Performing a Sign Test in Nonparametric Statistics, Example 183

In this video, we demonstrate how to use the Sign Test to test a hypothesis using a left-tailed test. We will explain the best way to define your test statistic so that the p-value can be calculated in a consistent way for each type of hypothesis test conducted (left, right, and two-tailed).
This video is part of the content available for free at www.statsprofe...

[@IsaacOffin]

I suggest that the test stat..should be the smallest number of (+ and -) sign or

[@DilipYadav-rt5cd]

thanks for the great video. Can you please explain why have you taken P(X>=7) and not P(X

[@IsaacOffin]

why didn''t you give your values sings ,that is, + and - sign

[@danielm9463]

This video is great--thank you for making it! So concise and clear. It helped me a lot. I've learned H0 should be written as η = 12 (not η ≥ 12) for the reason that the hypothesis test evaluates how likely the observed data are according to the null distribution/ null model centered on η = 12. (IOW, the hypothesis test doesn't compare the observed sample to other distributions centered on larger η's greater than 12.) Do you agree? Or are there other considerations/ reasons to write it as η ≥ 12?

[@dmcguckian]

Either approach is fine. In a test using strictly = for the null hypothesis, the significance level of the test is the same as the probability of a type I error. If you use the greater than or equal to in the null, your significance level is the maximum likelihood of the type I error. That means it could be much less depending on the true value of the parameter. Either approach is valid, but you should be aware of the effect on the type I error probability.

[@danielm9463]

@@dmcguckian Thanks for the detailed reply!

[@snehapakhi1908]

What's the value of x

[@trendynigs.got.biz.8268]

I fail to understand why we find the probability greater than 7. Why not lesser than 7?

[@Dr_Techo]

Hi. Thank you for your video. This is a one-tail test. What differences would there be if it was a two-tail test? i.e.: if we wanted to test that the median is not equal to 12?

[@dmcguckian]

My website has a couple of examples of the two-tailed case: www.statsprofessor.com/video.php?chapterId=7&id=804

[@itzboyon6983]

Thank you, It was easy.

[@dmcguckian]

Great!

[@sabinajavier9978]

Hello! What table did you use you here? Do you have link or??? Thank you!

[@dmcguckian]

www.statsprofessor.com/pdf/Binomial%20Table.pdf

[@sabinajavier9978]

@@dmcguckian Thank you so muchhh!

[@EAli-ec1xp]

thanks so much, plz i want Binominal tables

[@dmcguckian]

You can find it here: www.statsprofessor.com/video.php?chapterId=20&id=836

[@EAli-ec1xp]

n=7, k=1, 9=.05 i need theses tables

[@eashawasim5558]

Why are we taking @ as 0.5 when it is mentioned as 0.05.,?

[@dmcguckian]

The significance level here is 0.05. That is not the reason we are using p = 0.50. That is due to the fact that the probability that a value is more than (or less than the median) is 50%. Check my playlist on this topic for the concept video on this topic.

[@tvgathu9845]

wee mzee;