# Re: [NMusers] PsN general

From: Sebastien Bihorel <sebastien.bihorel_at_cognigencorp.com>
Date: Thu, 3 Jan 2019 23:28:44 -0500 (EST)

Hi,

As I already mentioned to you in a separate thread:

1. SAS and PsN have different methods to calculate percentiles, so expecting these 2 pieces of softwares to give you the same answers is unrealistic.
2. VPC statistics are very much dependent on your data and the binning options you included in your vpc command.

That being said, and seeing your table in a much more reasonable format, I must also say that your questions are very much not clear:

* " a confidence interval for the 95 percentile at this time ". Which time are you referring to?
* " Is there a reason that this occurred". What does "this" refer to?
* "can it be corrected to give the proper N=14?" Where do you expect to see an N of 14?

So overall, it would help if you gave us more details, included the vpc command you submitted.

Cheers

Sebastien

From: "jacksonan1945" <jacksonan1945_at_gmail.com>
To: nmusers_at_globomaxnm.com
Sent: Thursday, January 3, 2019 9:23:31 PM
Subject: [NMusers] PsN general

I am running a VPC stratified on CMT with N=14. When I looked at my file VPC_results I found the following for my CMT=11 and CMT=23. This resulted in a confidence interval for the 95 percentile at this time to be much higher than calculated independently with SAS. Is there a reason that this occurred and can it be corrected to give the proper N=14?

Andre

VPC results strata CMT = 11

Continuous data

154 observations out of 294

< TIME

<=

median.idv

no. of obs

mean real

first interval is closed

-0.5

0.5

0

14

0

0.5

2.5

1.5

28

7.601134

2.5

3.5

3

14

9.25575

VPC results strata CMT = 23

Continuous data

140 observations out of 294

< TIME

<=

median.idv

no. of obs

mean real

first interval is closed

0.5

1.5

1

14

0.2890893

1.5

2.5

2

14

0.5110357

Received on Thu Jan 03 2019 - 23:28:44 EST

This archive was generated by hypermail 2.3.0 : Fri Sep 27 2019 - 17:02:02 EDT