ARIN 2000: Research Methods Sampling Techniques

1
ARIN 2000 Research Methods Sampling Techniques
2
Designing Samples

• Apart from a national census most research is
  based on using a sample of a population rather
  than the whole population
• To conduct research, you need to decide how you
  will select your research sample (units of
  analysis)

3
Lets say you want to find out about peoples
attitudes and behaviours regarding online
shopping How will you find out?
4
Common sampling techniques

• What problems can you identify with these
  sampling techniques?
• vox pop samples
• TV viewer polls
• Chatting with friends
• Door-to-door surveys

5
Sampling Techniques

• What do we need to know before we can select a
  research sample?
• Define the population
• Who are you gathering data on?
• Examples Students at Sydney University
• Web Design companies in Sydney
• Female computer users aged 14-17 in Sydney
• Select a sampling frame
• How will you find your population?
• A sampling frame is a list of population members.
  
• Examples telephone book, electoral lists,
  company records etc
• How would you choose a sampling frame for the
  populations listed above?

6
Sampling Methods

• Once we have
• Defined our population, and
• Selected a sampling frame
• We need to select a sampling method - the two
  main methods are
• Probability sampling
• Non-probability sampling (purposive sampling)

7
Probability sampling

• Probability sampling involves random selection
  using statistical methods
• Used for selecting large, representative samples
• Used for describing characteristics of a
  population
• Used for testing empirical hypotheses
• This sampling technique is favoured in
  quantitative, positivist research

8
Probability Sampling

• Simple random sample (SRS)
• Each element in the population has an equal
  chance of selection
• Lottery method
• Random numbers
• Each element is numbered and selected by random
  numbers
• Random sampling usually done without replacement,
  ie. once an element is chosen it is removed from
  the population so it cant be chosen again
• Systematic selection
• All elements of a population are numbered and a
  sampling interval is used to select sample
  members (egg every 7th element chosen starting
  with a random number between 1 and 7)
• This system ensures a more even spread of the
  sample than SRS

9
Probability Sampling

• Stratified Samples
• The population is stratified before sampling
• Example The population of Sydney University
  Students might be stratified by Faculty or
  Department or by year before sampling
• Stratified samples make the sample more
  representative
• Cluster Samples
• Clusters are chosen from within the population
  before sampling
• Examples Choosing a cluster of suburbs, schools,
  organisations before sampling
• Cluster samples are cheaper and easier to
  administer but are less representative of the
  total population

10
Non-probability sampling

• Non-probability sampling is used when probability
  sampling is difficult or inappropriate
• Often used for small-scale qualitative research
• Good for exploratory research and theory
  development
• Good for developing and testing survey research
  instruments or doing a pilot study

11
Non-Probability Sampling

• Purposive Sampling
• Researcher decides on the sample based on their
  own knowledge of the population and the aims of
  the research
• Reliance on available subjects
• For example, stopping people on the street or
  using students in your lecture class
• This method is convenient but it is risky to
  generalise from this type of sample

12
Non-Probability Sampling

• Snowball sampling
• This form of sampling is useful when members of a
  population are difficult to locate, e.g. computer
  hackers, homeless people, drug addicts etc
• You start by collecting data on members of the
  population you can locate and ask them to
  introduce you to others from that population and
  continue in that way until you have collected
  enough data

13
Non-Probability Sampling

• Quota Sampling
• Sets quota controls to make the sample
  representative of the entire population, often
  used in market research, polling and audience
  research
• Example a study of computer use by women might
  set quotas for women who are single, married,
  with children, without children, employed,
  unemployed, different age groups etc
• To do this you need to know the proportion of
  people with each characteristic in the population

14
Quota Sampling

• Advantages
• Interviewing costs are cheaper than probability
  sampling (e.g. lower travel costs)
• Can be organised quickly - useful when research
  results need to be obtained quickly(e.g. polling
  attitudes to a political speech)

• Disadvantages
• Sample can be biased depending on the time and
  location of the research
• Danger of interviewer bias in selection
• Cant estimate the standard error or use
  inferential statistics because its not a
  probability sample