Test
Further Hypothesis Testing
Este test tiene tipos de pregunta que se abren en el reproductor de siempre.
Preguntas del test (3)
1Assumptions of T tests
> we assume that observations are [blank_start]independent[blank_end] > the observations are in the 2 groups are unrelated
> [blank_start]normal[blank_end] distribution for each sample
> equal [blank_start]variances[blank_end] in the two samples
if the variances prove not to be equal you can perform a t test with unequal variances > this carries out a more strict t test with [blank_start]decreased[blank_end] degrees of [blank_start]freedom[blank_end]- Aindependent
- Bnormal
- Cvariances
- Ddecreased
- Efreedom
2if normal distribution and/or equal variance of the data is [blank_start]not[blank_end] met we can [blank_start]transform[blank_end] the data
transformation
- e.g. takings logs or [blank_start]square[blank_end] roots > brings in the extreme values that might be effecting the [blank_start]normal[blank_end] distribution
after transformation we can test the means of the transformed data in a t-test- Anot
- Btransform
- Csquare
- Dnormal
32 dependent samples
- sometimes we have situations were every measurement in one group is related to a particular measurement in the other group (before and after treatments)
the data is in '[blank_start]matched[blank_end] pairs' form
so instead of doing 2 sample t test (independent samples) we do a [blank_start]paired[blank_end] t test
ASSUMPTIONS OF PAIRED T TESTS
- the paired t test assumes the [blank_start]differences[blank_end] (e.g. before and after) are approximately [blank_start]normally[blank_end] distributed > it makes no assumptions about the before and after samples separately
in an example with 6 data points, this does not give us much chance to prove a difference between a variable > the study is [blank_start]not[blank_end] very powerful due to its small sample size
[blank_start]bigger[blank_end] (er) sample size = more power!
stat > power and sample size > 2 sample t
- leave sample size empty because we want minitab to tell us what we need
- if we wanted to find a difference of 4 between means, put that in box
- power value = 0.8 (8%)
- standard deviation = whatever it is
sample size = if wanted to find that difference/significant result (if there is one) 80% of the time then we need a [blank_start]sample[blank_end] size of at least..... answer- Amatched
- Bpaired
- Cdifferences
- Dnormally
- Enot
- Fbigger
- Gsample
Información
- Descripción
- Stats Test sobre Further Hypothesis Testing, creado por sophie waddington el 28/12/2018.
- Autor
- sophie waddington
- Idioma
- English
- Visitas
- 0
- Etiquetas
- Valoración
- Aún sin valoraciones Valora este test