- the traditional method – Clustering (Non-hierarchical and Hierarchical techniques)
- the confirmatory method – discriminant analysis
- analysis requiring a target variable – Automatic Interaction Detection (AID)/ CHAID
- a more recent method – Latent Class Analysis (See above)
- other techniques partly related to segmentation such as Conjoint or mapping techniques
Perceptual and Preference Maps
Once the researcher has decided which customer groups within which market segments to target, he has to determine how to present the product to this target audience. Positioning shows how the brands/ products/ services are perceived by customers.
There is a wide range of mapping techniques which Digitab can offer to present customers’ perception including:
- Correspondence Analysis and Multidimensional Scaling
- Principal Component Analysis/ Factor Analysis
Launching a new product
In order to develop new products and services which will be competitive in the company’s relevant markets it is necessary to know individual customer needs and preferences. Therefore it is essential to first identify which attributes of a specific product consumers will prefer and then the likelihood that consumers will buy the product. Conjoint Analysis may be used to measure perceived values of specific product features and to forecast what the likely acceptance of a product will be.
Pricing Research
One of the prime determinants of a product’s success in the market place is clearly its price. The analysis of price is a key area in market research.
Digitab has a wealth of experience in this area which includes the following techniques:
- Brand Price Trade Off
- Gabor Granger
- Van Westendorp
- Price Elasticity of Demand
Other areas of research
- Data mining (Neural Networks, Genetic Algorithms)
- Simulation models
- Time series and forecasting models
