Minor learning curve here but worth it:

Download the R package at http://www.r-project.org/

Install the 'rms' package in R. A 'nomogram' function is part of the package.

If this is a logistic regression, then you want something like

mdl=lrm(Y~x1+x2+x3+x4+x5,data=myData)

nomogram(mdl,fun=plogis)

Thank you very much; I downloaded R package but kindly tell me how to install the 'rms' package in R?

type

help("install.packages")

in R

I did so but a window opened as in the attached file?!

Right. Hit 'Get List' from the top left and select 'install dependencies' from the bottom right. Then select 'rms' from the list produced by 'Get List'. Finally, type in "library(rms)" and you will be ready to go

The list that appeared is for countries!!

I have chosen UK(Bristol) then rms and it should be installed now.

So select the country you're in.

If it's installed, type library(rms) and then help("nomogram") and help("rms")

were shall I type library(rms)?

I got it, thanks for your kind patience.

Dear Mr Maltenfort.

I am an ophthalmologist, neither statistic specialist nor programmer,

Would you please advise me how to use the software.

Thank you very much

As important as having capable software can be, I would start with the hypotheses. What is are the theoretical effects of these 5 therapies? Are you combining them? i.e. is this a genuine factorial design...or simply a horse race? If the former, what are the mechanisms and do they lead to predictions of synergism (i.e. interactions)? Laying these out should preceed what machinery will be used to test the deduced hypotheses. Once you're to that point, any flavor of linear modeling will probably be able to handle things. The type of linear model will depend on the way in whcih the dependent variable changes...what ditribution arises from the its mechanisms? (generalized vs. not) Is there any nesting (mixed/correlated error vs. not) ...and so on.

according to our primary observations of the results of the treatment these five factors actually affect the results; for example under age 30 y/o there was undercorrection while over 30 y/o there was overcorrection.

Yes, I want to combine them.

It is a genuine factorial design.

There are mechanisms explaing their effects.

I do not know whether there are predictions of synergism, and this is what I want to find out!

Thank you

First of all, you medical study should include experiments and observations in all levels in order determine the effect of each factor on one another. Two way or multiple analysis of variance in your study are factors you can use to determine response variable as well a interractions between factors. Note that factorial experiment would have four treatment combinations in total using 2*2 factorial design. Donald. Medical Reserach, USA.

Sounds like you should find a thoughtful collaborator to help you navigate through the higher-order aspects of your design and appropriately model the effects. It's very easy to lose track...and no automated or semi-automated system that I know is smart enough to understand YOUR factors and how they apply to the real world.

Do you need to LEARN how to do and interpret multivariate stats, or just need software? I use SPSS but any stat package will do it. If you need training, I HIGHLY recommend you get a multivariate course from the nearest university. The course has to be at the PhD level to include relevant data. From that, you can then teach yourself procedures the class does not teach. Dr. Zinn

Just read post above mine and it is accurate. Basically, if you have not done or interpreted multivariate stats, you will spending a year or more to learn and utilize correctly for article publication. BTW, SPSS does do modeling but you have to know how to create SPSS code (called syntax). If you really want precision, use R.

Thank you all

Indeed, I have never done or interpreted multivariate stats;

but may I ask: I have R, and need to know how to use it.

many thank for all.

I think I might give you some useful tips here. I'm a physician and epidemiologist with good statistical and computer skills.

You are actually asking how to estimate the effect size of 5 selected factors on an outcome, which is treatment success. What Andrea previously said is a basic fact, you need first to specify the level of measurement for each variable, before we can proceed to measure the effect size of each factor in a univariate model first and after that proceed to multivariate modeling.

I agree with Mr. Jason that you really need guidance (thoughtful collaborator), since the other alternative of studying statistical package like "R" or attending PhD level statistics course is not a practical alternative for your current research question, however you might give it a strategic thought if you plan to carry on with research career.

For factorial analyses, you need to read 'exploratory factor analysis' and 'confirmatory factor analysis' in any book on multivariate statistics. For example: Discovering statistics using SPSS baby Andy Field OR ' Multivariate Data Analysis by Joseph F. Hair Jr, William C. Black, Barry J. Babin, and Rolph E. Anderson.

Good luck

I advise you to teach yourself on a daily basis. That is what Ido.

For learning R I advise to use www.datacamp.com. But learning takes some time and maybe it would be faster to hire an accomplished statitian who knows HOW to do it and you could just focus on WHAT you want him to do.

You can use a comparison between different factors and show the best using statistical method and comparisons using statical programs as SPSS or R