Ok, I have looked at the code. The output text in the listing seems correct but perhaps I should have used more words. Both are through the origin. The first just uses the raw Procrustes distances and the second, as it says, uses distances from the reference. Both seem reasonable statistics but I would prefer the first as a more absolute measure of fit because one knows that distances cannot be negative. That is why I gave it first.Sorry about my initial a answer from sea where I could not check and it has been quite a while since I last thought about that program. __________________F. James Rohlf, Distinguished Prof. Emeritus Dept. Anthropology and Ecology & Evolution Stonybrook University -------- Original message --------From: Anneke van Heteren <[email protected]> Date: 5/18/22 12:18 PM (GMT+09:30) To: "F. James Rohlf" <[email protected]> Cc: andrea cardini <[email protected]>, Morphmet2 <[email protected]> Subject: Re: [MORPHMET2] tpsSmall Dear James,Thank you very much for your explanation. I am still a little bit confused though. Both regressions claim to be uncentered and through the origin (see attachment). So which one is which? Is the top one centered and through the means and the bottom one through the origin? Or the other way around?Thank you in advance.Best regards,AnnekeOp wo 18 mei 2022 om 03:22 schreef F. James Rohlf <[email protected]>:The two regression lines are: the usual centered regression through the means and an uncensored regression through the origin that makes more sense for distances that cannot be negative. Often hard to see the difference when the correlation is very high. No test intended. Was just intended to indicate how confidently you could use a tangent space approximation rather than more complex methods.__________________F. James Rohlf, Distinguished Prof. Emeritus Dept. Anthropology and Ecology & Evolution Stonybrook University-------- Original message --------From: andrea cardini <[email protected]> Date: 5/17/22 10:18 PM (GMT+09:30) To: Anneke van Heteren <[email protected]>, Morphmet2 <[email protected]> Subject: Re: [MORPHMET2] tpsSmall Hi Anneke,Jim will answer, I am sure, if he's not away.If I get your question right, I think that the two regressions have always been there:- One is shape distances computed pairwise in your sample (Procrustes vs tangent space): with 10 individuals, you'll have 10*9/2=45 distances.- The other one is the distance of each individual to the sample mean, i.e. with N=10 the corresponding 10 distances.It's two ways of looking whether the approximation in the tg space is good. In my experience they produce highly congruent results. With the second approach, however, it might be easier to spot if one or a few individuals are strong outliers, and to be able to say which they are. I am not advising this as a method to detect outliers but it's something you'll see in some cases and maybe realize that there's a problem with some individuals. For outliers, I use multiple strategies.CheersAndreaOn 16/05/2022 09:23, Anneke van Heteren wrote:> Dear friends and colleagues,> > I have a question regarding the report of tpsSmall. When doing an analysis,> the report displays two regression outputs with different slopes,> correlations and root MS errors (see attached for a simple example). Other> than those numbers, they are accompanied by the same text and I cannot tell> the difference. I do not remember it being like that before and the sample> report in the help file also only shows one regression output. I would very> much appreciate it if somebody could enlighten me on what both these> regressions mean and what their differences are.> Thank you in advance.> > Best regards,> > Anneke van Heteren> -- Dr. Andrea CardiniResearcher, Dipartimento di Scienze Chimiche e Geologiche, Università di Modena e Reggio Emilia, Via Campi, 103 - 41125 Modena - Italytel. 0039 059 4223140Adjunct Associate Professor, Centre for Forensic Anthropology, The University of Western Australia, 35 Stirling Highway, Crawley WA 6009, AustraliaE-mail address: [email protected], [email protected]: https://sites.google.com/view/alcardini2/or https://tinyurl.com/andreacardini-- You received this message because you are subscribed to the Google Groups "Morphmet" group.To unsubscribe from this group and stop receiving emails from it, send an email to [email protected] view this discussion on the web visit https://groups.google.com/d/msgid/morphmet2/7e919190-da7e-374c-4bce-2310edec61b6%40gmail.com.
-- You received this message because you are subscribed to the Google Groups "Morphmet" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/morphmet2/6285a2b3.1c69fb81.8f949.62fc%40mx.google.com.
