Competitive Topography is a turnkey solution created to explore a number of entities, such as brands, that can be rated on a list of attributes.

Default Setup

In a default setup, respondents are presented with a series of screens, each for a separate entity, asking them to rate attributes using a 5-star rating. The rating scheme can be changed to 7-star or 9-star rating. Alternatively, researchers may use Sliders instead.

The ratings are then processed and displayed in three main views: Multidimensional Scale, Topography View, and Quadrant View. The rating process consists of two stages. First, a simple transformation is applied — each star increment gives 10 score points for star rating, and each slider increment adds 10 score points for slider rating. Second, Bayesian Averaging is applied over the mean scores to prevent excessive values at low sample sizes:

Where parameter C is set to 100, and m is mean attribute score. More: Wikipedia — Bayesian average.

Multidimensional Scale

The Multidimensional Scale view displays results of the Nonmetric Multidimensional Scaling (NMDS) ordination method. Each entity and attribute is placed on a 2D plane while preserving their pairwise distances in the original multidimensional entity portfolio / brand matrix. It is important to note that for the purposes of the ordination, a portfolio of attributes in each brand is studied.

If an entity point is close to an attribute point, then the entity has a higher share of the attribute in its portfolio compared to other entities. If an entity point is far from an attribute point, the entity has a lower share. If an entity is close to another entity, their attribute portfolios are similar. If an entity is far from another entity, their attribute portfolios are different.

Topography View

This is an extension of the Multidimensional Scale view. Besides the original 2D NMDS points for entities and attributes, a third dimension is used to display average scores for either entities or attributes. In this view it is possible to find patterns in entity and attribute evaluation.

Quadrant View

Quadrant View is designed to give freedom of selecting an arbitrary slice of data and presenting it on a 2D plot, with the possibility of additional variables being added through point size and color.