Qualitative Research Initiative

Areas of activities:
Consultancy
Advice on survey design
Discussion-guide design
Recruitment/Moderation
Transcriptions
Analysis & interpretation
Specialist viewing
Areas of concerns

- understanding the needs of the project
- adding value at all stages of the research process from input on project specification to questionnaire design to analytical options to data interpretation
- focus on analysis rather than reportage of facts and figures
- provide guidance and recommendations on the implications of the research
- reasonable expectations in terms of what can be achieved and within what timeframe
- reports and presentations to be concise, incisive and focused on clearly communicating what is actionable for future tactical and strategic planning

Research techniques:

Focus groups
Depth interviews (face-to-face/via telephone)
On-site mini-depths
Accompanied shops/ visits
Advertising evaluation
Citizens Juries
Home Visits
Mystery Shopping
National and International research
New Store Concept evaluation
Projective Techniques
Web surfs/ accompanied surfs
Workshops


Statistical Techniques:


CORRELATION ANALYSIS

What is correlation analysis?
Correlation analysis technique that looks at the indirect relationships in survey data
When would you use it?
To objectively establish which variables are most closely associated with a given action or mindset
What are the advantages?
It can provide a more discriminatory analysis than asking a direct question
Any disadvantages?
Has potential shortcomings when dealing with mixed scales
Anyone who has ever carried out a survey in which they ask respondents to rate the importance of various aspects of a product or service invariably finds that most people say everything is very important. What this disguises is that while respondents say everything is very important, in reality some things are more important to them than others.
This is where correlation analysis comes into play as it looks at the indirect relationships between variables and can help in objectively assessing the extent to which one variable really influences another.
The starting point for correlation analysis is to identify a ‘dependent variable’ – for example, overall satisfaction – and then to see to what extent the responses given to each of the variables correlates to the responses given to overall satisfaction. This analysis takes place at respondent level and enables us to establish, for each aspect, how closely related it is to the dependent variable – this is measured by the co-efficient of correlation.
The correlation co-efficient is usually scored between 0 and 1; a score of 1 would mean there was complete correlation between responses, a score of 0 would mean there was none. The higher the co-efficient, the greater the correlation.

Regression/Multiple Regression
What is regression analysis?
Regression is a technique used to predict the value of one variable based on results of one or more other variables
When would you use it?
To work out the simultaneous impact of more than one variable at a time
What are the advantages?
Allows you to work out ‘what if …’ scenarios
Any disadvantages?
Good predictive powers cannot be guaranteed. Intercorrelation of data can mean that not all variables are included in the regression model. Works best with binary variables (i.e. ‘yes’ / ‘no’ responses)
Regression analysis is used to help us predict the value of one variable from one or more other variables whose values can be predetermined.
The first stage of the process is to identify the variable we want to predict (the dependent variable) and to then carry out multiple regression analysis focusing on the variable(s) we want to use as predictors (explanatory variables). For example, the dependent variable might be overall satisfaction, the explanatory variables price, quality, value for money, delivery time and staff knowledge.
The multiple regression analysis would then identify the relationship between the dependent variable and the explanatory variables – this is presented as a model (formula) that might look like this:
Overall satisfaction =
1.37 x price rating + 0.91 x quality rating + 0.64 delivery time rating
+ 2.42 (a constant)
Invariably not all of the possible explanatory variables are included in the model due to inter-correlation between them: for example, the ratings that people give on price and value for money may be very closely correlated and are therefore not both required in the formula.
The overall predictive powers of the model can be calculated and expressed as the co-efficient of determination R2 (= the explained variation / total variation). The co-efficient of determination will lie between 0 and 1: 1 would mean that it is able to explain 100% of the variation although a figure of less than 50% is more common.

FACTOR ANALYSIS

Factor analysis is all about reduction. Simply put it reduces a large number of variables to a smaller number of variables.
Factor analysis does this by looking at the relationships (i.e. correlations) between the responses to each of the larger number of variables and grouping them together in combinations of variables that are closely correlated (i.e. they ‘behave’ in similar ways).
Factor analysis is extremely useful where you are dealing with a very large number of variables that would be cumbersome, time-consuming or simply impractical to analyse individually as it reduces them into smaller, homogeneous groupings. However, the compromise is some loss of sensitivity although this is usually more than compensated for by increased usability.
Factor analysis is widely used as input into other analysis – for example cluster analysis/market segmentation

Cluster Analysis

What isclusteranalysis?

Cluster analysis – also known as market segmentation – is a technique that is used to measure market composition

When would you use it?

To provide an alternative, more focused profile to what would be possible using basic socio-demographics, or other single-dimensional measures

What are the advantages?
It provides a classification that primarily describes the make-up of the market in attitudinal or behavioural terms
Any disadvantages?
While members of each cluster group share the same characteristics, each member is not all necessarily identical to every other member

Cluster analysis is often referred to as market segmentation. It is a technique that is used to help establish market composition by sub-dividing in into discrete groups (known as ‘clusters’).
While conventional ‘demographic’ analysis is based on tangible characteristics such as sex, age and social class, cluster analysis primarily relies on either subjective elements such as attitude, motivation, aspiration etc or on behavioural traits such as awareness, trial, weight of usership etc.
The rationale for cluster analysis is to sub-divide the sample into homogeneous groupings (i.e. clusters), each of whom share as many characteristics as possible with each other, while being as ‘different’ to everybody else as possible. This it does by looking at the response patterns and relationships between responses (for example from a large attitude battery such as TGI or series of factual responses based on awareness, trial and usership). In the interpretation of cluster analysis we should be aware that while everyone in the same cluster group shares the same broad traits and characteristics, they are not necessarily identical.
Cluster analysis allows us to identify the number and nature of different customer groupings within the market. By establishing the needs, requirements, opportunities and threats presented by each one we can ascertain their current and future [potential] worth to our business.

CHAID
What is CHAID?
CHAID stands for CHi-squared Automatic Interaction Detector. It is a technique that detects interaction between variables.
When would you use it?
It is usedto identify discrete groups of consumer and predict how their responses to somevariables affect other variables
What are the advantages?
Highly visual output, no equations.
Any disadvantages?
Needs large sample sizes to work effectively
CHAID detects interaction between variables in the data set. Using this technique we can establish relationships between a ‘dependent variable’ – for example readership of a certain newspaper – and other explanatory variables such as price, size, supplements etc.
CHAID does this by identifying discrete groups of respondents and, by taking their responses to explanatory variables, seeks to predict what the impact will be on the dependent variable.
CHAID is often used as an exploratory technique and is an alternative to multiple regression, especially when the data set is not well suited to regression analysis.
It is a highly visual means of data presentation that commonly takes the form of an organisation chart and does not entail any formulae or equations.
CHAID does not work well with small sample sizes as respondent groups can quickly become too small for reliable analysis



Source: http://www.icmresearch.co.uk/specialist_areas/qualitative.asp#research