Regression Analysis

Cluster Analysis – A technique of analysis that attempts to classify sample persons into a small number of segments (types, clusters) on the basis of a number of characteristics. Essentially, the persons within a given segment are made to look as similar as possible (as far as the stipulated characteristics are concerned); at the same time, the segments should be different to the greatest possible degree. See also: Cluster

Common Factor – In a factor analysis, each factor that occurs in two or more variables is called a common factor. When the factor occurs in all variables, it is called a group factor. A factor that occurs in only one single variable is termed specific. See also: Cluster

Conjoint Measurement – Syn: Conjoint Analysis A statistical research method that involves the measurement of the collective effects of two or more independent variables (product attributes) on the classification of a dependent variable (“over liking,” purchasing intention, “best buy,” or any other evaluative measurement). The stimulus is a combination of product-attributes. They may be presented in the form of descriptions, illustrations, concept or prototypes of products. Respondents are requested to study the set combinations and to classify or evaluate them according to the dependent variable selected. Because each stimulus is a combination of attributes, the classification or evaluation reflects the balancing of the conflicting product-attributes by the consumer. For example: do they prefer a large, powerful spacious automobile that is relatively expensive in its operation, or one that is smaller, less powerful, but more economical to operate? In the ultimate analysis the combinations of attributes are broken down.

Contrast Analysis – An analysis of research material in which it is attempted to form groups that are strong or of maximum contrast. For example: buyers of a specific product versus non-buyers, men versus women, young women (up to 30 older) versus older women (31 years and older). The analysis may be conducted either manually or (usually) by computer. Such an analysis is useful in the determination of the target group for a product or service. For example: what are the maximum differences between Coca Cola buyers and non-buyers. The “ Automatic interaction detector” is an example of a contrast analysis. See also: Target group determination / Automatic interaction detector

Covariance Analysis – A statistical method employed to reduce data from experiments or research projects with two or more variables that have been measured in different groups. This method uses (the concepts of ) analysis of variance and regression analysis simultaneously. See also: Analysis of variance / Regression analysis

Factor Analysis – A mathematical – statistical method used to simplify as series of analysis (for example : information from respondents) to primary, mutually independent, characteristics. This method makes it possible to reduce the number of features that characterize the respondent. The results of the factor of analysis may be graphically represented in a system of coordinates.

Factor Loading – In a factor analysis: 1. The regression of a response of an individual to an item of factor. 2. The weight assigned to a factor by a model in order to determine the response of an individual to a question. See also: Factor analysis / Regression analysis / Model

Linear Regression Analysis – A statistical method used to predict the value of a quantitative variable from the non-quantitative or categorical scores of correlated variable.

Multidimensional – The property of having more than two dimensions. Measured by a number of dimensions, magnitudes, variables, units of measure. For example: to measure an object according to the variables of length, width, height and weight. See also: Dimension / Unidimensional

Multidimensional Analysis – An analysis of data involving more than two variables. A number of diverse, advanced techniques are available for this purpose. See also: Dimension

Multidimensional Scaling – The utilization of scales on many dimensions (variables) simultaneously. Special techniques are available for this purpose. The aims is to unravel a complex whole (of variables) as it occurs in reality. See also: Multivariate techniques/Non-metric Multidimensional Scaling/Scale

Multivariate – Comprised of several (at least three) variables. See also: Multivariate Techniques

Multivariate Techniques – Syn: Multivariate Methods Advanced statistical techniques that involve simultaneous processing of several ( more than two) variables. For example: Simple and multiple regression techniques, factor analysis, cluster analysis. See also: Multidimensional Scaling/ Factor Analysis/ Cluster Analysis

N.M.S. – Abbreviation for non-metric multidimensional scaling. See also: Non-metric multidimensional scaling

Non-metric Multidimensional Scaling – This technique involves non-metric data, that is, nominal (A is similar to B) or ordinal (C is greater, more, better, etc. than D) data. Respondents make statements concerning the similarity of or their preference for products, brands or concepts. In this manner, a measure of similarity or preference is established. These data are placed into a matrix and the computer executes the analysis. The computer searches for the optimum structure (as small a number of dimensions as possible) that describes the coherence of the data and renders them capable of interpretation. See also: Nominal Scale/Interval Scale/Multidimensional Scaling

Orthogonal Rotations – A procedure in factor analysis that attempts to establish the simplest description of the independent factors exposed by the analysis. See also: Factor Analysis

Principal Component Analysis – A form of factor analysis that yields the smallest number of factors required to reproduce the original measurements. See also: Factor Analysis

Regression Analysis – Statistical techniques that measure the degree of coherence between variables. For this purpose, the researches uses his knowledge concerning the independent variables (for example: advertising) to determine the magnitude of the dependent variables (for example: sales) See also: Correlation/Independent Variables/Dependent Variables

Segmentation Analysis – Statistical procedure designed to distinguish market segments of the population that are relatively homogeneous, for example, the purposes of the advertiser. Such segments maybe approached in different ways (different media, different concepts) and, if necessary, different products can be presented. See also: Segment/Segmentation


Leave a Comment