Comparing Variances

[Jason Schoeneberger]

Hi Group,

 

Let’s assume I have test scores for two groups of students, group A and group B.  Let’s say that I notice that group A has more variance (less tight distribution) while group B has less variance (tighter distribution). Can I test, statistically, that there is more variance among A than B?

 

I’m not interested in knowing whether the mean is higher/lower, but whether group A has more variance in their scores than group B?

 

Any help is appreciated.

 

Jason

 

 

 

 

[Matthias Spörrle]

Hi Jason,

even though you are not interested in the significance of the mean difference, you should use the Independent-Samples T Test procedure as you will find the Levene's Test in the output which will answer your question of significant difference between the variances of the two groups.

Note that the dependent variable (at least interval data) has to be measured identically in both samples.

HTH
Matthias

On Tue, Feb 17, 2009 at 6:46 PM, Jason Schoeneberger <[hidden email]> wrote:

Hi Group,

 

Let's assume I have test scores for two groups of students, group A and group B.  Let's say that I notice that group A has more variance (less tight distribution) while group B has less variance (tighter distribution). Can I test, statistically, that there is more variance among A than B?

 

I'm not interested in knowing whether the mean is higher/lower, but whether group A has more variance in their scores than group B?

 

Any help is appreciated.

 

Jason

 

 

 

 

[Swank, Paul R]

If the groups are independent and reasonably distributed, then the F test will let you test the ratio of the variances. Just put the larger variance in the numerator and refer to F table with p = .025.

 

Dr. Paul R. Swank,

Professor and Director of Research

Children's Learning Institute

University of Texas Health Science Center-Houston

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Jason Schoeneberger
Sent: Tuesday, February 17, 2009 11:46 AM
To: [hidden email]
Subject: Comparing Variances

 

Hi Group,

 

Let’s assume I have test scores for two groups of students, group A and group B.  Let’s say that I notice that group A has more variance (less tight distribution) while group B has less variance (tighter distribution). Can I test, statistically, that there is more variance among A than B?

 

I’m not interested in knowing whether the mean is higher/lower, but whether group A has more variance in their scores than group B?

 

Any help is appreciated.

 

Jason

 

 

 

 

[Arthur Burke]

Jason ... In addition to the statistical test of the equality of variances that you can get from the T Test routine, the Explore option on the Analyze menu will give you statistics and plots useful for comparing the distributions of scores in the two groups. 

 

Art

---

From: Jason Schoeneberger [mailto:[hidden email]]
Sent: Tuesday, February 17, 2009 9:46 AM
To: [hidden email]
Subject: Comparing Variances

Hi Group,

Let’s assume I have test scores for two groups of students, group A and group B.  Let’s say that I notice that group A has more variance (less tight distribution) while group B has less variance (tighter distribution). Can I test, statistically, that there is more variance among A than B?

I’m not interested in knowing whether the mean is higher/lower, but whether group A has more variance in their scores than group B?

Any help is appreciated.

Jason

[David Hitchin]

Quoting Jason Schoeneberger <[hidden email]>:

> Let's assume I have test scores for two groups of students, group A
> and group B.  Let's say that I notice that group A has more variance
> while group B has less variance.
> Can I test, statistically, that there is more variance among A than B?
>
The previous answers are appropriate, but I'm a little worried about the
question.

If, before obtaining the test scores, there was a decision to test for
differences in variance, then the tests proposed would give the correct
p-values.

However, if after conducting the test the researcher just happens to
notice that there is something apparently interesting about the data,
and then conducts a test, the p-values will not be valid.

The usual procedure for testing, say at the 5% level, is such that when
you are conducting a large number of tests and there are no more than
chance differences between the two samples, only 5% of the tests will
have a type 1 error (wrongly concluding that there is a population
difference between the groups).

If you only test those groups for which you see something interesting,
such as a difference in means, variances, skews or kurtosis, then you
are picking the tests most likely to show a significant result, and you
will make more than 5% type 1 errors.

