Sample
A part or portion of a population.
The group of tickets we actually draw from a box.
Inferences
Making generalizations about a population based on results from the sample.
Parameter
Numerical facts of the population we want to study.
What we are trying to estimate.
Statistics vs. Parameters
Statistics are what we currently know, parameters are what we want to know.
Selection Bias
A systematic tendency on the part of the sampling to exclude a type of person or representation from the sample.
Quota Sampling
Fixed number of people given to interviewers that they must meet within certain categories assigned.
- Does not account for differences within categories.
- Human selection of the subjects does not eliminate unintentional human bias.
Simple Random Sampling (SRS)
Drawing from a box at random without replacement.
- Eliminates unintended human bias.
- Law of averages guarantees the percentage of subjects in a sample is likely close to the percentage in the population.
Multi-Stage Cluster Sampling
Breaking down a very large population group into smaller sub groups, each time using random chance to ensure unintended human selection bias is eliminated while still being able to narrow down to a very specific group we want to sample.
Probability Methods of Sampling
- Simple Random Sampling. (SRS)
2.
Response Bias
A type of bias that causes respondents to answer a particular way based on how questions are phrased or in the tone they are asked in.
Estimate =
Parameter + bias + chance error
Sampling Error
Often referred to as the ‘error’ or ‘chance error’ because of the fact that the sample is only part of the population.
Non-Sampling Error
This is considered as bias and usually comes from other sources than the sample itself such as non-response, selection bias, or any other systematic error.
