A textbook that might be very useful as a comprehensive review of nonparametric is Siegel's:
https://www.amazon.com/Nonparametric-Statistics-Behavioral-Sciences-Sidney/dp/0070573573 I like this text because in a very intuitive way describes the foundations and procedures for the techniques. Concerning CROSSTABS, the phi or tetrachoric coefficient for a 2x2 crosstab is a particular case of Pearson correlation when the two variables are dichotomic. As we know, in general, CROSSTABS for nxm dimension tables, Chi-square statistics is used and for tables with spare data, Fisher exact test: https://mathworld.wolfram.com/FishersExactTest.html Basically, the story for small samples would lay on combinatorial procedures to generate ad hoc empirical distributions. Alex On 12/26/2020 2:57 PM, Alan Mead wrote: My understanding was that Spearman's Rho is simply Pearson's r calculated on ranked data. If so, you can interpret them identically (other than the effect of ranking, which may be substantial), including using the same hypothesis test. Isn't it a bug if there is a space for p-value and it's blank? If you're saying that PSPP censors the p-value for small samples, that seems like the main use-case for using p-values. I just checked and the introductory stats book I taught from (Howell's undergraduate "Fundamental Statistics for the Behavioral Sciences") only has a couple pages on Spearman's Rho, but confirms the above and doesn't say anything about the hypothesis test being "wildly inaccurate" for small samples? I mean, all estimates from small samples are "wildly inaccurate," right? -Alan El sáb, 26 dic 2020 a las 14:57, Alan Mead (<ame...@alanmead.org>) escribió: > John, > > My understanding was that Spearman's Rho is simply Pearson's r calculated > on ranked data. If so, you can interpret them identically (other than the > effect of ranking, which may be substantial), including using the same > hypothesis test. > > Isn't it a bug if there is a space for p-value and it's blank? > > If you're saying that PSPP censors the p-value for small samples, that > seems like the main use-case for using p-values. I just checked and the > introductory stats book I taught from (Howell's undergraduate "Fundamental > Statistics for the Behavioral Sciences") only has a couple pages on > Spearman's Rho, but confirms the above and doesn't say anything about the > hypothesis test being "wildly inaccurate" for small samples? I mean, all > estimates from small samples are "wildly inaccurate," right? > > -Alan > > > On 12/26/2020 6:22 AM, John Darrington wrote: > > It N is large the significance of Spearman's Rho can be estimated using the > T statistic. This is what SPSS does in its NPAR TEST CORR command. However > it is wildly inaccurate for small values of N and SPSS has taken a lot of > criticism for it. > > J' > > On Sat, Dec 26, 2020 at 12:45:26PM +0530, Ramagokula Krishnan wrote: > Hi, > > Hope you are safe and well. First of all, thanks for making this > software a > reality. Many of my students in India are benefiting from this. > > This is a small query I have with regard to Spearman's Rank Correlation > using ordinal variables. There is an option to perform the test under the > crosstab option which is great. However, the p value (significance) is > not > shown in the output. > > I've attached a screenshot of the output. Please have a look at the > bottom > most table which has the correlation values but the significance has not > been mentioned. > > It would be great if there was a workaround through which I could > estimate > the p value as well. > > Thank you in advance > Dr. S. Rama Gokula Krishnan > > > > > > -- > > Alan D. Mead, Ph.D. > President, Talent Algorithms Inc. > > science + technology = better workers > http://www.alanmead.org > > The irony of this ... is that the Internet is > both almost-infinitely expandable, while at the > same time constrained within its own pre-defined > box. And if that makes no sense to you, just > reflect on the existence of Facebook. We have > the vastness of the internet and yet billions > of people decided to spend most of them time > within a horribly designed, fake-news emporium > of a website that sucks every possible piece of > personal information out of you so it can sell it > to others. And they see nothing wrong with that. > > -- Kieren McCarthy, commenting on why we are not > all using IPv6 > >
