07_Sampling Bias and Error Flashcards
Sampling error is____
a random error (chance)
• Attributed to sampling, no matter the method used
Sampling bias is ___
a systematic error (not chance)
• Attributed to sampling method used
What are Types of Sampling Bias?
- Undercoverage bias (aka coverage bias) (Roosevelt/phone)
- Self-selection bias (aka volunteer bias) (computers)
- Non-response bias (male surveys)
Cause: Convenience sampling (and other non-probability sampling techniques); inaccurate definition of the target population
What is the difference between Sampling vs. Selection Bias?
- Random Sampling ≠ Random Assignment (aka Randomization)
- Sampling bias: participants in sample are not representative of the population
• High sampling bias = low external validity - Selection bias: participants choose which group to be in, treatment or control; or researcher chooses for participant (non-random assignment)
• High selection bias = low internal validity
Recall: Increasing the sample size can reduce the variance in the sample and ___
minimize sampling error.
Can a sample be too large?
yes
Sample Size and Statistical Significance
The simple one sample t-test
as n->∞, t also approaches ∞
Result: Large samples can inflate the statistical significance, and drastically lower the practical significance.
Power is the probability of ___
rejecting the null
hypothesis when it is false (correct decision).
There are formulas for calculating the appropriate sample size.
• Power analysis
• And there are sample size calculators available for free on the Internet.
Increasing the sample size reduces -____
the sampling error and increases the power.
if the sample size gets too large, the statistical significance might ___
not be meaningful (excess power).
True or False: Power = 1 - β
true
