terminology Flashcards
Empirical Rule
Any unimodal and symmetric distribution fall under:
68 Percent of data are located on the interval
95 percent of data are located on the interval
99.2 percent of data are located on the interval
z- scores
Signed value that indicated the number of standard deviations a quantity is away from the mean.
Resampling
nonparametric technique for determining statistical significance by comparing an outcome with a set of outcomes obtained by randomly assigning the data points among groups.
randomization test
resampling technique that permutes the data points to obtain the comparison set of random outcomes, selecting each data point only once
A randomization test is classified as random resampling without replacement, because the data point can no longer be drawn from the population
randomization test
To conduct the randomization test:
Calculate the test statistic for the original data.
Calculate the number of combinations possible for the combined set of data.
Calculate the number of combinations that will produce the original or a more extreme value of the test statistic.
The p-value is found from the ratio of the two previous steps.
Bootstrapping
resampling technique that randomly selects a set of data points, allowing the same data point to potentially be selected more than once, to create an approximate sampling distribution.
parametric method
makes inferences based on data assuming some statistical distribution of a population or for a statistic
nonparametric method
makes inferences based on data requiring fewer assumptions about the statistical distribution of the population
Skew
measure of asymmetry about the mean.
skew is useful when….
Nonparametric methods are useful when data is skewed
standard error of skewness
measure of the deviations that exist between random subsamples selected from the data set, given by:
When is a sample too skewed
ratio is around 2 or greater
Parametrics When its best used
data is drawn from a population with a known distribution or when the sample is large
Parametric methods (Advantage vs Disadvantages)
Advantages
More powerful if the assumptions of the parametric test are true.
Disadvantages
Require large samples. Samples must approximate a distribution.
Nonparametric methods (Advantage vs Disadvantages)
Advantages
Do not require large samples, or for the sample to be drawn from a specific distribution
Disadvantages
Less powerful when the assumptions for a parametric test are met. Also, more difficult to compute.
