Statistical Services

Statistics come in many forms. We want to ensure you get the very best out of your data, and understand what it is saying.

Types of Statistics in Data Analysis

This is not a full list, but gives you some idea of the range of statistics we deal with, from the relatively simple to the complex.


Method / Technique



Analysis of variance (ANOVA, MANOVA and MANCOVA)

General techniques that test whether means between groups are equal or not
Answers the question are these groups really different


BPTO (brand price trade off)

Respondents select preference for different brands at different prices


Canonical correlation

An extension of multiple regression, where there are multiple dependent variables as well as multiple independent variables



A decision tree model with multiple splits
Useful for identifying the demographic characteristics of likely buyers


Classification trees (decision trees)

An analysis that predicts group membership


Cluster analysis

Allocates records (people, businesses, etc) into mutually exclusive groups, where the members of a cluster are more like other members of the same cluster than members of any other cluster
Useful for identifying customer segments, attitudinal segments, and target markets


Conjoint  analysis

Measures the relative importance of attributes so that the most attractive offers are identified for target groups
Often used to evaluate new products, services or ideas



Level of association between two measures
Does not assume causality, but frequently used to identify key drivers


Correspondence mapping

A method of dimension reduction and perceptual mapping
It is used to compare brands, companies, segments in terms of their relative positioning


Discriminant analysis

Predicting the likelihood that an individual belongs to a series of groups or segments
Allocating individuals to the most likely group / segment
Identifying he key characteristics that distinguish each group / segment


Factor analysis

Reduces many variables to fewer  factors which represent the underlying dimensions of the data
It provides insight through the variable combinations, and clarity by reducing the number of variables
Often used to clarify key drivers, to assist in model building, and reduce data requirements from repeat surveys


Gabor Grainger

Respondents say how likely they are to buy a product at a series of prices
Provides measures of perceived value and price elasticity


Key driver analysis (KDA)

Identifies what measures have the most impact on a dependent variable, such as customer satisfaction, employee satisfaction, likelihood of choice


Locational analysis

Systematic method using evaluation of distances or cost-distance-time calculations for site planning and market  analysis
Identifies and evaluates catchment areas


Logistic regression analysis

Predicts the probability of group membership


Logit analysis

Produces linear probability models, usually used if predicting one of two alternatives
Used to model purchase intention


Multidimensional scaling (MDS)

Restructures respondent comparisons of similarity or preference into distances plotted in a multidimensional space
Usually displayed as a spatial or perceptual map


Multiple regression analysis

Understanding the relationship between multiple variables and how these influence a dependent variable, such as likelihood to buy, or customer satisfaction, or employee satisfaction
Measuring the likely impact of changes in the values of the influencing measures on the predicted variable, such as improved sales or higher satisfaction


Perceptual maps

The position of brands relative to their competitors, and the inter-relationship of brand attributes
Reduces multidimensional data to a two dimensional map
Methods include MDS and correspondence analysis


Price sensitivity meter (Van Westendorp)

Uses set of four questions to establish cheap and too cheap price, expensive and too expensive price for each respondent
Calculates range of acceptable pricing through the lowest and highest reasonable prices, indicating price elasticity


Principal components analysis

A form of factor analysis that uses correlation to identify super-variables


Quadrant charts

Divides results into four, identifying what is important and needs maintaining, what is important and needs improving, what is less important and needs reduced investment, and what is not important and requires no change
Often used to present KDA results
Useful to compare segments on key differentiators


Social network analysis

Understanding relationships in terms of volumes, direction, and spread. For example, transactions between companies, buying and selling behaviour, communications between group members.


Structural equation modelling / path analysis

Maps the relationships between variables using factor analysis, canonical correlation and multiple regression to evaluate predictive power


Time series analysis

Used both for forecasting and influences over time to analysis a series of events / measures
For example, advertising spend over time on sales
Underlying seasonal variances are generally removed


TURF (Total Unduplicated Reach and Frequency Analysis)

Used to provide estimates of market potential and identifying the optimal mix of products / flavours / options

Looking for something that isn't listed? Get in touch and see if we can help - the list is not exhaustive!




Model Sum of Squares df Mean Square F Sig.
1 Regression 36.197 9 4.022 8.844 .000
Residual 55.024 121 .455    
Total 91.221 130