SPSSX Discussion

How to Compute p-values

Classic

List

Threaded

4 messages Options

E. Bernardo

How to Compute p-values

Hello everyone,

Below is the syntax I used to estimate the parameters of a conjoint model. Provided below also is the sample output of the syntax. ( Please dont bother of the large standard errors in the outputs because these are just hypothetical data.)

The syntax provides the mean utility and standard error for each factor category. Is there a way to compute the p-value associated for each mean utility and standard error. That is, can we add a line in the syntax asking spss to compute the p-values?

Or, as an alternative, is there a way to compute the p-value using only the mean and standard error?

Thanks.

Eins

CONJOINT PLAN 'c:\eins\Design2.sav'
   /DATA 'c:\eins\dataforanalysis.sav'
   /RANK = com_1 to com_12
   /SUBJECT=Res#
   /FACTORS CurrOff(discrete) Tuition(discrete less) Location Program Prof Instruction Facility ClassSched
/PLOT SUMMARY.

Sample Outputs:

Utilities
		Utility Estimate	Std. Error
CurrOff	MBA & Doctoral Degrees	1.667	1.126
CurrOff	Double Courses for MBA	-1.667	1.126
Tuition	P1500 Below	-.667	1.126
Tuition	P1500 Above	.667	1.126
Location	University Belt	-.167	1.126
Location	Near Place of Work	.167	1.126
Program	With Thesis	.333	1.126
Program	None-Thesis	-.333	1.126
Prof	Doctoral Degree Full-time Academician	-.500	1.126
Prof	Doctoral Degree with Industry Experience	.500	1.126
Instruction	Online	-1.833	1.126
Instruction	Classroom	1.833	1.126
Facility	Library Internet & Manual Payment System	-.833	1.126
Facility	Library Internet & Online Payment System	.833	1.126
ClassSched	Weekdays & Weekends	.667	1.126
ClassSched	Weekends Only	-.667	1.126
(Constant)		6.500	1.126

What are the top 10 Pinoy street foods?
Yummy! Find out on Yahoo! Answers

Bruce Weaver

Re: How to Compute p-values

Administrator

I suspect that you are asking how to compute a p-value for utility divided by its standard error. Is that right? I don't know anything about conjoint analysis. Does one use the standard normal here? Or a t-distribution? If the latter, what are the degrees of freedom?

When you have figured out which distribution you should be using, use the appropriate CDF function to get the p-value. E.g., if it is the standard normal you want, it will be something like this:

compute z = utility/SE .
compute p = 2*(1-(CDF.NORMAL(abs(z),0,1))) .
exe.

eins wrote

Hello everyone,

Below is the syntax I used to estimate the parameters of a conjoint model. Provided below also is the sample output of the syntax. ( Please dont bother of the large standard errors in the outputs because these are just hypothetical data.)

The syntax provides the mean utility and standard error for each factor category. Is there a way to compute the p-value associated for each mean utility and standard error. That is, can we add a line in the syntax asking spss to compute the p-values?

Or, as an alternative, is there a way to compute the p-value using only the mean and standard error?

Thanks.
Eins

CONJOINT PLAN 'c:\eins\Design2.sav'
   /DATA 'c:\eins\dataforanalysis.sav'
   /RANK = com_1 to com_12
   /SUBJECT=Res#
   /FACTORS CurrOff(discrete) Tuition(discrete less) Location Program Prof Instruction Facility ClassSched
/PLOT SUMMARY.

Sample Outputs:

Utilities

Utility Estimate

Std. Error

CurrOff

MBA & Doctoral Degrees

1.667

1.126

Double Courses for MBA

-1.667

1.126

Tuition

P1500 Below

-.667

1.126

P1500 Above

.667

1.126

Location

University Belt

-.167

1.126

Near Place of Work

.167

1.126

Program

With Thesis

.333

1.126

None-Thesis

-.333

1.126

Prof

Doctoral Degree Full-time Academician

-.500

1.126

Doctoral Degree with Industry Experience

.500

1.126

Instruction

Online

-1.833

1.126

Classroom

1.833

1.126

Facility

Library Internet & Manual Payment System

-.833

1.126

Library Internet & Online Payment System

.833

1.126

ClassSched

Weekdays & Weekends

.667

1.126

Weekends Only

-.667

1.126

(Constant)

6.500

1.126

"Try the new FASTER Yahoo! Mail. Experience it today at http://ph.mail.yahoo.com"

--
Bruce Weaver
bweaver@lakeheadu.ca
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

PLEASE NOTE THE FOLLOWING:
1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above.
2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/).

E. Bernardo

Re: How to Compute p-values

I need the formula to compute the p-value of a t-distribution. The sample size n is available. I hope somebody help me on this.

Thank you.

