SPSSX Discussion

calculate t-scores based on all responses of each participant

Classic

List

Threaded

10 messages Options

Irie

calculate t-scores based on all responses of each participant

hello, i need to calculate t-scores for my physiological data based on all
responses of each participant (instead of all participants) in order to
increase comparability of individual parameters between subjects. How can I
do this? Is there a tool for this in SPSS? thank you!

--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

=====================
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

Andy W

Re: calculate t-scores based on all responses of each participant

Can you give an fake dataset and then an example calculation you want to do?
Not exactly sure what you mean by your current description.

-----
Andy W
[hidden email]
http://andrewpwheeler.wordpress.com/
--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

=====================
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

Andy W
apwheele@gmail.com
http://andrewpwheeler.wordpress.com/

Ware, William B

Re: calculate t-scores based on all responses of each participant

Assuming that you want to convert raw scores to McCall's T-scores (mean = 50, sd = 10)...

Create a new variable by z = (raw - mean of raw)/sd of raw... Then change it T = 50 + 10*z

At least this is what I think is being asked... If you want to make the distribution normal, you will have to go through an area transformation...

wbw

William B. Ware, Retired Professor
Educational Psychology, Measurement, and Evaluation
Learning Sciences and Psychological Studies
University of North Carolina at Chapel Hill
McMichael Term Professor of Education, 2011-2013
Adjunct Professor, School of Social Work
Academy of Distinguished Teaching Scholars at UNC-Chapel Hill, Charter Member
EMAIL: [hidden email]

-----Original Message-----
From: SPSSX(r) Discussion <[hidden email]> On Behalf Of Andy W
Sent: Thursday, August 16, 2018 11:01 AM
To: [hidden email]
Subject: Re: calculate t-scores based on all responses of each participant

Can you give an fake dataset and then an example calculation you want to do?
Not exactly sure what you mean by your current description.

-----
Andy W
[hidden email]
http://andrewpwheeler.wordpress.com/
--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

=====================
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

=====================
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

Mike

Re: calculate t-scores based on all responses of each participant

In reply to this post by Irie

By T scores I assume that you want to convert your original scale data

into one the has a mean of 50 and a standard deviation of 10 but using

the subject mean and subject SD instead of the group mean and SD

for a specific variable.. I can provide some code but others may have

more efficient code to use.

First, compute the mean of the original scores, say Y1 to YK: K should

be replaced by the actual number (especially in vector and loop below)

or the last score name.

compute Y_SubjMean=mean(Y1 to YK). /* Note: check for missing data.

Second, compute the SD of the original scores Y1 to YK.

compute Y_SubjSD=SD(Y1 to YK). /* Check for missing data.

Convert original scores into z-scores and then convert z-scores

into T scores.

Vector Y = Y1 to YK.

Vector Z = Z1 to ZK(K)..

Vector T = T1 to TK(K). /* You can provide a format like (F5.2).

Loop #I = 1 to K.

compute Z(#I)= (Y(#I) - Y_SubjMean)/Y_SubjSD.

compute T(#I)= (Z(#I) + 50) * 10.

End loop.

If there are any errors above (syntactic or logical), one of the list

readers should point it out.

If you have missing data, things become a little more complicated

and you'll have to make decisions about how to deal with those cases.

HTH

-Mike Palij

New York University

[hidden email]

On Thu, Aug 16, 2018 at 10:12 AM, Irie <[hidden email]> wrote:

hello, i need to calculate t-scores for my physiological data based on all
responses of each participant (instead of all participants) in order to
increase comparability of individual parameters between subjects. How can I
do this? Is there a tool for this in SPSS? thank you!

--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

===================== 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

Irie

Re: calculate t-scores based on all responses of each participant

Cool, thank you Mike! That`s what I need, you got me.

I tried the first compute and it didn`t work. You said I should replace K
with the actual number or the last score name, what do you mean by that?
shouldn`t it be the case number? But than it was Y1 to Y1...

Thank you!!

--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

=====================
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

Mike

Re: calculate t-scores based on all responses of each participant

I use "YK" to indicate the last score that a person has. You don't

identify how many scores you have for each person but let's assume

that you have five scores: Y1, Y2, Y3, Y4, and Y5. In this case

K = 5 and the SPSS code would be:

Vector Y = Y1 to Y5.

Vector Z = Z1 to Z5(5)..

Vector T = T1 to T5(5).

and

Loop #I = 1 to 5.

So K just refers to the number of original scores you have as well

as the number of T scores corresponding to the original scores.

-Mike Palij

New York University

[hidden email]

On Thu, Aug 16, 2018 at 1:10 PM, Irie <[hidden email]> wrote:

Cool, thank you Mike! That`s what I need, you got me.

I tried the first compute and it didn`t work. You said I should replace K
with the actual number or the last score name, what do you mean by that?
shouldn`t it be the case number? But than it was Y1 to Y1...

Thank you!!

--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

=====================
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

Mike

Re: calculate t-scores based on all responses of each participant

In reply to this post by Irie

One more thing: I wrote the T score formula wrong. I originally wrote:

compute T(#I)= (Z(#I) + 50) * 10.

But it should be:

compute T(#I)= 50 + (Z(#I) * 10).

The z-score should first be multiplied by the SD=10 and then added

to the T score mean = 50. Sorry for the error.

-Mike Palij

New York University

[hidden email]

On Thu, Aug 16, 2018 at 1:10 PM, Irie <[hidden email]> wrote:

Cool, thank you Mike! That`s what I need, you got me.

I tried the first compute and it didn`t work. You said I should replace K
with the actual number or the last score name, what do you mean by that?
shouldn`t it be the case number? But than it was Y1 to Y1...

Thank you!!

--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

=====================
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

Art Kendall

Re: calculate t-scores based on all responses of each participant

In reply to this post by Mike

Is there a reason you are using T-scores ratrher than Z-scores?

Are T-scores more commonly used in some specialty area?

I terms of ipsatizing data, z-scores are much more common.

-----
Art Kendall
Social Research Consultants
--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

=====================
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

Art Kendall
Social Research Consultants

Rich Ulrich

Re: calculate t-scores based on all responses of each participant

For continuous measures, ipsatized data lose both the mean and SD. Those usually

matter, in my usual data. I can't remember when I may have ipsatized scores.

So, I don't know what is used by people who ipsatize, but I have a strong preference

for showing T-scores rather than z-scores. They are much easier to look at -- 2 digits

instead of 4 or 5, no minus-signs to grab or confuse the eye (especially if you want to

show ranges). T-scores usually serve well as composite scores of multiple items or

sub-scores, whenever the raw mean is not meaningful.

Rich Ulrich

From: SPSSX(r) Discussion <[hidden email]> on behalf of Art Kendall <[hidden email]>
Sent: Friday, August 17, 2018 11:52 AM
To: [hidden email]
Subject: Re: calculate t-scores based on all responses of each participant

Art Kendall

Re: calculate t-scores based on all responses of each participant

The only time I have seen ipsatized data used was in Q Factor analyses as a
means of fuzzy clustering.

In the many realizations of the General Linear Model (regression, ANOVA,
etc.) it seems that Mean and SD would be important.

Perhaps the OP would clarify the rationale and circumstances.

-----
Art Kendall
Social Research Consultants
--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

=====================
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

Art Kendall
Social Research Consultants