Consider the two possibilities; (1) you bet on a horse before the race
and (2) you bet on a horse after the race when you have been told that
it finished among the front runners. (1) is a fair bet, but if you find
anyone offering good odds for (2), please let me know.

David Hitchin

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

[Guerrero, Rodrigo]

Hello everyone,

 

I just upgraded to V17.0.1 and I am having trouble working with time and date fields.  May be you all can help me.

 

I have a simple date in the format yyyy/mm/dd and a time field in the format of hh:mm as follows:

 

CreateDate  CreateTime

20009/01/26 10:19

2009/01/27 14:50

 

 

I am trying to run a simple command from this data: 

 

 IF (CreateDate = 2009/01/26 and CreateTime >=10:19) var1 = 1

 

 

But I cannot get it to run.  I get the following error:

 

Error # 4007 in column 48.  Text: :

The expression is incomplete.  Check for missing operands, invalid operators,

unmatched parentheses or excessive string length.

This command not executed.

 

 

Can anyone tell me what I am doing wrong?  I have tried putting quotes around my values, but I get a different error about data types.

 

Thank you.

 

RG

 

Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317

 

The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information.

[Oliver, Richard]

Date and time literal strings are problematic in expressions. Try:

 

if CreateDate=date.dmy(26,1,2009) and CreateTime>=time.hms(10,19) var1=1.

 

---

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Guerrero, Rodrigo
Sent: Friday, February 20, 2009 10:30 AM
To: [hidden email]
Subject: Working with Dates

 

Hello everyone,

 

I just upgraded to V17.0.1 and I am having trouble working with time and date fields.  May be you all can help me.

 

I have a simple date in the format yyyy/mm/dd and a time field in the format of hh:mm as follows:

 

CreateDate  CreateTime

20009/01/26 10:19

2009/01/27 14:50

 

 

I am trying to run a simple command from this data: 

 

 IF (CreateDate = 2009/01/26 and CreateTime >=10:19) var1 = 1

 

 

But I cannot get it to run.  I get the following error:

 

Error # 4007 in column 48.  Text: :

The expression is incomplete.  Check for missing operands, invalid operators,

unmatched parentheses or excessive string length.

This command not executed.

 

 

Can anyone tell me what I am doing wrong?  I have tried putting quotes around my values, but I get a different error about data types.

 

Thank you.

 

RG

 

Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317

 

The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information.

[Guerrero, Rodrigo]

Thanks Richard.  That works great!

 

RG

 

Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317

 

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Oliver, Richard
Sent: Friday, February 20, 2009 10:43 AM
To: [hidden email]
Subject: Re: Working with Dates

 

Date and time literal strings are problematic in expressions. Try:

 

if CreateDate=date.dmy(26,1,2009) and CreateTime>=time.hms(10,19) var1=1.

 

---

From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Guerrero, Rodrigo
Sent: Friday, February 20, 2009 10:30 AM
To: [hidden email]
Subject: Working with Dates

 

Hello everyone,

 

I just upgraded to V17.0.1 and I am having trouble working with time and date fields.  May be you all can help me.

 

I have a simple date in the format yyyy/mm/dd and a time field in the format of hh:mm as follows:

 

CreateDate  CreateTime

20009/01/26 10:19

2009/01/27 14:50

 

 

I am trying to run a simple command from this data: 

 

 IF (CreateDate = 2009/01/26 and CreateTime >=10:19) var1 = 1

 

 

But I cannot get it to run.  I get the following error:

 

Error # 4007 in column 48.  Text: :

The expression is incomplete.  Check for missing operands, invalid operators,

unmatched parentheses or excessive string length.

This command not executed.

 

 

Can anyone tell me what I am doing wrong?  I have tried putting quotes around my values, but I get a different error about data types.

 

Thank you.

 

RG

 

Rodrigo A. Guerrero | Director Of Marketing Research and Analysis | The Scooter Store | 830.627.4317

 

The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information.

The information transmitted is intended only for the addressee(s) and may contain confidential or privileged material, or both. Any review, receipt, dissemination or other use of this information by non-addressees is prohibited. If you received this in error or are a non-addressee, please contact the sender and delete the transmitted information.