Why Only Means in the Six Sigma Salary Survey?

Six Sigma – iSixSigma Forums Old Forums General Why Only Means in the Six Sigma Salary Survey?

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
  • #35920


    As I´m a MBB who contributed to this survey. I was actually quite excited to see the results. Reading the findings I got the impression, that the data contained much more information than revealed.
    Isn’t Six Sigma about reducing variation? Don’t we all tell the others, that only looking at mean values is worthless unless shape and spread are taken into account as well? Maybe I´m to harsh here, I really appreciate the initiative. But a data analysis of this kind, which is viewed by many stat professionaly around the globe should serve at least the very basic principles of data mining: A) Is the data normally distributed? If yes, indicate mean & standard deviation. B) Is it not normal, nor following any other major distribution? Then use non parametrics metrics as median and quartiles or percentiles. This procedure is used in many other surveys on salary!C) Confidence Intervals are always a good thing to show.
    I hope that Matthew LeVeque and his team will take this serious and provide further analysis!
    Again, good start, but only a medium finish!
    Awaiting controverse discussions,best Regards,Oli


    Shereen A. Mosallam

    Very interesting comment!
    Graphical presentation will be excellent too.



    No controversial discussion on my part here Oli.  I’m a BB and I contributed to the survey as well and I agree 100% with your comments.  Here’s what I’d like to see apart from just the mean:
    MIN, Q1, Median, Q3, MAX
    This should be done for each industry, for each job description and for each education level.
    It’s absurd that they didn’t already include this info in the original article.



    I agree with all of you. At the same time, I wanted to share with you that when I was offered to dive into Sig Sigma, I had no place to look for the salary information. This aritcle is a great one given there is nothing at least I could find out there. Hats off to everyone participated and put this article together. Like any other process improvement project we should let these guys go through few more cycles where the voice of the customer is heard.


    Mike Allen

    Box and Whisker plots are a good way to graphically represent the stats which Anonymous mentions, and also help show center, shape and spread (must indicate sample size, though).
    I agree with the feedback given so far – this was a great initiative, one which undoubtedly required much time and effort, but mean data only shows part of the picture, and doesn’t indicate center, shape or spread.  Typically, managers are looking for one number to tell them everything, and it’s our job as BBs and MBBs to educate them that this isn’t feasible.  (rough analogy: the story of the 5 blind men and the elephant…)

Viewing 5 posts - 1 through 5 (of 5 total)

The forum ‘General’ is closed to new topics and replies.