Help
Options
Didn't know it?
click below
click below
Knew it?
click below
click below
Don't Know
Remaining cards (0)
Know
retry
shuffle
restart
0:00
Embed Code - If you would like this activity on your web page, copy the script below and paste it into your web page.
Normal Size Small Size show me how
Normal Size Small Size show me how
Marketing research
Lecture 3
| Question | Answer |
|---|---|
| What conditions need to be checked to determine causality? | 1. concomitant variation: extent cause X and an effect Y occur together or vary 2. Time order of occurrence: First X then Y 3. Absence of other possible causal factors: control other factors or hold constant |
| Third cause fallacy | Does one cause the other? no there is a third cause |
| Experiments | Common way to determine causality |
| Types of experiments | - field experiment: implement in one market and not the other - lab experiment : show to one group and not the other |
| What can disturb the outcome and solutions? | - other things going on in test markets and other markets - two experimental groups not being similar |
| How to eliminate all 'other causes' | - Control for other variable things in analysis - Random assignment to groups |
| What're the three main approaches for experimental research? | -true experiment -field experiment -quasi experiment |
| What are true experiments? | Can determine causal relationships because the researcher systematically orchestrates a situation so that people in one way or another way. Often performed in a lab. also possible in surveys |
| Disadvantages of true experiments | - manipulation makes the situation less like a real world situation - Evaluation apprehension -manipulation is usually between subjects |
| within subjects design | The same participant tests all conditions corresponding to a variable |
| Between subjects design | different participants are assigned to different conditions corresponding to a variable |
| Advantages of field experiments | - setting more similar to real life situation - shoppers in their natural environment - less evaluation apprehension |
| Disadvantages of field experiments | - Less experimental control - intervening variables |
| Advantages of quasi- experiments | - Natural setting - heterogeneity in respondents - less evaluation apprehension |
| Disadvantages of quasi- experiments | - Existing groups, no manipulation - no random assignment - less experimental control |
| 5 important concepts for regression | 1. Dependent variable Y, Independent variable X 2. Check R squared and ANOVA F for overall diagnostics 3. Check p-values( significance) 4. standardized coefficients 5. including nominal variables |
| Explanatory variables in conjoint analysis are often not interval/ ratio but nominal/categorical, so what extra effort is needed? | - If independent variable is yes or no, recode to no=0 and yes=1 (1 is dummy variable) - If its categorical make dummy variable 1 less than there are levels |
| Important issue in regression : multicollinearity | - Can't distinguish what causes sales change - hows up by surprising signs and parameters cancel each other |
| How to check for multicollinearity | - high correlations in correlation matrix - Variance inflation factor per variable, VIF>5 is suspicious |
| New product development | Conjoint analysis as a tool for better decision making |
| What're the 3 main conjoint methods? | 1. Conjoint value analysis: respondents rate profiles on a scale 2. Choice based conjoint: choose preferred profile from a set 3. Rank ordering: rank order the profiles |
| Choice based conjoint advantage | Greater realism/easier task |
| choice based conjoint disadvantages | - less information per individual - Respondents have to understand several profiles at once - linear regression wont work, requires multinomial logit |
| Conjoint value analysis | Respondents perform tasks; give evaluation grade of product profile, do this many times for different profiles. Use these grades to infer the utilities of individual attributes |
| Conjoint analysis decomposes utility into separate parts | Assumption: product is a bundle of attributes, each attribute has a certain utility for a level. Utilities of these characteristics are computed with linear regression. Show mulitple manipulated products to get enough info |
| How to determine the attributes and levels that will be experimented with | - use exploratory research, in house expertise, find the attributes and levels the customers care about - focus on attributes that can be changed - try have attributes have same levels to avoid number of levels effect (many levels seem more important) |
| What is orthogonal design? | a way to reduce the number of questions, effects of attributes can be separated |
| Which attribute level do you leave out? | The estimate of the levels indicate the change in rate relative to the one left out |
| Which attribute is most important? | importance is calculated from the range of betas ( betamax- betamin). could also compute importance per segment. |
| Benefits of conjoint analysis | -Estimate preference quantitatively instead of direct questioning and guessing -See which attribute levels are valued highly - focus on these in product development |
| limitations of conjoint analysis | - all features cannot be incorporated -many observations needed - is utility really compensatory - answering more than 9 questions can be taxing - respondents don't have to live with their choice - metric conjoint |
Created by:
janasultan
Popular Marketing sets