Statistics Flashcards
The ideal sample should be selected at random. Why is this?
Makes the sample unbiased - each person/object has an equal chance of being picked
Makes the sample independent - if one person/object is chosen, it has no influence on another being chosen
What are the different sampling methods?
stratified - divide population units into homogenous group (strata) and draw a simple random sample from each strata (e.g. population of all 20-50yr olds can be grouped according to SIMD - Scottish Index of Multiple Deprivation, and within each SIMD group, a simple random sample can be selected)
Clustered - groups population into small clusters, naturally occurring clusters (e.g. health boards, schools, practices), draws a simple random sample of clusters
Systematic - start with randomly chosen unit then select every nth unit, ok if pattern smooth over time
Convenience - avoid the bother of designing a procedure and just grab the first n population units that come along, especially prone to bias
Self-Selected - volunteers come forward, all sorts of problems can arise from these
How can you create a sampling frame?
random number tables
computer generated random numbers
How certain can we be that our sample mean is CLOSE TO the target population mean?
only select one sample
calculate 95% confidence intervals around the sampling mean
-95 times out of 100, the CI will contain the true population mean
Interpretation of a confidence interval and data is only valid when?
the sample is a representative sample from the target population
the sample size is large enough for the distribution of the sample mean to be normal
If a more precise estimate of the mean is required (i.e. a narrower CI interval), what would you do?
increase the sample size
Why are confidence intervals used?
They allow us to generalise the result to the target population provided the sample is representative
What should be assessed when looking at data?
How sampling is done - random?
Is sample representative - any biases?
Is sample large enough - look at width of confidence interval
Can findings be generalised to the target population?
Are the findings clinically important?
The ______ the sample, the _______ the CI, the more accurate the mean.
larger
narrower
What are the advantages of sampling a population?
lower cost
less time required
less resources needed