MAR 4832 Exam 1

What is the role of exploratory research—such as focus group interviews—in conducting a conjoint study (also, how are such interviews conducted, and what is a discussion guide)?

- Focus groups show their preferences and perceptions toward a set of products and that is analyzed and converted into value system that shows the importance and desired value of each attribute -Guide of questions for conducting a focus group discussion

What is the purpose of Maximum Difference Scaling?

- Forces consumers to make comparisons among the alternatives - Used for measuring the importance or preference within a list of items

What is the difference between full-factorial and partial-factorial conjoint study designs?

- Full-Factorial: all combinations are shown to user/Sawtooth is different - Choice Based Conjoint: they make a choice on combinations Partial-Factorial

How is conjoint analysis different from MaxDiff?

- MaxDiff: the items are scaled on a single dimension (or attribute) - Conjoint tackles multiple attributes

Are there conditions under which conjoint analysis may not be appropriate?

- Product cannot be a bundle of attributes - category is revolutionary or customers cannot reasonably rate the product-if a firm cannot change the attribute levels-trade - off is reasonable = yes conjoint, no tradeoffs available = no Conjoint

Which attributes and levels tend to be useful for a given conjoint study?

- Attributes are characteristics of a product. Levels are the degrees of the characteristics - Attribute = Color, Level= Red, Purple, Orange - Attribute levels should cover the full range of possibilities for existing products as well as products that may not yet exist

What is a dummy variable?

- Takes only the value 1 or 0 to indicate the absence or presence of some categorical effect that may change the outcome. - Shows preference levels for each consumer, on different attributes and levels

What is assumed in conducting such simulations?

- all attributes that affect buyer choices in the real world have been accounted for - each product has equal availability (distribution) - respondents are aware of all products - the products reach long range equilibrium - equal effectiveness of sales force - no out-of-stock conditions

MaxDiff tends to be more useful under what conditions

- the list has more than 7 items - but fewer than 30

How might conjoint analysis help the brand manager?

- what aspects of product drives customer's brand choice - Conjoint simulators predict respondents' interest in product - "what-if" games investigate the value of modifications to an existing product or alternative, and investigates product line extensions

What is a good criterion to use in selecting among the approaches? Is there a preferred approach?

-Logit is more predictable (find part worth for attributes and add to constant) -Highest Utility (First choice approach) -Logit (More Accurate)

Why is it important to consider competitors' products in conducting simulations?

Because we do market simulations to predict market shares

What might be a weakness in such (traditional) approaches that we discussed in class?

Doesn't mimic the real world or take into account trade-offs

Why might some software provide no constant term in the CA report (alternatively: how do we generate coefficients analogous to the commercial software from the Excel output for traditional CA)?

It gives a value for each level instead

Given the Excel output, how is the total value of a specific bundle of attributes (with specified levels) determined?

K+(Dummy1 x Utility1)+(Dummy2 x Utility2)

How are 'Linear' and 'Logit' based simulations different?

Linear contains dependent variable Logit has limited # of possible values

What is a design matrix (note: you are expected to write the design matrix corresponding to a given set of attributes and levels) and what is its role in conducting conjoint analysis (CA)?

Shows preference levels for each consumer, on different attributes and levels

What are the steps to conduct traditional CA using Excel for a sample of respondents?

Step 1: Design number of: attributes, levels for each attribute, and combinations shown to consumers Step 2: Design matrix using dummy variables. Rule: required dummy variables = no. of levels- 1 dummy variables can take on only values 0 or 1 Step 3: Use LINEST to Count number of dummy variables; add 1 to that number, write names in reverse order, Select as many cells as that total Step 4: Interpretation (and comparing answers)two decimal places sufficient

Full-Factorial Design

all combinations are shown to user/Sawtooth is different

Market Simulations

competitive market scenarios to predict which products respondents would choose

The terms "Connectivity", "Frequency Balance", "Orthogonality" and "Positional Balance" refer to what aspects of a MaxDiff study (and how does software such as Sawtooth help in this context)?

crucial aspects of the study design that ensure the data collected is reliable and accurately reflects respondent preferences, and software like Sawtooth helps by automatically generating designs that optimize these factors, minimizing potential bias and maximizing the information gathered from each respondent.

How many dummy variables do we need to represent the levels of an attribute?

dummy variables Rule: required dummy variables = no. of levels- 1 dummy variables can take on only values 0 or 1

Logit Model

estimated share of, say, product A is: eVA / eVA | eVB | eVC

Linear Model

estimated share of, say, product A is: VA/ VA | VB | VC

Importance Scores

measure how much influence each attribute has on someone's choice

Utilities (part worths)

numeric values that reflect the desirability of different features

How is the relative importance of different attributes obtained?

Range of one attribute divided by sum of all attribute ranges

Three Models for the Simulations

First-choice Model Linear Model Logit Model

A typical product may have many features (e.g., think of a product like a cellphone);are all those features explicitly used in a typical conjoint study?

No, only the most important ones

How should one use the resulting Excel output—in particular, how does one interpret the regression coefficient of a dummy variable, and what is the role for a constant term in the regression equation?

The coefficient reflects effect relative to the benchmark category so we have to add the constant to get the total effect

First-choice Model

The product with the highest total value is always selected

Identify the distinct 'outputs' one obtains from conjoint analysis

Utilities Relative Importance of Attributes Simulations