Please enable JavaScript.

Coggle requires JavaScript to display documents.

SU4- Causal Rearch, Sampling Theory - Coggle Diagram

- - - - Test marketing- controlled field experiments
        
        Data usage
        
        Pilot test product intro/modifications
        
        Explore different marketing mix options
        
        Examine product pros, cons, inconsistencies
        
        Importance
        
        Predict sales
        
        Identify possible cust reactions
        
        Anticipate adverse marketing consequences
        
        Vary X = marketing mix variables (4Ps) to get market performance indicator e.g. sales, website traffic
  - - - Data collected- confirms X + Y meaningful association
      - Control- all other possible variables that might change Y
      - Temporal order- X before Y
    - - Internal- research accurately identifies r/s i.e. controls all other variables than X
      - External (data generalisability)- increase by random sampling/treatment assignment
- - - - Sample representativeness; data generalisability
      - Discrepany- btw sample stats and population actual value
      - Reduce error- increase sample size
    - - Population frame error- wrongly defined target population
      - Measurement error- wrong qns/scale
      - Response error- poorly designed questionnaire
      - Data logging error
      - Extensive study- more error
  - - - Types
        
        Simple random sampling (SRS)- equal selection chance
        
        Pros
        
        Easily understood
        
        Generalisable results- with prescribed margin of error
        
        Unbiased estimate of population's characteristics, parameters
        
        Cons
        
        Difficult to get complete list of target population
        
        All sampling units need to be identified
        
        Systematic random sampling- order
        
        Pros
        
        Fast
        
        Cheap
        
        Cons
        
        Potential hidden data patterns- not observed by researchers
        
        Need to know how many sampling units in defined target population
        
        Steps
        
        Skip interval- defined target population size/desired sample size
        
        Random starting point
        
        Stratified random sampling
        
        Steps
        
        Form homogeneous subgroups (strata) within target populatioln
        
        Choose unit randomly from each strata
        
        Combine all chosen units into 1 single sample
        
        Methods
        
        Proportionate stratified- strata size relative to proportion of total population
        
        Disproportionate stratified- strata size independent proportion to total population
        
        Pros
        
        Assurance of sample representativeness
        
        Opportunity to compare stratas
        
        More precise- less error
        
        Cons- hard to determine stratifying factor
        
        Goal
        
        Within each strata- as homogeneous as possible; minimal variability
        
        Between stratum- as heterogeneous as possible; maximise variability
        
        Cluster sampling- mutually exclusive subpopulation (cluster)
        
        Each cluster- represents target population heterogenity
        
        Area sampling- geographic designation cluster
        
        One-step approach (assumes all cluster same)- randomly pick 1 cluster as sample
        
        Two-step approach- randomly select several clusters, probability method to select units within selected clusters
        
        Pros- cheap, easy to implement
        
        Cons- clusters likely homogeneous, appropriateness of
    - - Types
        
        Convenience
        
        Assumes target population homogeneous
        
        Pros- large sample, easy to use
        
        Cons- representativeness
        
        Judgement/purposive
        
        Based on researcher's belief that respondent will meet study requirement
        
        Pros- convenient, better than convenience sampling
        
        Cons- representativeness
        
        Quota
        
        Prespecified quota of some criteria (demographics, behaviours, etc)
        
        Pros
        
        Assures prespecified subgroup represented; contains specific subgroups desired by researcher
        
        Reduces selection bias
        
        Snowball
        
        Respondents help researcher identify more respondents
        
        Usage- niche market
        
        Cons- bias presence, non-generalisable results
      - Based on researcher's judgement
      - Representativeness/generalisability concerns
  - - - Precise definition (elements, sampling units, time frames) of target population
      - Full set identified for investigation based on research objectives
    - - Source- customer lists
      - List of all eligible sampling units