12 Flashcards
Mean-squared error?
How far my estimator is to what im trying to estimate squared?
Why is mean squared error important?
Since you can have high variance but low bias, or low variance or high bias, high variance and high bias, and low variance and low bias.
Probability sampling method?
Is where the probability of selecting a unit or subject with a nonzero probability and is known in advance.
What does the probability sampling method tell you? 3•
•Sampling error can be assessed
•Results can be projected to population
•Often more expensive than non-probability samples
Does every possible sample size n have the same chance of being selected?
Yes
If you have a simple random sample size of n, then the results will have some? What is an exception?
Dependence between those selected UNLESS the sample size is relatively small compared to the overall population size, then selections can be treated as independent.
What are the methods of selecting a simple random sample of n subjects from a population of N subjects? 3•
•Number subjects from 1 to N
•generate list of n numbers taken from 1 to N.
•Select those subjects assigned these numbers
Stratified sampling?
Given a population is partitioned into mutually exclusive and exhaustive homogeneous strata, a stratified sample is a sample composed of simple random samples taken from within each stratum. Ensure representation of important characteristics and that no characteristic gets ignored.
Mutually exclusive in stratified sampling?
They DO NOT overlap.
Exhaustive?
No one is left out.
Homogenous strata?
Same the important characteristic.
What is the goal of a strata sample?
Ensure representation of important characteristics so you can get some from each region. This ensures that a characteristic doesn’t get ignored.
Inhomogenous?
Different
Cluster sampling?
Grouped individuals in which every individual in that group is measured. Sample of each cluster. Exhaustive
What is the main goal of cluster sampling?
Each cluster looks like overall population.
!!!What do you do when you have several clusters when doing cluster sampling?
You take the simple random sample of each cluster (n_cluster). The sample size are the number of subjects in one cluster.
What are the benefits of cluster sampling? 2•
•more economical
•sample of city blocks, sample of households in each block.
Systematic sampling?
Given a population, the measurements is ordered in someway. This sample is formed by selecting a unit from the population and then every k^th unit around the original selected unit.
Replacement?
With or without replacement but is a simple where selected units (measurements) are returned to the population after being selected or not return after being selected.
If you sample many times with replacement, then the results are?
Results are Independent.
If you want new information when taking several samples, then?
You do not have replacement when sampling.
Are computations easier or harder with replacements?
Easier
If n ««< N, what does this mean?
Sample size really small compared to population.
If sample size n ««< N, then you can treat it with ______ by getting away with it?
Replacement
