About Liamb

SteveDenham · ‎07-23-2024

If you are locked into model selection techniques like stepwise, and you realize that it really isn't suited to mixed models, consider treating the difference between the treated and control subjects for each pair as your response variable. Without knowing your data structure, consider this: proc hpgenselect data=tab; class paired var1 var2 var3 var4 ; model wbc_diff= var1 var2 var3 var4/ dist = negbin link=log ; selection method=stepwise details=all; run; The key here is in calculating the response variable wbc_diff. There are several ways, but all of them depend on the current structure of the dataset "tab". You might be happier using method=lasso/choose=aicc, as it provides at lest some protection against type I error inflation due to multiple testing. SteveDenham

StatDave · ‎07-09-2024

I assume that what you mean is that you want the difference in the group means which is not what the DIFF option in the LSMEANS statement gives you. If you add the ILINK and CL options in the LSMEANS statement, Mean and confidence limits columns will be added in the LS-means table showing the individual group means and confidence intervals on the response. But the estimates in the LS-means differences table are differences of log means (or log ratios of means) and the values in the Exponentiated columns are ratios of means. To get estimated differences of means and confidence limits on the differences, use the NLMeans macro. See Example 1 in the Results tab of the macro documentation.

Liamb · ‎07-02-2024

Thanks a lot

Liamb · ‎05-20-2024

I'm not sure but I think it is the following formula for standard mean difference (SMD):

Ksharp · ‎08-17-2023

I don't understand, there is not Paired Wilcoxon test, Wilcoxon signed rank is just the non-parameter version of pair ttest ,it is for continuous variable ,not category variable. But McNemar test is for the paired category variable(a.k.a freq/count data). There are two different statistical method. And since it is about stat question,better post it at stat forum: https://communities.sas.com/t5/Statistical-Procedures/bd-p/statistical_procedures and @StatDave maybe give you a help.

Liamb · ‎08-10-2023

Thanks, it work

Kurt_Bremser · ‎03-17-2023

Maxim 19: Long Beats Wide. proc transpose data=have out=long; by id; var _numeric_; run; data want; set long; if _n_ = 1 then do; declare hash _1 (dataset:"long (where=(col1 = 1))"); _1.definekey("_name_"); _1.definedone(); end; if _1.check() = 0; run; The hash is just my tool of choice (no prior sorting needed), you could also do SQL joins or a DATA step MERGE.

PaigeMiller · ‎04-15-2022

@Liamb The %TRANSPOSE macro that I gave a link to will do this.

ballardw · ‎09-24-2021

@Liamb wrote: Thanks With these options I can only change the font (example font=B) in insetgroup but I want to change the size of the lines (axes and box) on the graph and the texts in insetgroup (example font_size=16), which options do I use? AXIS appearance options are in AXIS statements, set with the STYLE= option values 0 to 46 (0 means no line), WIDTH affects the axis line thickness unless Tick mark options also use width, in which case the ticks get the width, NOFRAME is required for this to change appearance for just the axis, otherwise the frame will get the thickness as well. Line options on which portion of the box to what? LBOXES=linetype or variablename if you want the different boxes with different lines. The value or variable would be limited to the integers 1 to 46 corresponding to the line type. Or perhaps it is time to change to Proc SGPLOT/SGPANEL.

Liamb · ‎03-15-2021

Sorry for my late reply I did not have the authorizations to modify the sasv9_usermods.cfg file and if I convert my Excel file to CSV I lose my values written in Kazakhe. Finally, I used SAS Unicode supported to process my data file

Liamb · ‎09-06-2018

Thanks a lot Rob

Online Status	Offline
Date Last Visited	‎07-18-2024 12:40 PM

Re: Proc hpgenselect with matched data

Proc hpgenselect with matched data

Proc genmod: Exponentiated with 95% confidence interval

Re: Proc genmod with normal distribution, log link, unstructured varia...

Proc genmod with normal distribution, log link, unstructured variance-...

Re: Propensity score and IPW with imputed data

Re: Propensity score and IPW with imputed data

Propensity score and IPW with imputed data

Re: Wilcoxon signed rank and Mc Nemar tests

Re: Wilcoxon signed rank and Mc Nemar tests

Re: Proc genmod: Exponentiated with 95% confidence interval

Re: Check matching criteria

Re: How to select several variables containing same value

Re: How to select several variables containing same value

Re: Macro transpose : How to add a 2nd variable

Re: Proc hpgenselect with matched data

Re: Proc genmod: Exponentiated with 95% confidence interval

Re: Proc genmod with normal distribution, log link, unstructured varia...

Re: Propensity score and IPW with imputed data

Re: Wilcoxon signed rank and Mc Nemar tests

Re: Check matching criteria

Re: How to select several variables containing same value

Re: Macro transpose : How to add a 2nd variable

Re: Ods graphic with proc boxplot : Font size of insetgroup, axis and ...

Re: Proc import : language issues

Re: Paired t-test and proc mianalyze using Rubin's method