# Statistics – Attribute or Variable Data

Six Sigma – iSixSigma › Forums › General Forums › Methodology › Statistics – Attribute or Variable Data

- This topic has 6 replies, 5 voices, and was last updated 7 years, 2 months ago by Cliff Norman.

- AuthorPosts
- May 31, 2013 at 2:39 am #54424

CarolineParticipant@cazzasibley**Include @cazzasibley in your post and this person will**

be notified via email.I have a question if I have variable data ( that is not normally distributed) I then transferred it to Attribute data and worked out the DPMO from the opps/defects. If the DPMO is normally distributed can I carry on using stats such at t – tests etc. Or because it is originally attribute data I should use chi squared etc? Any advise appreciated.

0May 31, 2013 at 3:39 am #195319

Vyanktesh KulkarniGuest@kulkarni.vyanktesh**Include @kulkarni.vyanktesh in your post and this person will**

be notified via email.Hi,

Why don’t you think of using non-parametric test instead of converting the data to attribute?

0May 31, 2013 at 6:22 am #195320

Mayank GuptaParticipant@mayankgupta**Include @mayankgupta in your post and this person will**

be notified via email.Typically Attribute data is converted to Variable data (Discrete or Continuous) so that statistical tests could be applied.

As suggested you can directly use non-parametric tests on your original data set. Further only a variable continuous data can be converted to variable discrete or attribute but not vice versa. So there is no way that you can use any of the parametric tests posts such conversion.0May 31, 2013 at 7:46 am #195321

Chris SeiderParticipant@cseider**Include @cseider in your post and this person will**

be notified via email.@cazzasibley

It sounds like you are getting enamored with the statistics. I’m not sure what you are trying to solve but if your project is to reduce the ppm defective of the variable being measured with continuous type of data, I’d much more suggest finding a curve that best fits your data (Weibull or 3-parameter Weibull) or use the Distribution Identification in Minitab and find the ppm defective outside your spec limits using non-normal capability analysis found in Minitab. Then keeping the same distribution type, calculate the new ppm defective of the aspect you are measuring. I’m assuming you are trying to make ppm defective as a metric which is QUITE acceptable.If you want to prove a statistical difference in the mean/median/standard deviation shift, you can use non-parametric statistics without doing the hard work you are attempting.

Converting continuous data to attribute is ALWAYS the wrong approach… My two cents. :)

0May 31, 2013 at 12:40 pm #195323

Mohamed SalaheldinParticipant@msalaheldin**Include @msalaheldin in your post and this person will**

be notified via email.I also aggree with Chris, Do Not convert continuous data to attribute.

0June 3, 2013 at 12:42 am #195335

Vinodkumar.BMember@vinod059**Include @vinod059 in your post and this person will**

be notified via email.I ALSO AGREE,

0June 3, 2013 at 8:48 am #195343

Cliff NormanGuest@cliffnorman**Include @cliffnorman in your post and this person will**

be notified via email.As others have said, stay with the continuous data. Before doing anything else put the data on an appropriate control chart and learn from the special causes. As Shewhart noted: things in nature are stable, man made processes are inherently unstable. I have taken this from Shewhart’s postulates. T test and other tests all rest on the assumption of IID; Independent and Identically Distributed. If there are special causes present these assumptions are violated and the tests are useless. Even though the “control chart” show up in DMAIC under C for many novices, it should be used early. Getting the process that produced the data stable is an achievement. It is also where the learning should start. Calculating DPMO, and other outcome measures can come later; after learning and some work. Best, Cliff

0 - AuthorPosts

You must be logged in to reply to this topic.