Z Test

[Michael Score]

Hello, I just got SPSS 19.0 and am trying to run a Z test.  I am a bit of a novice with SPSS and can easily find the T test but cannot locate where to find the Z test within the software. 

 

Can anyone out there help?

 

Thanks.

[Bruce Weaver]

There are many variants on the z-test.  Which one are you trying to perform?  I.e., are you comparing two means, two proportions, or something else?

Michael Score wrote

Hello, I just got SPSS 19.0 and am trying to run a Z test.  I am a bit of a
novice with SPSS and can easily find the T test but cannot locate where to
find the Z test within the software.

Can anyone out there help?

Thanks.

[Bruce Weaver]

I received an off-list reply from Mike, but didn't see it right away, because noted in my sig file, I don't check my hotmail account very often.

--- start of reply from Mike ---

I am trying to run a one tailed z test. I am evaluating the difference between a sample mean from the sample's population mean.
Thanks for the help.
Mike

--- end of reply from Mike ----

I am not aware of a canned procedure for this.  But you can roll your own very easily.  Something like:

* Read in the data.
data list free / xbar mu sd (3f5.2) n (f5.0).
begin data
108 100 15 25
end data.

compute #se = sd/SQRT(n).
compute z = abs((xbar - mu) / #se).
* Compute the one-tailed p-value .
compute p = 2*(1 - cdfnorm(z)).
format z (f5.3) / p (f5.3).
list.

Variable sd is the population SD, of course.

Bruce Weaver wrote

There are many variants on the z-test.  Which one are you trying to perform?  I.e., are you comparing two means, two proportions, or something else?

Michael Score wrote

Hello, I just got SPSS 19.0 and am trying to run a Z test.  I am a bit of a
novice with SPSS and can easily find the T test but cannot locate where to
find the Z test within the software.

Can anyone out there help?

Thanks.

[David Marso]

Run a t-test using the pop mean as test value.
T-TEST TESTVAL n /VARIABLE=varlist.

Bruce Weaver wrote

I received an off-list reply from Mike, but didn't see it right away, because noted in my sig file, I don't check my hotmail account very often.

--- start of reply from Mike ---

I am trying to run a one tailed z test. I am evaluating the difference between a sample mean from the sample's population mean.
Thanks for the help.
Mike

--- end of reply from Mike ----

I am not aware of a canned procedure for this.  But you can roll your own very easily.  Something like:

* Read in the data.
data list free / xbar mu sd (3f5.2) n (f5.0).
begin data
108 100 15 25
end data.

compute #se = sd/SQRT(n).
compute z = abs((xbar - mu) / #se).
* Compute the one-tailed p-value .
compute p = 2*(1 - cdfnorm(z)).
format z (f5.3) / p (f5.3).
list.

Variable sd is the population SD, of course.

Bruce Weaver wrote

There are many variants on the z-test.  Which one are you trying to perform?  I.e., are you comparing two means, two proportions, or something else?

Michael Score wrote

Hello, I just got SPSS 19.0 and am trying to run a Z test.  I am a bit of a
novice with SPSS and can easily find the T test but cannot locate where to
find the Z test within the software.

Can anyone out there help?

Thanks.

[Bruce Weaver]

The t-test will not give exactly the same result, of course, because it uses the sample SD rather than the population SD, and it uses the t-distribution (with df = n-1) when calculating the p-value rather than the standard normal distribution.

Question for Mike:  Is the population SD really known?  If not, then you should be using the single-sample t-test David shows below.  

David Marso wrote

Run a t-test using the pop mean as test value.
T-TEST TESTVAL n /VARIABLE=varlist.

Bruce Weaver wrote

I received an off-list reply from Mike, but didn't see it right away, because noted in my sig file, I don't check my hotmail account very often.

--- start of reply from Mike ---

I am trying to run a one tailed z test. I am evaluating the difference between a sample mean from the sample's population mean.
Thanks for the help.
Mike

--- end of reply from Mike ----

I am not aware of a canned procedure for this.  But you can roll your own very easily.  Something like:

* Read in the data.
data list free / xbar mu sd (3f5.2) n (f5.0).
begin data
108 100 15 25
end data.

compute #se = sd/SQRT(n).
compute z = abs((xbar - mu) / #se).
* Compute the one-tailed p-value .
compute p = 2*(1 - cdfnorm(z)).
format z (f5.3) / p (f5.3).
list.

Variable sd is the population SD, of course.

Bruce Weaver wrote

There are many variants on the z-test.  Which one are you trying to perform?  I.e., are you comparing two means, two proportions, or something else?

Michael Score wrote

Hello, I just got SPSS 19.0 and am trying to run a Z test.  I am a bit of a
novice with SPSS and can easily find the T test but cannot locate where to
find the Z test within the software.

Can anyone out there help?

Thanks.

[David Marso]

CEFGW  
} ;-)>

[Bruce Weaver]

I copied that code from somewhere else where I was illustrating a one-tailed test, but did not do a great job of editing it.  Specifically, the "one-tailed p-value" should have been removed or edited, because the p-value that is computed is for a two-tailed test.  Here is a cleaned up version.  As before, the sd variable is the population SD.

* Read in the data.
data list free / xbar mu sd (3f5.2) n (f5.0).
begin data
108 100 15 25
end data.

compute #se = sd/SQRT(n).
compute z = abs((xbar - mu) / #se).
compute p = 2*(1 - cdfnorm(z)).
format z p (f5.3).
list.

p.s. - I'm still trying to work out what David's "CEFGW" means.  

Bruce Weaver wrote

--- snip ---

* Read in the data.
data list free / xbar mu sd (3f5.2) n (f5.0).
begin data
108 100 15 25
end data.

compute #se = sd/SQRT(n).
compute z = abs((xbar - mu) / #se).
* Compute the one-tailed p-value .
compute p = 2*(1 - cdfnorm(z)).
format z (f5.3) / p (f5.3).
list.

Variable sd is the population SD, of course.

Bruce Weaver wrote

There are many variants on the z-test.  Which one are you trying to perform?  I.e., are you comparing two means, two proportions, or something else?

Michael Score wrote

Hello, I just got SPSS 19.0 and am trying to run a Z test.  I am a bit of a
novice with SPSS and can easily find the T test but cannot locate where to
find the Z test within the software.

Can anyone out there help?

Thanks.

[David Marso]

"p.s. - I'm still trying to work out what David's "CEFGW" means.  "
CLOSE ENOUGH FOR GOVERNMENT WORK ;-))))

[Bruce Weaver]

Aha!  I should have got that.  I'll blame it on the Friday afternoon effect.  

David Marso wrote

"p.s. - I'm still trying to work out what David's "CEFGW" means.  "
CLOSE ENOUGH FOR GOVERNMENT WORK ;-))))