Dear all, I am trying to fit a simple (intercept-only) meta-analytic model using the robumeta package using the following code: anx_cont_mean<-robu(formula = es_fisher ~ 1, var.eff.size variance_fisher, studynum = study_ID, modelweights = "CORR", rho = 0.8, small=TRUE, data = anxiety_control) When I try to run this model, the following error message pops up: Error in solve.default(sumXWX) : system is computationally singular: reciprocal condition number = 0 What exactly does this mean in the context of meta-analysis? I haven't been able to find any answers. Thank you, Kristina ----------------------------------------- Kristina Loderer Ludwig-Maximilians-Universit?t M?nchen Department Psychologie Leopoldstr. 13 D-80802 M?nchen Telefon: +49 (89) 2180-6047 Email: Kristina.Loderer at psy.lmu.de -----------------------------------------
Dear Kristina I do not use that package so cannot offer any direct help but 1 - can you fit the model with any other combination of parameters? 2 - what happens if you vary rho? 3 - if your data-set is small and not confidential can you share it, otherwise can you show us str(anxiety_control) or summary(anxiety_control) preferable without any variables you are not using in this model? 4 - there is a robust option in Wolfgang Viechtbauer's metafor package which might help although I am not sure how equivalent the analysis approaches are. On 01/09/2016 17:52, Kristina Loderer wrote:> Dear all, > > I am trying to fit a simple (intercept-only) meta-analytic model using > the robumeta package using the following code: > anx_cont_mean<-robu(formula = es_fisher ~ 1, var.eff.size > variance_fisher, studynum = study_ID, modelweights = "CORR", rho = 0.8, > small=TRUE, data = anxiety_control) > > When I try to run this model, the following error message pops up: > Error in solve.default(sumXWX) : > system is computationally singular: reciprocal condition number = 0 > > What exactly does this mean in the context of meta-analysis? I haven't > been able to find any answers. > > Thank you, > Kristina > > > ----------------------------------------- > Kristina Loderer > Ludwig-Maximilians-Universit?t M?nchen > Department Psychologie > Leopoldstr. 13 > D-80802 M?nchen > > Telefon: +49 (89) 2180-6047 > Email: Kristina.Loderer at psy.lmu.de > > ----------------------------------------- > > ______________________________________________ > R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. >-- Michael http://www.dewey.myzen.co.uk/home.html
Dear Michael, thank you for your reply! I managed to solve the problem in the meantime, there was an issue in the effect size conversion (correlations to Fisher's z) that lead to zero values for some effect size variances. Regards, Kristina ----------------------------------------- Kristina Loderer Ludwig-Maximilians-Universit?t M?nchen Department Psychologie Leopoldstr. 13 D-80802 M?nchen Telefon: +49 (89) 2180-6047 Email: Kristina.Loderer at psy.lmu.de ----------------------------------------->>> Michael Dewey <lists at dewey.myzen.co.uk> 02.09.16 13.38 Uhr >>>Dear Kristina I do not use that package so cannot offer any direct help but 1 - can you fit the model with any other combination of parameters? 2 - what happens if you vary rho? 3 - if your data-set is small and not confidential can you share it, otherwise can you show us str(anxiety_control) or summary(anxiety_control) preferable without any variables you are not using in this model? 4 - there is a robust option in Wolfgang Viechtbauer's metafor package which might help although I am not sure how equivalent the analysis approaches are. On 01/09/2016 17:52, Kristina Loderer wrote:> Dear all, > > I am trying to fit a simple (intercept-only) meta-analytic model using > the robumeta package using the following code: > anx_cont_mean<-robu(formula = es_fisher ~ 1, var.eff.size > variance_fisher, studynum = study_ID, modelweights = "CORR", rho 0.8, > small=TRUE, data = anxiety_control) > > When I try to run this model, the following error message pops up: > Error in solve.default(sumXWX) : > system is computationally singular: reciprocal condition number = 0 > > What exactly does this mean in the context of meta-analysis? I haven't > been able to find any answers. > > Thank you, > Kristina > > > ----------------------------------------- > Kristina Loderer > Ludwig-Maximilians-Universit?t M?nchen > Department Psychologie > Leopoldstr. 13 > D-80802 M?nchen > > Telefon: +49 (89) 2180-6047 > Email: Kristina.Loderer at psy.lmu.de > > ----------------------------------------- > > ______________________________________________ > R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guidehttp://www.R-project.org/posting-guide.html> and provide commented, minimal, self-contained, reproducible code. >-- Michael http://www.dewey.myzen.co.uk/home.html