Sampling populations Flashcards
What are the two reasons for why how sampling is done is important?
- Obtaining reliable measurements of the population parameters
- Assessing the uncertainty of those measurements
What were the results of the 1984 study on Feline Hi-Rise syndrome (cats falling out of windows)?
There was a positive correlation between survival and height fallen. Cats that fell from higher stories were more likely to survive
What did the vets behind the 1984 study on Feline Hi-Rise syndrome attribute their results to?
- Cats had a high surface area to volume ratio with their size and leg shape
- Had excellent vestibular systems
- They reached terminal velocity quickly, relaxed and were better able to absorb the impact
- They could land on their feet and absorb the shock through soft tissue
What was the problem with the sample in the 1984 study on Feline Hi-Rise syndrome?
It was a sample of convenience. The only cats studied were the ones that fell out of windows by chance and were brought to the vets
How was the sample in the 1984 study on Feline Hi-Rise syndrome biased? What were the vets missing?
Only cats that were brought into the clinic were studied. Many cats that fell out of higher floors died and weren’t brought in, and many that fell out of lower floors weren’t injured
What is a population?
Every single individual/unit of interest
What is a sample?
A subset of units from a population
What is a sampling unit?
The basic unit of study. Can be a group or an individual thing
What are 3 characteristics of good samples?
- Independent selection
- Random selection
- Sufficiently large
What are 3 challenges with sampling?
- Sampling error
- Sampling bias
- Precision
What is sampling error?
Departure from the true population parameter due to chance. Is just luck
What is the precision of an estimate?
The spread of estimates from sampling error
What is precision?
The idea of being reproducible in amount or performance. Estimates are tightly grouped and repeatable
Does low sampling error result in high or low precision? High or low reproducibility?
High precision and high reproducibility
How do you minimize sampling error?
Increase sample size and take multiple samples
What is sampling bias?
A systemic discrepancy between estimates and the true population parameter. A consistent over or underestimation
What is accuracy?
How close to the true population parameter the estimate is
What is the difference between sampling error and sampling bias in terms of the direction of deviation?
Sampling error deviates in both directions. Sampling bias only deviates in one
Does low sampling bias result in high or low accuracy?
High
What is measurement error?
Imprecision caused by variation in measurements on an individual due to chance
What is measurement bias?
Systemic differences between the truth and the measured value of each individual
What are the two criteria for a sample to be random?
- Every individual has an equal chance to be sampled
2. Independence: sampling one has no effect on another
What happens to the sample if every individual isn’t equally likely to be chosen?
Get a biased sample
What happens to the sample if there’s no independence?
Smaller sample size and miscalculated precision
What would be 2 ways to do a random sample?
Random number generators and random plots
What is a sample of convenience?
A sample based on individuals that are easily available to the researcher. Not a good thing to have
What is volunteer bias?
A systemic difference in human studies between the pool of volunteers and the population where they came from
Why is volunteer bias a thing?
The behaviour of subjects affects their ability to be chosen
