Hypothesis Testing – Analyze Phase
Six Sigma – iSixSigma › Forums › Old Forums › General › Hypothesis Testing – Analyze Phase
 This topic has 13 replies, 13 voices, and was last updated 13 years, 11 months ago by Yadav.

AuthorPosts

October 10, 2003 at 4:38 pm #33539
I have a GreenBelt candidate working on a project in the Analyze Phase. She is trying to determine the best Hypothesis Test to use (in Minitab) and is having some problems.
The data that she is using is Discrete Data, where the information is either True or False or 1 or 2… type of data.
What would be the best type of Hypothesis test to use? Can you please suggest a few different tools to use?
Thank you!0October 10, 2003 at 4:45 pm #90871
BirdmanParticipant@Birdman Include @Birdman in your post and this person will
be notified via email.I would use a 2 proportion test as the data is Discrete
0October 10, 2003 at 5:07 pm #90873
facemanParticipant@faceman Include @faceman in your post and this person will
be notified via email.I agree with birdman that a two proportion test is a good option. Also, chi squared test would be appropriate, especially if sample size is small.
0October 16, 2003 at 5:31 am #91091
Alexis DiomampoParticipant@AlexisDiomampo Include @AlexisDiomampo in your post and this person will
be notified via email.You could use this as guide:
Type of Data Type of Test
1 Proportion data = 1 Proportion test
2 Proportion data = 2 Proportion Test
3 or more proportion = Contingency Table0October 16, 2003 at 6:37 am #91093
Jan AkerstromParticipant@JanAkerstrom Include @JanAkerstrom in your post and this person will
be notified via email.Try JMP instead of MINITAB. The structure of the software helps you determine the “correct” analysis. You classify the binary data as Nominal ord Ordinal and use the test probability function in the distribution platform.
0October 16, 2003 at 8:53 am #91099
HemanthParticipant@Hemanth Include @Hemanth in your post and this person will
be notified via email.Hi Brett,
Well you would have got the answer from the previous posts. What bugged me was your green belt not being able to figure out the test for a simple application. So I guess there must be some complication to that and if not then you might consider checking the training program for your green belts.
0October 16, 2003 at 9:27 am #91102
Chandrashekhar DeshmukhParticipant@ChandrashekharDeshmukh Include @ChandrashekharDeshmukh in your post and this person will
be notified via email.Dear Mr. Brett,
Thank you very much for giving us the opportunity to share our thoughts.
Method of collecting data in % figures will give more input to best analysis which inturn will give digitised answer.
Thanking you,
Chandrashekhar Deshmukh
0October 16, 2003 at 10:24 am #91106Have a look at the tables of tests for different types of data in the link below from the British Medical Journal.
http://bmj.bmjjournals.com/collections/statsbk/13.shtml
It is a good summary of what can be done. Minitab should be able to do most if not all.
Data is data so do not get hung up about medical things.
0October 16, 2003 at 1:37 pm #91124
Judy KlomanParticipant@JudyKloman Include @JudyKloman in your post and this person will
be notified via email.I am working a transactional process where my data is discrete. I have been able to successfully use 2 sample t hypothesis testing. The practical problem being: is A equal to B or better than.
0October 16, 2003 at 2:09 pm #91128Hi, Brett.
I learned a matrix of tools about analysis phase:
If you Y response variable is DISCRETE and you have only one X DISCRETE factor, you can use Chi Square Test or 2 Proportion Test.
If you Y response variable is CONTINUOS and you have two or more X DISCRETES factors, you must use ANOVA.
Regards.
Juan0October 18, 2003 at 2:29 pm #91208How do I get JMP software?
0October 21, 2003 at 4:43 pm #91298
sathish.NMember@sathish.N Include @sathish.N in your post and this person will
be notified via email.1) You should use Chi Square if only your “Y” (Output) & “X” (Input/Process variables ) factors are discrete data type or You can compare two data (Ex: June & July) to come out with Variation in Outputs
2) You should use 2 Sample Ttest to determine whether or not two processes or data set have equal means (or averages). Perform Hypothesis where:
Ho = Mean of Sample A is Equal to Mean of Sample B(Pvalue>.05)
Ha = Mean of Sample A is Not Equal to Mean of Sample B (PValue<0.05) .This is in subjective to Transaction Process, for manaufacturing industry use P value as 0.01.0October 21, 2003 at 5:27 pm #91300You need to purchase it. See the third banner down on the bottom of this page. It’s a link for JMP in blue with the red border.
Tim0February 7, 2008 at 12:29 pm #168302CAN YOU PROVIDE THE MATRIX OF POTENTIAL VARIBLE FOR TEST
0 
AuthorPosts
The forum ‘General’ is closed to new topics and replies.