Analytical Techniques
Perhaps the most important aspect of designing research projects is the selection of appropriate analytical techniques. Not surprisingly, each type of analysis has a specific set of parameters that determine its suitability to a particular study design. Most can be deployed using in-person, telephone and web-based interview technologies.
The table below shows various types of analyses and the types of research that can benefit from each technique.
Analytical Techniques |
||
| Technique | What It Does | What It's Used For |
| Conjoint Analysis/ Choice Modeling | Allows consumer preferences for a product or service to be broken down into trade-offs among its individual attributes, without separating those attributes from the context in which overall judgments are made. | Optimizing product configurations; studying price elasticities of demand; simulating market response to new or modified offerings; diagnosing competitive strengths and weaknesses. |
| Analytical Hierarchy Process (AHP) | Analyzes trade-offs made in pair-wise comparisons of competing features, feature levels, products, etc. Used best when the number of items to be investigated is limited to seven or fewer. Can be implemented over the phone. | Determining "most-preferred" feature/level/product/service configurations; estimating shares among competing products/services. |
| Factor Analysis | Identifies a set of underlying dimensions ("factors") within a set of variables, revealing unobserved structure in the data. | Reducing the number of variables for analysis; identifying conceptual or benefit dimensions underlying expressed product perceptions and preferences. |
| Discriminant Analysis | Examines how two or more groups (generally of respondents) differ from one another on the basis of a number of predictor variables. | Understanding and modeling differences between/among groups: buyers vs. non-buyers, buyers of different brands, etc.; predicting market behavior based on demographic and psychographic variables. |
| Cluster Analysis | Uses any of several techniques to classify people, objects, or variables into more homogeneous groups. | Identifying/describing market segments; developing typologies; finding and describing target markets. |
| Regression Analysis (simple and multiple) | Studies the dependence of a single, interval scale variable (such as market share) on one (simple) or more (multiple) other variables. | Forecasting sales, market share, profitability; modeling buying patterns and impact of marketing programs; estimating elasticities and response functions. |
| Perceptual Mapping / Multi- dimensional Scaling | Illustrates the relationships between variables or groups of objects or people by placing them in a multi-dimensional space ("map"). | Uncovering "hidden order" in data: clusters, perceptual dimensions, competitors; evaluating positioning and image; facilitating product or advertisement placement. |
| Price Sensitivity Measurement | Establishes "least resistance" price and the range of acceptable pricing. | Setting price levels; developing competitive pricing strategies. |
| Quality Function Deployment (QFD) | Customer needs are identified and broken into their smallest components and studied for importance, opportunities and competitive advantages. Involves both internal and external research. | Focuses engineering and marketing resources on key features for maximum product impact. Keeps feature-rich products from being "over-engineered" so that new products are easy and intuitive to use. |
Conjoint Analysis
Conjoint Analysis is a "trade-off" exercise in which participants are forced to make preference decisions based on the relative importance of one product's features and attributes versus another product's features and attributes. A conjoint exercise forces the consumer to make "real world" decisions by looking at complete descriptions of an array of products and then selecting the one(s) they would choose in an actual "buying" environment. Conjoint is especially useful for complex trade-offs involving many choices, and is easily implemented through a card sorting or rating exercise.
Conjoint Analysis lends sophistication and accuracy to the process of product concept modeling. Conjoint also provides the researcher with a reliable method that can predict how study participants would evaluate untested products. The technique works very well in qualitative implementation, such as focus groups, but may also be used quantitatively in web surveys, mail studies, one-on-one interviews, and even in telephone research.
Analysis of Conjoint Results
In a conjoint design, every level of every attribute is compared with all other levels of all other attributes. Therefore, all product attributes are equally weighted. Subsequently, research outcome can:
1) Determine the RELATIVE IMPORTANCE of all attributes. Stated as a percentage, the relative importance represents the maximum amount of variation in the rankings or ratings of the cards, which can be explained by changes in the levels or values for that attribute. In other words, this is a measure of the degree of influence changes in attribute levels have on card rankings. The more important an attribute is, the more frequently cards with the most preferred level of that attribute will appear toward the top of the sorted deck (or show the highest ratings).
2) Quantify the RELATIVE SENSITIVITY to changes in the individual levels for each attribute, which is expressed as the "utility" score. Utility scores provide a clear means of revealing the sensitivities of the attributes and their various levels. In particular, utility scores permit the following analyses to be performed:
- Determine the level at which significant changes in importance, or value, occur.
- Allow comparisons of total utility scores for complete products, or comparisons of individual attribute levels.
- Utility scores for untested products, or untested attribute level values, can be predicted and compared through interpolation.
Conjoint's relative importance and utility scores allow us to model "What If?" scenarios and clearly see the likely marketplace results of features and feature-level trade-offs for conceptualized products.
Follow these guidelines -- use Conjoint Analysis when:
- Multiple product attributes in varying levels impact product desirability and manufacturing costs.
- The sophistication of the consumer's decision process transcends commodity level purchase behavior.
- The product configuration creates conflict/interaction among attributes/levels.
Price Sensitivity Analysis
Market Decisions' model for conducting Price Sensitivity Analysis requires respondents to relate the pricing of goods or services to perceived "value" rather than absolute price. It identifies the high and low range of acceptable prices and the Point of Least Resistance pricing. The basic steps taken to conduct the analysis are: 1) implementation of a value-based pricing question sequence worded in the context of the product, 2) accumulation of response percentages from computer tables and 3) observational analysis of plotted data.
The Question Series
A typical question series is shown below. Note how the questions are worded in the context of the product, in this case a flat rate billing system for telephone dial tone service.
Thinking just about the flat rate billing option we defined earlier (UNLIMITED CALLING FOR A SINGLE MONTHLY FEE), I'd like to know your opinion on the following pricing questions:
1 At what monthly flat rate for unlimited calling would you just start to feel that Extended Area Service to these communities is EXPENSIVE?2 At what monthly price would you just start to feel that this service is INEXPENSIVE?
3 At what price would you feel that this service is definitely TOO EXPENSIVE?
4 And, finally, at what point would you just start to feel that Extended Area Service to these communities was TOO CHEAP and you would have concerns about the company's ability to supply the service at this price?
Plotting the Data
Once all of the accumulations have been calculated, all data is plotted on a single line graph and specific observations regarding price sensitivity can be made based on the intersection points of the lines.
Analyzing the Results