YouGov Consulting is supported by a dedicated quantitative analytics team; it is our job to help provide insight for clients. By working closely with the various consultancy sectors we develop a good understanding of the needs of a particular client, the needs of their business and the objectives of the research.

Involved from the proposal stage to the final presentation of findings, we use our expertise to provide guidance throughout the research process so as to ensure that the data collected is suitable for the application of advanced statistical techniques that can be used to inform our response to the needs of our clients.

By using advanced techniques we can go beyond merely describing the data, we can begin to explain and even predict attitudes, behaviours and harder business outcomes. These explanations and predictions can help our clients to inform a number of things, including their own behaviour, both internally and externally, decisions about their products/policies and how they approach their marketing, communications and people strategies.

The team uses a number of advanced techniques in order to achieve the maximum value from the data collected. Depending on the needs of the project any one or combinations of these techniques may be used. The techniques include:

• Regression Analysis: a technique that allows us to model the effect of one or more variables on something that we are trying to explain, for example we can address questions such as: What are the key determinants of customer / employee satisfaction? What factors drive profit levels? What are the key drivers of audience viewing figures? Why do people use a particular product?
• Cluster Analysis: used to distinguish between groups of people on the basis of attitudes and/or behaviour. This information can be used in developing marketing strategies, such as which product(s) or services to target to which group(s). Examples of groups you might look at are: first-time visitors/buyers, repeat visitors, returning customers, new registrants, respondents to certain marketing campaigns, employees by sector or grade.
• Factor Analysis: used to reduce the amount of information analysed by picking out underlying dimensions within the data. An individual’s score on a dimension can be incorporated into other models to predict behaviour and/ or attitudes or be used to segment respondents. Examples where factor analysis may be useful is picking out dimensions of leadership attributes from a list of characteristics, looking at how various attributes of an insurance policy group together or looking at what types of food produce people buy.
• CHAID Analysis: a way of visually representing the relationship between variables in the form of a tree diagram. It can be used both to predict an outcome and segment respondents by how they answer the relevant survey questions. For example we can show the effect of price, size, features of a product on purchase of that product by identifying discrete groups of people based on their responses to these variables and looking at the impact on purchase.
• Conjoint Analysis: a way of measuring how people respond to trade-offs between different attributes when, for example, considering which product to buy or service to invest in. Other examples include the trade-off between location and waiting time for a particular service, perhaps medical care, and for a company the trade-off between price and market share.
• Correspondence Analysis / Mapping: a way of visually presenting the relationship between categorical variables, such as brand and attribute. The technique can be used to map products, services or organisations against any number of traits, features or attributes.
• Structural Equation Modelling: builds on regression analysis allowing us to model relationships where causal effects may be reciprocal or mediated. For example, someone’s attitudes towards a particular product may influence their willingness to buy that product but their willingness to purchase may also influence their attitudes; in looking at what causes absenteeism in the workplace it may be that both socio-demographics and job satisfaction have an effect, but also the former may have an indirect effect through the latter.
• Monte Carlo Simulations: provide a way of modelling uncertainty by using probability distributions to run scenarios in order to establish a range of possibilities rather than relying on point estimates. For example if you want to model the risks involved in financial transactions or the relationship between supply of a particular product or service and consumption then you might want to know what will happen if demand falls or the cost of production increases. These effects can be modelled using simulations.