Inferential Stats Flashcards
(11 cards)
For a single sample,
What are the criteria to use z?
Large sample size, n ≥ 30
Population variance can be known or unknown.
OR
Small sample size, n < 30
Population variance known
For a single sample,
What are the criteria to use t?
Small sample size, n < 30
AND
Population variance is unknown.
Assumption for:
Two independent populations mean test using t
- Underlaying data are normally distributed
- Population variance of both populations are the same
Assumptions for:
Paired samples mean test using t
- Population differences are normally distributed
- Population variance of both populations are the same
When to use
x̅ – y̅
VS
d̅
x̅ – y̅ : two independent samples
d̅ : two paired samples
A question of two paired samples wants you to find confidence intervals of μx–μy, will you use find x̅ – y̅ or d̅
d̅ as it is paired sample.
It is believed that the means of two population differs.
Therefore,
H0:D̅ = 0
H1:D̅ ≠ 0
True or false?
False, hypothesis must refer to population; D̅ is mean of samples differences. μx–μy OR μd are acceptable.
5 samples are taken before and after an adjustment.
A statistician uses paired samples t-test using two distinct sample variances. Comment on his method.
He should be fired.
Populations tested using t-test are always assumed to share a common variance.
When doing mean tests using z, or finding confidence interval of mean using z, is it necessary for a sample to be normally distributed?
No. If the sample size of the population is sufficiently large (n > 30), due to central limit theorem, the mean of sample, X bar, appears to be/is approximately normally distributed.
A random sample of 50 batteries, claimed to be identical, taken from factories A and B each, are tested for the mean lives. Should you find the common variance and do paired test?
No and no.
It is not known whether they share a common variance and the batteries are not paired in any way meaningful
It is claimed that A and B are the same.
Confidence interval is given to be (-15,8).
Does the confidence interval support the claim?
Yes.
If A and B have no difference,
μa–μb
or
μd
was claimed to = 0,
And 0 is within the interval.