Request for help (quntative estimation of two sampes
Six Sigma – iSixSigma › Forums › Old Forums › General › Request for help (quntative estimation of two sampes
 This topic has 9 replies, 4 voices, and was last updated 17 years, 1 month ago by Lomax.

AuthorPosts

August 17, 2004 at 1:14 pm #36566
Dear All, please help
Does anyone know how to estimate the differance in means of two sets of samples.
For example, if we have tested new device in technology process, how can we estimate the differance in average time spending for this process if we have two sets: 1st is the time for process (averaging value for each hour ) during 1 week before deployment of new device and 2d is the same one after staring device usage.
Thanks in advance.
Regards
Eugene0August 17, 2004 at 3:37 pm #105857Depending . This is a general answer based on the description provided
You can use ttest or Anova analysis.
Before doing that you must take a look at the normality test (andersondarling) and look to the graphics looking for some cyclic behavior. (use one column for “before” data , and another for “after” data…)
Maybe you would need to use a paired ttest …
these tests provide you easily ( in minitab) the estimated difference among 2 means in the results window. The significance depends on the pvalue .
BR
0August 17, 2004 at 3:37 pm #105858Eugene,
PairedT test.0August 18, 2004 at 7:00 am #105905Renato, thank you for help at first.
But I still have some question, I just started to learn statistics, so the question.
I use ANOVA single factor annalysis, as I am aware, the rations between F and F critical determine the statistical significans of the two means, but it does not answer the question about in what times it less/bigger. I am interested in quantative estimation.
You say the pvalue represent the quantative estimation, does not it?
Could you tell me about this, please?
Thank you again
Regards
Eugene0August 18, 2004 at 7:02 am #105906Ritz, thank you for help .
But I still have some question, I just started to learn statistics, so the question.
I use ANOVA single factor annalysis (it is the Ttest also?), but it does not answer the question about in what times it less/bigger. I am interested in quantative estimation.
Can you explain what factors are responsible for quantative ratios between means?
Thank you again
Regards
Eugene0August 18, 2004 at 1:44 pm #105916Eugene,
First: You are attempting to determine if sample1 is different from sample2. In comparing only 2 means, the ttest is the appropriate test, not ANOVA (Ftest).
Second: Your samples seem to be dependent, essentially indicating that the only change to the technology process is the new device. If so, then the paired ttest is the correct test to use for this situation. If you are using Minitab, but the “before device installed” hourly average time data in column 1, and the “after device installed” hourly average time data in column 2. You will want to check for normality before using the paired ttest. If the data are not normal, use the nonparametric alternative (sign test or signed rank test).
Third: Assuming normal data and using the paired ttest, a pvalue below 0.05 indicates a significant difference between the two sample means. Understanding which sample has a larger mean is as simple as looking at the results table — the sample means are given.
Fourth: Assuming all of the above (no confounding process changes, normality, setup and run test correctly, etc.), you can state (given a level of confidence, typically 95%) that the device caused a significant change in process mean.
Hope this helps provide some clarity – please provide further clarification of what you are looking for if this post does not help.
Regards.
Ritz0August 19, 2004 at 7:05 am #105969Ritz, Thank you a lot for this complete information!
Yesterday I have improve my statistic lnowlrge and today I am understand you entirely.
I have additional quaestions, why ANOVA is inappropriate test, I mean if befor testing I phave berformed the AndersonDarling test then ANOVA is suitable, is not it?
And I wil try to use your goodness one more time. You see I am from Ukraine and have no softwares tatistical tools, so I rty to eleborate the algorithm and implemented it by myself. So, Can you suggest me some links to the web sights with such detailed explanation?
Thank you a lot, you have helped me very greatly.
Regards
Eugene
0August 30, 2004 at 2:35 pm #106602Sorry for the delay, Eugene. Took some time off. Nice to hear from a fellow Ukranian (my Father’s side came to America in c.1900)
It isn’t so much that the ANOVA is inappropriate – it is essentially a ttest as well. ANOVA is a robust test — it is likely to pick up significance in your case. The question is more of precision of testing methods – the pairedt test is more appropriate since your samples are dependent on each other. ANOVA and 2sample ttest use an assumption of independence. ANOVA and 2sample ttest will yield the same result when comparing 2 (and only 2!) samples.
A few websites I have found useful for knowledge and reference are:
http://davidmlane.com/hyperstat/
http://www.statsoft.com/textbook/stathome.html
http://statwww.berkeley.edu/users/stark/SticiGui/Text/index.htm
I would recommend purchasing a stats software tool when you have the money and opportunity – Minitab is commonly used in 6Sigma deployments, but Jump, Statistica, SPSS, are also good packages. At minimum, Excel has an ANOVA and ttest module you can load in for free…go to “tools” menu, select “addins” and look for the “data analysis” addin.
Good luck to you.
Regards,
Ritz0August 30, 2004 at 2:49 pm #106603Ritz, hello.
Thank you for your information. It is very useful for me.
I am using excel data analysis tool now, so thank you for your advices.
It is very pleasent to hear that there are some connections between you and Ukraine.
So I will rise my level in statistics.
Have a nice day!
Regards
P.S. If you have wish to visit a Kiev please let me know, I hope I will help you with this.0August 30, 2004 at 3:21 pm #106605Hi everybody,
Personally, I use the following for test between two samples:
Comparison of Means::: If sample size <30 – t test . else. ztest
Comparison of variances :: F Test
These are based on statistical equations and are quite robust. I have made them into easily usable excel file – But do not know how to attach it here. If anybody is interested, can write to me at [email protected] – provided of course, you promise to help me in future if I have a technical problem and you are in a position to help ;)
Neil
0 
AuthorPosts
The forum ‘General’ is closed to new topics and replies.