iSixSigma

Sampling Technique

Six Sigma – iSixSigma › Forums › Old Forums › General › Sampling Technique

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
    Posts
  • May 19, 2005 at 12:07 am #39402

    Dewi
    ★ 10 Years ★
    Participant

    Hello!
    I am planning to do a survey to measure employee satisfaction with a specific IT application. I am thinking of inviting all employees in 3 companies to do the survey. However, one of my colleagues advised me that if I do that, it will create selectivity bias. He thought that only employees who use the application will participate in the survey and since the survey will be launched on the web, he also thought that it is unfair for employees who seldom access the net.
    My target is employees who use the application and since the application could be accessed from office or through the web, I do not think that it will be a bias targetting all employees to participate.
    Would you be able to give me advice on this matter?
    Thank you.
     

    0
    May 19, 2005 at 1:47 am #119701

    Issa
    ★ 10 Years ★
    Participant

    Whoever told you to proceed with a stratified sampling is right. If the application is of no use to some of the employees even if they can access it, including them in the survey will result in a biased (or even erroneous) result.

    0
    May 19, 2005 at 2:00 am #119703

    Dewi
    ★ 10 Years ★
    Participant

    Thanks for your feedback. However, the application is available to be used for all employees. Unfortunately, it is not mandatory for employees to use it as they have alternative ways to perform their work. Would you be able to give me some advice on which sampling method is best for my case? Thank you.

    0
    May 19, 2005 at 3:53 am #119706

    Issa
    ★ 10 Years ★
    Participant

    I would still proceed with a stratified sampling. I would conduct two surveys one for the heavy users and one for the others. Or I conduct one survey with 80% (or more or less) of the respondants comming from the group of the heavy users and a smaller percentage comming from the other group. The coefficient (percentage) that you affect to the groups should not be arbitrary, it should reflect the amount of time or the number of people in each group that use the application.

    0
    June 7, 2005 at 1:42 am #120790

    Jonathon Andell
    ★ 10 Years ★
    Participant

    Start with this question: why do you want to conduct the survey? You may decide to ask “Why?” after each response, until the final answer stops changing.That exercise will tell you whether you need to survey only current users, or whether to survey potential users. As for whether you are creating selectivity bias, that depends on many more factors than we have covered so far.

    0
  • Author
    Posts
Viewing 5 posts - 1 through 5 (of 5 total)

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