Sie sind auf Seite 1von 2

1

A
.
No, because the two samples have very different variances (violates
homogeneity of variance). Also, because the Solar Flare distribution
appears to be highly skewed, it does not meet the normality assumption

B By taking a large number of simple random samples of size n from a


. given sample of size n, we can estimate the sampling distribution of the
mean of our original sample. The sampling distribution of mean
differences of two samples can be used to determine if the hypothesized
mean difference under the null hypothesis (0) falls within some
confidence interval. If zero falls within the confidence interval, we
cannot reject the null hypothesis. If zero falls outside the confidence
interval, we reject the null hypothsis.

A and C

1. Examine your data


2. Try a t-test or other appropriate parametric test
3. If the t-test is non-significant, then try Booststrap
4. resample a large number of samples of size n randomly from your sample of
size n.
5. Compute the mean of each of these re-samples to make a sampling
distribution
6. if you have a second sample, do the same for it.
7. create a sampling distribution of mean differences by subtracting your
bootstrap sampling distributions
8. Determine the probability that 0, the mean difference under the null
hypothesis, falls within the sampling distribution of differences
9. if p(0) is less than alpha, reject the null hypothesis. If not, fail to reject the
null hypothesis

1. Resample a large number of samples of size n randomly from your sample of


size n.
2

2. Compute the mean of each of these re-samples to make a sampling


distribution
3. if you have a second sample, do the same for it.
4. create a sampling distribution of mean differences by subtracting your
bootstrap sampling distributions
5. Determine the probability that 0, the mean difference under the null
hypothesis, falls within the sampling distribution of differences
1. if p(0) is less than alpha, reject the null hypothesis. If not, fail to reject the
null hypothesis
1. Examine your data
2. Take your two samples and subtract their ordered values
3. create a sampling distribution of mean differences by taking a large number of random samples of size n from
these differences.
4. Determine the probability that 0, the mean difference under the null hypothesis, falls within the sampling
distribution of differences
5. if p(0) is less than alpha, reject the null hypothesis. If not, fail to reject the null hypothesis

Sort your sampling distribution and count from the tails of the distribution to
the appropriate percentile to find the critical value. Compare the value of the
differences in sample statistics under the null hypothesis (usually zero) to the
critical value. If zero falls towards the center of the distribution from the
critical value, conclude that it is more probable than you woud like for your
given alpha level and fail to reject the null hypothesis. if zero falls towards
the tail when compared to the critical value, conclude that it is unlikely to
have occurred solely as a function of chance, and reject the null hypothesis.

Flare Sample 7.8,

Normal Sample 5.5

A & D, depending on which test you think is appropriate for this context.

1.17

No because the mean always provides more power than the median.

Das könnte Ihnen auch gefallen