Attribute or Variable
Six Sigma – iSixSigma › Forums › Old Forums › General › Attribute or Variable
 This topic has 18 replies, 7 voices, and was last updated 12 years, 6 months ago by Truth Teller.

AuthorPosts

February 1, 2010 at 5:20 pm #53199
Hi,
I have a process where a client raises a ticket requesting information or other. Our teams works on the same and closes the ticket by providing the right resolution. Our software tracks the ticket requested time and closure time. Difference is what we call as TAT.
Now since the Time is tracked and calculated by software automatically, will my Y i.e., TAT is attribute or variable?
I understood that a variable data can take a range, whereas attribute data cannot. Since in my above case it cannot take any range when measured as it is done my software, how should I consider the same i.e., variable or attribute.
Thanks in advance.0February 1, 2010 at 5:42 pm #188918Hi,
I have one more doubt. A variable data can take any value in a range or can be broken down into smaller parts and still have meaning. Correct?
One of my process generates defects. I am counting the same and these are attribue data. Using the count of defects I am calculating the percentage defective using the formula (Defects/Produced)*100. Now since the result can also be broken down into decimals, should I call the percentage defective as Variable data? Same in the case of accuracy, using the formula (1Defects/Produced)*100, should I call accuracy as variable or attriute?
Please advice.0February 1, 2010 at 5:42 pm #188919
StewartParticipant@Markiemark Include @Markiemark in your post and this person will
be notified via email.Is this Homework?
What do you mean, cannot take any range, since software collecting/tabulating?
Time or lengths (time between start & complete) are Variable data types, does not matter if software collecting or not.
Examples of Attribute are (Male/female, heads/tails, Y/N), categorical type data.0February 1, 2010 at 5:50 pm #188920
MrMHeadParticipant@MrMHead Include @MrMHead in your post and this person will
be notified via email.Attribute data is more of a nonlinear description – Red, Green, Blue … Yes – No / Pass – Fail … short – medium – long.Do you mean Continuous or Discrete?Time can be treated as either – Depending on your situation, your gauge/measurement.How does your software report the results. In minutes (discrete) or decimals of minutes (continuous)?
How are you going to use the results / measurements?
If you have a requirement to close all calls within 5 min, you can convert your minutes to Attribute data based on + / – 5 min. But you lose a lot of valuable information doing that (imho)0February 1, 2010 at 5:59 pm #188921Hi,
Yes I mean continuous or discrete.
My software reports the reults in Hours:Minutes:Seconds format.
I am taking the difference of Closed Time – Received Time and the result will also be in Hours:Minutes:Seconds format. Hence it is clear in this case that the data is continous.
But can you please provide me answer for my second question i.e, % defective and accuracy.0February 1, 2010 at 6:02 pm #188922
MrMHeadParticipant@MrMHead Include @MrMHead in your post and this person will
be notified via email.Better than that – I can provide you with a whole thread on Percentages:https://www.isixsigma.com/forum/showthread.asp?messageID=157705 . . found via google
0February 1, 2010 at 6:48 pm #188924Variable
0February 1, 2010 at 7:01 pm #188925Hi stan,
Can you please explain the logic?
0February 1, 2010 at 8:22 pm #188931
MBBinWIParticipant@MBBinWI Include @MBBinWI in your post and this person will
be notified via email.yes, Stan, please explain the logic. My instructor won’t accept just a one word answer.
0February 1, 2010 at 8:33 pm #188932What are you trying to do with the data? Because most likely, it won’t really matter.
Stevo
0February 1, 2010 at 9:39 pm #188935The question was – is TAT is attribute or variable?I answered the question correctly. I am not sure logic applies.Create – if you choose to struggle with such simple stuff, you should
probably find another line of work.0February 2, 2010 at 1:16 am #188943hi stan, I thought you have given the answer as variable for % defective or accuracy. Hence I was asking the logic.
0February 2, 2010 at 1:18 am #188945WRONG
0February 2, 2010 at 2:04 am #188949Hi Stevo,
I am understanding the tool selection process. In Minitab15, tool selection process flow it is mentioned that Anova can be used only if the respone is continuous.
Let us say I am interested to do an analysis between process wise accuracy and want to see whether there is any statitical difference between the procsses accuracies or all of them are the same.
Let us say the Y i.e., accuracy is discrete and as processes also are discrete, hence we need to use ChiSquare. But if we conclude that accuracy is continous and as processes are discrete, Anova can be used. But I need to be fundamentally right before I jump in using the tool.
Plz advice.0February 2, 2010 at 2:05 am #188950
Truth TellerMember@TruthTeller Include @TruthTeller in your post and this person will
be notified via email.Apparently none of the brain trusts have directly answered your question regarding the count of defects. Here are some thoughts:1. Ratios of discrete data are still discrete regardless of the number of decimal places you have.
2. Counts of defects follow a Poisson distribution. At around 14 or so, the Poisson distribution can be approximated by the Normal Distribution and the counts can be treated as continuous. It is still discrete but falls in that gray area allowing you to use the assumptions of continuous.
3. The same for proportion defective. As the proportion approaches .5 the binomial distribution can be approximated by the normal and again can be treated as continuous. The underlying data is still discrete but you can treat it as continuous with care.0February 2, 2010 at 2:08 am #188952
Truth TellerMember@TruthTeller Include @TruthTeller in your post and this person will
be notified via email.Before you go jumping into the use of ANOVA you better be sure that you validate the assumptions otherwise your conclusions will be suspect. This may be a challenge is you are using discrete data and treating it as continuous. You may be better off with nonparametric tests depending on your data.
0February 2, 2010 at 2:55 am #188955Hint – there is no concluding that a factor is continuous, it is or it
isn’tCorrect answer for the original question is variable and it is also
continuous0February 2, 2010 at 2:58 am #188956Which question are you answering. The first has nothing to do with
the second.I think this guy is confused.0February 2, 2010 at 3:32 am #188958
Truth TellerMember@TruthTeller Include @TruthTeller in your post and this person will
be notified via email.The question regarding time is clear. Time is continuous unless you do something stupid and convert it to discrete. For example “on time” or “late”. Although the underlying data is still continuous it is now reported out as a binary discrete value. The question I was answering is this one:”One of my process generates defects. I am counting the same and these are attribue data. Using the count of defects I am calculating the percentage defective using the formula (Defects/Produced)*100. Now since the result can also be broken down into decimals, should I call the percentage defective as Variable data?”
0 
AuthorPosts
The forum ‘General’ is closed to new topics and replies.