Added Flashcards
Multi item scale ads
Ads: less variables in regression formula
Multi item scale mOstly has a higher reliability compared to a single item scale
Higher construct validity and criterion validity
More precisie segmentation possible
Four steps of sampling process
Define the population
Determine the sampling frame
Select the sampling procedure
Determine the sample size
Define the terms correlation and causality
Correlation: when two or more events are related to each other and change together
Causality : when one event contributes to the production of another event. Where the cause is partly responsible for the effect, and the effect is dependent on the cause
Why do we sample
Because of cost considerations
Because researches cannot analYse the whole population
Samples can produce sound results d proper rules are followed for the draw
How would you measure
What is your gender?
Answer options male and female
Scale type nominal
Measurement 1-0 codes variables
How would you measure: would you buy a smart refrigerator?
Answer option: yes or no
Scale type: nominal
Measurement: 1-0 coded variables
How would you meisje how likely are you willing to pay for a smart refrigerator
Answer Opiotn: 5 or 7 point likert scale
Scale type: interval
Measurement: 1-5 coded variables
What would you willing to pay for a smart refrigerator
Answer option: let respondents fill in any numerical answer
Scale type: ratio
Measurement: digit between o and k
How would you rank the following criteria for choosing a smart refrigerator
Answer option: let respondents check boxes with smart refrigerator characteristics
Scale type: nominal
Measurement: each characteristic has a rank from 1 till k
How many trips to the grocery market do you make a week?
Answer option: let respondents fill in any numerical answer
Scale type: ratio
Measurement: digit between 0 and k
How to Calc the sample mean
Sum of:
Xi * rel frequency
Or if without rel frequency:
1/N * sum of xi * absolute frequency
Optimal ad budget
Optimality condition is
aU /aA = 0
So au/aA = B (P - Cost var) * K * Pa * A^b-1 -1=0
If doesn’t equal 0 then optimality condition not fulfilled
Calc optimal ad budget
_
A optimal = B * (P-Cvar) *Q
Carry over effect
Marketing implies in period t leads to lagged Seles effect in later periods (eg because of awareness or image effects)
Features of an experiment
- Formulation of a casual relationship (hypothesis)
- Evaluation of the directional influence of one or more independent variables on one or more dependent variables.
- Controlling of all disturbing influences (control variables) to exclude distortion of the results
Factors influencing the a error
Size of the effect - the larger the measured effect, the lower the probability of error
Dispersion of the measurement values - the greater the dispersion of the measurements values, the greater the probability of error
Sample size - the larger the sample the lower je probability of error
T test for independent samples
T = difference of sample means / standard deviations of the difference of the sample means
For independent samples there is no correspondence between the respondents of the groups
The group sizes may be different
R squared
R squared = regression coefficient ^2 * (variance of x / variance of y )
Properties of R squared
R squared indicates how well the mode explains the variance of a dependent variable.
But there are no rules for how high R^2 must be
It says nothing about the importance of an influencing variable
It offers no info on how well the model performs outside of the sample
It is influenced by the properties of the sample: decreases with greater variance of Y as well as with lower variance or x
Correlation is not regression
Parameter to be estimated:
For correlation analysis constrained between -1 and +1
For regression not constrained
Intepretation of result:
Correlation analysis: linear correlation between two variables
Regression analysis: beteeen one dependent and multiple influencing
Theoretical understanding:
Correlation; not necessary
Regression: necessary and testable
Ads and dis of regression analysis
Pos of regression analysis:
Can be used for various purposes: explanation of relationships, simulation of effects, prediction
And works with various data types like:
Classic metric data
Data with 0/1 values
Frequencies
The dis of regression analysis:
The regression model needs
Mathematical formulation of the mental model
Sufficient data with sufficient variation
Good data (clean measurement)
Key assumptions of linear regressions
Multiple linear regression requires at least two independent variables
Requires a linear relationship between dependent and independent variable
Error term is normally distributed
No multicollinearity: multiple regression assumes Independent variable not high correlated with each other
Homoscedasticty: variance of error terms are similar across the values of independent variables
Regression analysis requires at least 20 cases per independent variable
Error sources
Over reporting - more positive than they actually are
Interview bias
Bias because of question order
Halo effect - when one answer influences another
Socially desired answers because of non anonymity
How to find most significant thing in data
One with lowest significant
Which variable has the strongest impact
One with highest standardised beta coefficient
Four generic strayed postures in multinational corporations
Low global integration and low local responsiveness = international strategy
Low global integration and high local responsiveness = multinational strategy
High global integration and low responsiveness = global strategy
High global integrationi and high local responsiveness = transnational strategy
Drivers for strategy orientated behaviour
Should = goals and action Want = incentives and attitudes Can = resources and skill
Benefits of databases
Data independence - seperately data storage from data retrieval
Consistency of data - no duplicates
Data accessibility and responsiveness - different ways of accessing required data
Uniform security and integrity controls - access control, backup and recovery