0
Still curious? Ask our experts.
Chat with our AI personalities
Ross
Fran
DevinAdd your answer:
Earn +20 pts
Is it necessary that sample sizes be equal for a two sample test?
No.
Does the central limit theorem allows the use of the normal distribution to analyze the sample mean if the sample sizes are large enough?
Yes. Roughly, very large samples are very likely to have subsets data points having very similar means and distributions. Large numbers of such subsets will tend to be normal distributed (Why?) and will tend to make the total sample be normally distributed.
Is it desirable to have a two sample test with equal sizes?
If they are not matched pairs, it does not really matter. If the combined sample size is fixed (because of costs, say) then it is better to have a larger sample where more variability is expected.
What can the difference between mean and median values infer. If the mean is 8 655.7 and the median -8 653.0?
First, I will give an example, similar to your question: -11000 -9000 +44000 mean = 8,000 and median = -9000. Symmetrical distributions after infinite sampling will show no difference in mean and median. Large differences are possible with small sample sizes even with symmetrical distributions. If the sample is large and the difference is large, this infers that the distribution is asymmetrical. The skewness of the distribution can be calculated.
Why is A bell shaped probability distribution curve is NOT necessarily a normal distribution?
There are infinitely many sets of parameters that will generate a bell shaped curves - or near approximations. The Student's t or binomial, for large sample sizes get very close to the Gaussian distribution. There are others, too.
