t-test more than 30 sample size is recommendable? if not, then why t-test value is present for 40, 60 & 120?

Question

Is it correct, if we use t-test when the sample size is
more than 30? If it's not recommendable, then why in the
Statistical Table, the value for t-test is present for
40, 60, 120 & for even infinity…?

Answer 1

t-test is used for testing values of coefficient. The higher the sample size, the more accurate any analysis is, however, sample size of 30 is large enough for any college level practical exercise. Whether or not doing t-test does not depend on sample size, but rather on the hypothesis you're testing. You use t-test for testing if the coefficient is equal to some constant (the most obvious is testing for significance of the coefficient - if the coefficient is zero), but you use F-test to test joint hypotheses.

Answer 2

the previous answer is correct, but it ignores one reality: real sampling costs both time and money. While greater sample sizes render a smaller degree of error, they incur a larger cost. You can run a valid analysis with a sample of only 1, but the degree of error grows (unless your purpose is only to learn one person's opinion) each time the sample size shrinks.



A sample size of 1 billion could give some very accurate results, but you have to draw the line somewhere -- or do you want to pay for, obtain, add up, and analyze 1 billion values in less than a week?

Answer 3

That makes no sense. The sample size should be larger, not smaller. Its been quite a few years since I took Data Analysis in b-school, but, I definitely remember that you want to use larger sample sizes, not smaller. The sample size SHOULD be larger than 30.



But, sometimes, you have no choice, not as many observations. So, you do what you can with 30 observations (or less).

Answer 4

to apply t-try ought to: Have effortless random pattern (SRS) n > 40 (robust) primary distribution and unknown mu and sigma Your t -cost relies upon on levels of freedom (n-a million) and value or self belief factor.