S2) Sampling Variation, Bias & Confounding Flashcards
Differentiate between statistics and probability, using a scenario
- Statistics: Given the information in your hand, what is in the pail?
- Probability: Given the information in the pail, what is in your hand?
What is sampling distribution?
Sampling distribution is distribution of a statistic across an infinite number of samples
Discuss how we analyse the following:
- Theoretical population
- The study population
- The sampling frame
- The sample
- Theoretical population: Who do you want to generalise to?
- The study population: Which population can you access?
- The sampling frame: How can you get access to them?
- The sample: Who is in your study?
What are the three main studies in population science?
- Survey
- Cohort study
- Case-Control study
In terms of the following image, describe the bias and precision.
- Large bias
- High precision
In terms of the following image, describe the bias and precision:
- Large bias
- Low precision
In terms of the following image, describe the bias and precision:
- No bias
- Low precision
In terms of the following image, describe the bias and precision:
- No bias
- High precision
What is a confidence interval?
Confidence interval is an estimate of the precision of the observed values in the sample
What is a 95% confidence interval?
95% Confidence interval is the range that is likely to contain the mean of the population 95% of the time
Which two factors widen the 95% confidence interval?
- Greater variation in population values
- Smaller sample sizes used to calculate it
What approximates to the mean of population values?
The mean of the ‘mean of samples’ approximates to the mean of population values
What approximates to normal Gaussian distribution?
The distribution of the mean of samples approximates to normal Gaussian distribution
Which kinds of errors result from selection bias?
- Errors in generalisation
- Errors in comparability
What are errors in generalisation?
Errors in generalisation: study participants are drawn from a sampling frame that is not representative of the general population