Exam 2: Participant Sampling Flashcards
Population
• All people (or items, or locations, etc.) of interest: example all 4 year old bilingual spanish-english speakers
- Who you want your results to be relevant for, generalize to
- Can be large: All 4- and 5-year old children who are Spanish-English bilinguals • Can be relatively small: All children at a particular education center
Sample
- The individuals actually in your study
- Representative of the population (ideally equal chance of selection; intended (who you want to generalize your results to) vs. accessible populations(the group from which the researcher recruits))
- Use sample statistics to make inferences about population parameters
Symbols for population and sample means
x(bar) = Sample Mean mu = Population Mean
Parameter vs. statistic
parameter- about whole population
statistic = conclusion from a sample of the population
Census
Census is when Population = Sample
Sampling Bias
Failing to identify/examine all members of a population; sources include samples of convenience (recruitment procedures) and volunteerism (cannot be avoided due to informed consent and some studies may be more susceptible to this than others)
geography matters- ethnicity, race, education levels
put out an ad and use the first 50 people who responded to the ad
motivation issues
Types of Sampling
Probability samples: SIMPLE RANDOM SYSTEMATIC RANDOM STRATIFIED RANDOM CLUSTER MULTISTAGE
Non-probability samples:
CONVENIENCE
PURPOSIVE
Probability sampling
Uses some form of random selection based on probability- requires setting up a procedure that ensures that the different members of your population have equal probabilities of being chosen
Simple Random Sampling
- Choose such that each sample in the population has an equal chance of being selected (e.g., picking a name out of a hat, choosing the short straw, generating random numbers)
- Advantages: Equal chance of selection; fair & free from sampling bias
- Disadvantages: Need to know the entire population; not the most statistically efficient method; Luck of the draw, may not represent subgroups well
simple random sampling negates sub-groups (even men/women, etc)
larger the population the more impossible this can be
Systematic Random Sampling
Selecting one member randomly and then choose additional members at evenly spaced intervals
Disadvantages: you need a complete listing & need to watch out for periodicity in your list
Advantages: Fairly easy to do, sometimes the easiest (e.g. what library books get the most circulation?)
Example:
• 100 students in your class
• Want a sample of 20
• Have a class listing in alphabetical order
• Interval: 100/20 = 5
• Start number
• Randomly select a number between 1-5 • Select every 5th until N = 20
what about systematic noise in the list?
you still need a complete list of the population
Stratified sample
- Population can be divided into different groups based on criteria (i.e., strata)
- Separate simple random sample from each population stratum
- Advantages over simple: Assures representation of overall population AND key subgroups; potentially apply your results to subgroups
Example-
• Survey ASHA membership on a topic
• Professional membership breakdown
Female (95.3%)
Male (4.7%)
• If we want to ensure that our sample has the same representation, need to stratify
• Divide population into groups and then randomly select ~20 women for every man
Cluster Sampling
• Select clusters from population on the basis of simple random sampling, then sample all people in the cluster
Example:
• Population: Kindergarteners in MD
• Cluster: Random sample of MD schools with kindergarten programs
• Sample: All kids in the sampled schools
- Economical, but susceptible to sampling bias
- Clusters are intrinsically more homogenous
clusters- make sure you sample sufficient clusters
Multistage sampling
- Combine different methods of probability sampling
* Example: Using cluster sampling to select certain schools, and then random sampling within each school
In Hearing and Speech
- Probability sampling often used for clinical field tests
- But otherwise, it is almost impossible to get a complete listing of all members of the population in order to select among them
- The cost of doing so would be prohibitive
Non-probability sampling
• Does not involve random selection
• May or may not represent the population well
Often hard for us to know how well
Even if we have a large sample size
• Susceptible to researcher bias
• Broad types: Convenience and purposive
• Matched-group design/Matched- subjects design
