Fichas
BIOL 315 Flashcards 7
Fichas del conjunto (11)
- 1Steps when Data is Non-NormalContinue with the test
Transform Data
Non-Parametric Tests
Randomization and Resampling
Change to an Alternate Test - 2Data Transformation RulesMust apply the exact same transformation to each individual
Must be backwards convertible to the original value without ambiguity (needs one-to-one correspondence) - 3Choosing Transformation RulesCan try different transformations until you find one that makes the data fit the assumption(s)
CANNOT keep trying transformations until p<0.05 and results are significant - 4Log TransformationGood when:
-variance and mean increase together
-data has a right skew
-good for count data - 5Arcsine TransformationGood for proportions
- 6Squared TransformationOpposite to the log transformation
Used for left-skewed data or data where the variance decreases with mean - 7Steps when Data is HeteroscedasticContinue with the Test
Use Welch's t-test
Transform the Data
Randomization and Resampling - 8Permutation/RandomizationUsed for hypothesis testing on 2 numerical and/or categorical variables
Shuffles the real data randomly, makes estimate on randomized data, repeat for 10000 times, make a distribution of estimates, treat this as your null hypothesis. Works best when designs are balanced. Faulty when there are unequal sample sizes and heteroscedasticity - 9BootstrappingEach group has a randomly chosen equal number of observations taken with replacement
With the bootstrap data, the estimate is calculate
This is repeated many times
Make a distribution of bootstrap replicate estimates - 10Faults of Computer Intensive MethodsConventional methods are more convenient and more powerful when assumptions are met (and sample sizes are small)
Can be very computationally intensive and can take a long time - 11Square-Root TransformationWill have a similar as the log transformation, but one advantage is that it can have zeroes in the dataset
Información
- Descripción
- (Bio 315) Science Fichas sobre BIOL 315 Flashcards 7, creado por Natasha Gidluck el 06/12/2020.
- Creado por
- Natasha Gidluck
- Visitas
- 2
- Valoración
Más sobre Probability and Sampling Distributions