Please enable JavaScript.

Coggle requires JavaScript to display documents.

Empirical Research (Data (Primary Data (Data scaling (Interval Scale…

- - - - Population
        
        Includes all elements of a population
      - Sample
        
        Mostly this one is used
        
        Most times used because population is too large or confidential
    - - Pros
        
        Very specific and suited for a given research question
      - Cons
        
        Requires significant time and money
    - - Gathered by us
      - Collected for given purpose
      - Directly fit specific research project
    - - Self Selection
        
        Reason people are participating
        
        Only selecting people with a positive opinion
        
        Ex: Comments on course or hotel
      - Selection Bias
        
        Examples
        
        Survey via phone
        
        Survey while TV show is on
        
        1PM survey at Marienplatz
    - - Nominal Scale
        
        Color (Blue/red/green)
        
        0 or 1 variable
      - Ordinal Scale
        
        School Grades
        
        Some subjective value
      - Interval Scale
        
        Temperature
        
        No meaningful zero
        
        Distances can be interpreted
      - Ratio Scale
        
        Income or rent
        
        Distances can be interpreted, but there is also a real 0, making the characteristic represented by the variable does not existing
      - Interval and Ratio can be combined and called Metric scale
    - - Time series
        
        Examples
        
        Beer drinking over the week
        
        Pros/cons
        
        Pros
        
        Often can be performed using historical data
        
        Cons
        
        Data collection takes time
      - Cross sectional
        
        Definition
        
        Measuring at one specific point in time
        
        Pros/Cons
        
        Pros
        
        Quick
        
        Cheap
        
        Cons
        
        Homogeneity
        
        R&D vs Sales Example. If you only look at manufacturing firms
        
        Not always negative (depends on research question)
        
        Examples
        
        Food consumption in 2016
      - Panel Data
        
        Definition
        
        Combination of Time series and Cross Sectional
        
        Examples
        
        N people are observed over 10 years
        
        Pros/cons
        
        Pros
        
        Able to control latent (time-invariant) hetereogenity between units of observation
        
        Cons
        
        Panel mortality (people drop out)
        
        Non responses at certain times
        
        leads to unbalanced panel (panels with non responses at certain times)
        
        Panels
        
        Balanced
        
        Each person gave data each year
        
        Unbalanced
        
        Some people didn't give data on come years
      - How to choose?
        
        Funding/resources?
        
        Time horizon?
        
        Change over time means time or panel data
        
        Type of research question
        
        Snapshot/Static
        
        Development/Dynamoc
        
        Type of hypothesis
        
        Correlation (One point in time sufficient)
        
        Change (Several points in time)
        
        Causality (several points in time)
  - - - Internal
        
        Data from inner departments, customer service data, etc.
      - External
        
        Data from the department of statistics
        
        Eurostat is free
        
        AMADEUS is very expensive but complete
    - - Pros
        
        Readily available for analysis
        
        Usually exact
        
        Can have errors also
        
        In general inexpensive or free
        
        Can also be quite expensive (AMADEUS)
      - Cons
        
        Might not exactly fit research problem
        
        Competitors have same data
        
        Data might be old
        
        Sometimes bad quality