Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

- JMP User Community
- :
- Discussions
- :
- pooled /unpooled t test

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

Highlighted

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Jul 17, 2020 4:55 PM
(853 views)

Hi

What is the appropriate two sample t procedure for the summary below milking periods of cow breeds pooled/unpooled (Welch) using R?

Frisian cow Xbar=342.5 S1=10.2 n1=126

Jersey cow Xbar =348.6 S2=12.2 n2=39

Construct 95% confidence interval for µ1-µ2

Give proper interpretation of the interval

Thanks

1 ACCEPTED SOLUTION

Accepted Solutions

Highlighted

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

@Alma1 , I do not know what data you have, an example would be helpful.

In the below script (when you execute) you see a dataset and both t-test performed, giving different results.

Generally, statistical Tests have preconditions, and t-test assumes normal Distribution of the dataset,

pooled t-test assumes equal variance,

t-test works also with different variance.

This is because they work with mean and variance, and if that two numbers do not represent the data correctly, the method doesn't work.

If you perform a t-test, and the assumptions are not valid, you cannot rely on the result.

In the below example you can see different results for both t-Tests, but be Aware, that normal Distribution is also not valid in this case.

You can find These results in the documentation also, and I found STIPS a quite good training to understand this stuff.

```
Names Default To Here( 1 );
cdt = Open( "$SAMPLE_DATA\companies.jmp" );
cdt << Oneway(
Y( :Name( "Profits ($M)" ) ),
X( :Type ),
Means( 1 ),
t Test( 1 ),
Mean Diamonds( 1 ),
SendToReport(
Dispatch( {"Oneway Anova"}, "Summary of Fit", OutlineBox, {Close( 1 )} ),
Dispatch( {"Oneway Anova"}, "Analysis of Variance", OutlineBox, {Close( 1 )} ),
Dispatch( {"Oneway Anova"}, "Means for Oneway Anova", OutlineBox, {Close( 1 )} )
)
);
```

Georg

10 REPLIES 10

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: pooled /unpooled t test

Is this a homework exercise?

Why are you asking the JMP community to answer a question that is directed for a solution using R?

Why are you asking the JMP community to answer a question that is directed for a solution using R?

Jim

Highlighted
##

Nop! it is not a homework. we are conducting a research and I didnt know whether to use pooled /unpooled test whats wrong asking for ideas? you are free not to answer if you wish

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: pooled /unpooled t test

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: pooled /unpooled t test

Unpooled is needed when variance is different

Georg

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: pooled /unpooled t test

Georg thanks for your response. Do you mean population variance and should we calculate it before deciding on a pooled / unpooled test?

Highlighted

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

@Alma1 , I do not know what data you have, an example would be helpful.

In the below script (when you execute) you see a dataset and both t-test performed, giving different results.

Generally, statistical Tests have preconditions, and t-test assumes normal Distribution of the dataset,

pooled t-test assumes equal variance,

t-test works also with different variance.

This is because they work with mean and variance, and if that two numbers do not represent the data correctly, the method doesn't work.

If you perform a t-test, and the assumptions are not valid, you cannot rely on the result.

In the below example you can see different results for both t-Tests, but be Aware, that normal Distribution is also not valid in this case.

You can find These results in the documentation also, and I found STIPS a quite good training to understand this stuff.

```
Names Default To Here( 1 );
cdt = Open( "$SAMPLE_DATA\companies.jmp" );
cdt << Oneway(
Y( :Name( "Profits ($M)" ) ),
X( :Type ),
Means( 1 ),
t Test( 1 ),
Mean Diamonds( 1 ),
SendToReport(
Dispatch( {"Oneway Anova"}, "Summary of Fit", OutlineBox, {Close( 1 )} ),
Dispatch( {"Oneway Anova"}, "Analysis of Variance", OutlineBox, {Close( 1 )} ),
Dispatch( {"Oneway Anova"}, "Means for Oneway Anova", OutlineBox, {Close( 1 )} )
)
);
```

Georg

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: pooled /unpooled t test

million thanks for your help

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: pooled /unpooled t test

by the way jmp background is R.

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: pooled /unpooled t test

@Alma1 wrote:

by the way jmp background is R.

I'm not sure exactly what you're thinking of here but JMP has no background with R.

JMP is a product from SAS and is developed solely by us.

There is integration with R through JSL, providing the ability to call R and submit R code but that's the only association between JMP and R.

-Jeff

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: pooled /unpooled t test

I do not think of anything that might harm you or any other person. I meant integration why is it this much important I just used wrong wording, but this raised because one of the users interrupted me by questioning my question. Any person has the right to ask any scientific question and the viewers can choose to answer or ignore. What is wrong! Why is he even asking if the question is a homework is this something related to him? Do researchers have homework?!

I would suggest keeping up the academic value and providing help if needed instead of bothering what is asked.

I would suggest keeping up the academic value and providing help if needed instead of bothering what is asked.