Marketing Research: Data Collection

Marketing Research: Data Collection

1. Sampling

   • Population
     • All individuals/objects that meet certain requirements to be in this group
       • All households in HK
       • All HKUST students
       • All cars sold in 2016
     • Characteristics
       • Parameter
   • Census
     • Information about every member of the population
       • Demographic Information of whole populaton
   • Sample
     • Only a subset of the population is studied
       • Used to draw conclusions about the target population
     • Why not use census?
       • Population data is not available
       • Too costly to analyze entire population
     • Characteristics
       • Statistics
   • Sampling Error = Parameter - Statistics
   • Procedure
     • Define target population
     • Identify sampling frame
       • List of population elements --> draw sample
         • Geographic areas
         • Institutions
         • Individuals
     • Select sampling procedure
       • Non-probability
         • Zero or unknown probability of being selected
         • Convenience Sample
         • Judgment Sampling
           • Snowball sampling
         • Quota Sample
           • Looks like the larger population in certain dimensions
       • Probability
         • Advantages
           • Allows quantification of sampling error
           • Generally more representative
         • Known and non-zero probability of being selected
         • Design
           • Simple random sampling
             • Advantages
               • Easy to implement
               • Doesn't always require a list
             • Disadvantages
               • Less efficient than stratified
               • Usually more expensive than cluster
             • Mechanics
               • Table of random numbers
               • Computer generated random numbers
               • Random digit dialing
             • Known and equal probability of being selected
           • Stratified sampling
             • 2-step procedure
               • Divide population into mutually exclusive & exhaustive subgroups (strata)
               • Randomly select a sample from each subgroup (strata)
             • Feature
               • Homogeneous within group
               • Heterogeneous across groups
             • Advantages
               • More efficient than simple random
               • More effective than simple random
             • Disadvantages
               • Costly & difficult to identify strata
               • Require larger sample with larger number of strata
           • Cluster sampling
             • 3-step procedure
               • Divide the population into mutually exclusive & exhaustive subgroups
               • Randomly delect 1 or more subgorups
               • Select either all (one-stage cluster sampling) or random sample of the elements (2-stage cluster sampling) from chosen subgroups
             • Feature
               • Large heterogeneous within group
           • Systematic sampling
             • Procedure
               • Determine sample size N
               • Divide population of size M by N
               • Pick random start
               • Select every d elements
             • Features
               • Systematically spread the sample
               • Efficiency depends on the ordering of the list
           • Multi-stage sampling
             • Combination of several methods
     • Select sample elements
     • Collect data