--- On Thu, 4/1/10, Bruce Weaver <[hidden email]> wrote:

From: Bruce Weaver <[hidden email]>
Subject: Re: How to Compute p-values
To: [hidden email]
Date: Thursday, 1 April, 2010, 5:33 PM

I suspect that you are asking how to compute a p-value for utility divided by
its standard error. Is that right? I don't know anything about conjoint
analysis. Does one use the standard normal here? Or a t-distribution? If
the latter, what are the degrees of freedom?

When you have figured out which distribution you should be using, use the
appropriate CDF function to get the p-value. E.g., if it is the standard
normal you want, it will be something like this:

compute z = utility/SE .
compute p = 2*(1-(CDF.NORMAL(abs(z),0,1))) .
exe.

eins wrote:

>
> Hello everyone,
> Â
> Below isÂ the syntax I used to estimate the parameters of a conjoint
> model.Â Provided below also is the sample output of the syntax.Â ( Please
> dont bother of the large standard errors in the outputs because these are
> just hypothetical data.)
> Â
> The syntax provides the mean utilityÂ and standard error for each factor
> category.Â Is there a way to compute the p-value associated for each mean
> utility and standard error.Â That is, can we add a line in the syntax
> asking spss to compute the p-values?
> Â
> Or, as an alternative, is there a way to compute the p-value using only
> the mean and standard error?
> Â
> Thanks.
> Eins
> Â
> CONJOINT PLAN 'c:\eins\Design2.sav'
> Â Â /DATAÂ 'c:\eins\dataforanalysis.sav'
> Â Â /RANK = com_1 to com_12
> Â Â /SUBJECT=Res#
> Â Â /FACTORS CurrOff(discrete) Tuition(discrete less) Location Program Prof
> Instruction Facility ClassSched
> Â /PLOT SUMMARY.
> Â
> Â
> Â
> Sample Outputs:
>
>
>
>
>
> Utilities
>
>
>
> Â
>
> Â
>
> Utility Estimate
>
> Std. Error
>
>
> CurrOff
>
> MBA & Doctoral Degrees
>
> 1.667
>
> 1.126
>
>
> Double Courses for MBA
>
> -1.667
>
> 1.126
>
>
> Tuition
>
> P1500 Below
>
> -.667
>
> 1.126
>
>
> P1500 Above
>
> .667
>
> 1.126
>
>
> Location
>
> University Belt
>
> -.167
>
> 1.126
>
>
> Near Place of Work
>
> .167
>
> 1.126
>
>
> Program
>
> With Thesis
>
> .333
>
> 1.126
>
>
> None-Thesis
>
> -.333
>
> 1.126
>
>
> Prof
>
> Doctoral Degree Full-time Academician
>
> -.500
>
> 1.126
>
>
> Doctoral Degree with Industry Experience
>
> .500
>
> 1.126
>
>
> Instruction
>
> Online
>
> -1.833
>
> 1.126
>
>
> Classroom
>
> 1.833
>
> 1.126
>
>
> Facility
>
> Library Internet & Manual Payment System
>
> -.833
>
> 1.126
>
>
> Library Internet & Online Payment System
>
> .833
>
> 1.126
>
>
> ClassSched
>
> Weekdays & Weekends
>
> .667
>
> 1.126
>
>
> Weekends Only
>
> -.667
>
> 1.126
>
>
>
> (Constant)
>
> 6.500
>
> 1.126
> Â
>
>
> "Try the new FASTER Yahoo! Mail. Experience it today at
> <A href='http://ph.mail.yahoo.com"/' target=_blank>http://ph.mail.yahoo.com"
>

-----
--
Bruce Weaver
bweaver@...
http://sites.google.com/a/lakeheadu.ca/bweaver/
"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.
--
View this message in context: http://old.nabble.com/How-to-Compute-p-values-tp28109723p28111586.html
Sent from the SPSSX Discussion mailing list archive at Nabble.com.

=====================
To manage your subscription to SPSSX-L, send a message to
LISTSERV@... (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

Start chatting with friends on the all-new Yahoo! Pingbox today!!
It's easy to create your personal chat space on your blogs

E. Bernardo

Re: How to Compute p-values

In reply to this post by E. Bernardo

On Thu, 1 Apr 2010 10:33:17 -0700, Bruce Weaver <[hidden email]>
wrote:

>I suspect that you are asking how to compute a p-value for utility
divided by
>its standard error. Is that right? I don't know anything about conjoint
>analysis. Does one use the standard normal here? Or a t-distribution?
If
>the latter, what are the degrees of freedom?
>
>When you have figured out which distribution you should be using, use the
>appropriate CDF function to get the p-value. E.g., if it is the standard
>normal you want, it will be something like this:
>
>compute z = utility/SE .
>compute p = 2*(1-(CDF.NORMAL(abs(z),0,1))) .
>exe.

I need the formula to compute the p-value of a t-distribution. The sample
size (n) is available. Hope there is generous one who would help me on
this.

Thank you.
Eins

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD