3.2 Sampling Flashcards
Define Sampling
Sampling is to draw a subset of the population that truly represents the target population as a whole i.e. infers that the researcher is able to draw conclusions about the target population from the sample or subset of that target population
Explain the importance of sampling in health research
Unnecessary and impractical to study the whole population
Obtain useful and valid information about the population
What are the two main types of sampling?
Probability
Non-probability
Validity
i.e. findings are accurate and indicative, the research measures what it set out to measure
Truth and accuracy of the data and findings.
It refers to
- concepts being investigated
- people or objects being studied
- methods by which data are collected
- findings that are produced
True or False. The extent to which research findings can be generalised to the target population will depend on the extent of sampling error or bias within the study
True
i.e. ↑ error/bias →↓representativeness of findings
Sampling frame
Members of a population eligible to be included in a sample
What is a sampling plan and what is its purpose?
Outlines the process by which the subsets or participants are dawn from target population.
Aim is to minimise sampling error and bias and maximise representativeness of the sample.
No sampling technique can guarantee against sampling errors – we can just aim to minimise them.
What is probability sampling?
Every member of the population has an equal (and independent) chance of recruitment vs. not equal chance.
Intent is to generalise the findings of the sample to the population from which it was taken.
Used when research is investigating a large population and requires a statistical description.
Helps to reduce sampling error, increase internal validity. Reduces bias as everyone has equal chance of recruitment.
What are the 4 types of random sampling?
Simple random sampling
Systematic random sampling,
Stratified random sampling
Cluster random sampling
Internal validity means….
Less bias
Scientific process is used
Study is accurate
Refers to the extent to which changes in the dependent variable (the observed effects) can be attributed to the independent variable, rather than EXTRANEOUS variables.
External validity
Generalisation of findings from sample to population (representativeness of sample)
Refers to the degree to which results of a study are generalisable beyond the immediate study sample and setting to other samples and settings.
True or False. You need internal validity to have external validity.
True
What is simple random sampling?
Establish sampling frame.
Decide on sample size.
Select the required number of units starting at a random point by either lottery method or use of random numbers
e.g. drawing names/numbers out of a hat, or assigning everyone a number and using a random number generator to select
Systematic sampling
Sampling frame is available.
Sample size has been predetermined.
Start at a random point on sampling frame but pick every “nth” unit (the gap)
The gap is determined by sampling frame # / sample size e.g. 25 / 5 = select every 5th persion
Stratified random sampling
Subdived groups. Randomly selected from each group. Represents important characteristics of the population.
Similar to simple random sampling but the population is first subdivided into homogenous subsets (strata) from which the required number of units can be randomly selected.
Stratification reflects the population characteristics significant or crucial to the research question – e.g. gender, age, occupation.
Allows for a greater degree of representativeness of the population.
Sampling ratio =
sample size to population size
Cluster sampling
Utilises simple random sampling of predetermined groups or clusters of individual elements of interest; e.g. villages, hospitals, schools families, health areas
Often used when cannot create a sampling frame that will include all the units of the accessible population.
Example:
Wish to investigate attitudes to GBLTI amongst nursing home employees in WA
List all the nursing homes in WA (this becomes the sampling frame)
Randomly select a number of nursing homes
Interview the employees from each selected nursing home
Does a larger sample size mean less or more sampling error?
Less
Do samples of more diverse populations need to be smaller or larger than samples of more homogenous population?
Larger
Non-probability sampling
The probability of a potential research participant being selected is not known in advance.
The findings cannot be generalised to a larger group of people
Non-randomness of the sampling increases the risk of what?
Bias
When is non-probability sampling commonly used?
Particularly when have limited resources (much less costly and time consuming than probability sampling )
or unable to identify or access all members of population.
Useful in exploratory research and qualitative studies; samples not statistically representative but can give in-depth insights into phenomena
What are the 5 types of non-probability sampling?
*Purposive
*Convenience
*Snowball
Theoretical
Quota
Purposive sampling …
Participants are recruited according to pre-selected criteria relevant to the research aims/questions of a given study.
Designed to recruit thosewho have the required status, experience or knowledge of interest to the researcher.
Typically choice of sample is based on judgement that it has
characteristics typical of the population.
Open to selection bias
Convenience sampling…
When people are invited to participate in the study because they are conveniently (opportunistically) available with regard to access, location, time and willingness.
(Utilises the most readily available convenient units of the population (cases/persons) who fit the criteria of the study.)
Open to self selection bias
Suffer from either under-representation or over-representation
Less bias if the population is homogenous
Most common form of sampling in qualitative research.
What are possible contributors to sampling errors in health research?
Accessible population is not representative of the target population; Do all groups want to participate or ‘be found’?
Sampling frame may not be representative of population → unrepresentative sample
Small sample size may over or under represent certain characteristics of the population
The researcher may not have accurately identified the defining characteristics of the population
What does rigour mean / refer to in research?
Rigour refers to the trustworthiness/credibility/strength of the research
What are traditional criteria for judging quantitative research?
Internal validity
External validity
Reliability
Objectivity
What are alternative criteria for judging quantitative research?
Credibility (instead of int validity in quant)
Transferability (instead of ext validity in quant)
Dependability (instead of reliability in quant)
Confirmability (instead of objectivity in quant)
What is snowball sampling?
When the researcher starts gathering information from one or a small number of people and then requests they put the researcher in touch with others who may be friends, relatives, colleagues or other significant contacts.
Especially useful in recruiting ‘hidden’ populations e.g. marginalised or stigmatised individuals - drug users, prostitutes, those not-registered with a medical practice
What is theoretical sampling?
The research starts from a homogenous (small) sample and moves to a heterogenous (larger) sampling. Sampling occurs sequentially and alonside data analysis. Analysed data guides the areas to be explored further in the next round of data collection and the focus of the sample in order to achieve this.
Mostly used in grounded theory.
What is quota sampling?
A non-probability sampling strategy where the researcher identifies the various strata of a population and ensures that all these strata are proportionately represented within the sample to increase its representativeness.
What is random sampling?
A process of selecting a sample whereby each member of the population has an equal chance of being included.
What are some ways to assess reliability?
Test-retest, split-half and alternate forms
Statistical tests such as Cronbach’s alpha and the Spearman rho correlation coefficient test
What is reliability?
Reliability tells you how consistently a method measures something. When you apply the same method to the same sample under the same conditions, you should get the same results. If not, the method of measurement may be unreliable or bias may have crept into your research.
Concerned with consistency and dependability of a measuring instrument, that is, the degree to which it gives the same answer over time, across similar groups, and irrespective of who administers it.
A reliable measuring instrument will always give the same result on different occasions, assuming that what is being measured has not changed during the intervening period.