iSixSigma

empowerment survey

Six Sigma – iSixSigma Forums Old Forums General empowerment survey

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
    Posts
  • #37798

    Fontanilla
    Participant

    Hello Forum
    We have conducted an empowerment survey based on 10 segments  further divided into 49 questions. The scale used is 1-5 with 1 being ‘strongly disagree’, 4 being ‘strongly agree’ and 5 being ‘cannot say’. How do we arrive at an analysis other than the percentage of people on each level of the scale.

    0
    #112487

    John Melby
    Participant

    Before you think about the analysis, you might consider the number of questions in your survey. In many working environments, as all know, completing surveys is usually a lower prioity work task. Any surveys over 10 simple questions have an increasing risk of being deleted before completed. Even those that are, when the survey is too long, it become less accurate as respondants tire in finishing it. Also, you may consider a scale that eliminates the nuetral, forcing the choice to one side or the other. Or, simply structure it to; Yes, No. A shorter survey will increase the number of responses received which usually increases the overall accuracy of the data. Lastly, no survey is as accurate as a statistically valid sampling live interview survey process. Electronic volunteer surveys only measure those that are willing to fill out the survey…a situation that can provide very inaccurate data. My suggestion is you spend more time doing short, live interviews to a small, diverse group. Comments welcome.

    0
    #112496

    Deanb
    Participant

    Dan,
    If your data has relatively normal distributions you can use correltation, regression modelling and anovas to identify interesting and potentially useful relationships. To get decent normal distributions you have to do many things right the first time (good question and survey design, sufficient sample size, etc) which is why it is important to design the survey to suit the statistics you intend to use (rule #1 in statistical studies), and not merely matching the statistical tools to the data after the fact. I find averaging averages of sub-factors (X-bar) to form main categories improves the development of normal distributions, however comingling needs to be done with care to assure you are truly measuring what the main categories represent.
    Stepping back a bit, it is important to understand where surveys stand in the heirarchy of knowledge. They are quite weak on the validity and reliability scale. At best they alert you to a possible relationship in nature which motivates you to study something in greater depth. They are a form of open dialog and listening, and a weak form at that. You can ID critical “relationships” with more validity and reliability using direct group methods such as FMEAs or Focus Groups, which is what I usually progress to after doing a survey.
    Good Luck.

    0
    #112552

    sathish chandran
    Member

    Hi,
    You should have used the scale of 1 to 7 for the survey (where 1 means strongly disagree and 7 being strongly agree. Neutral being 4).
    any way, having done the analysis. Use radar chart to plot the ratings under different cateories you had chosen to survey.
    Any rating in the neutral zone is a -ve score, you need to work towards converting them to +ve ones.
    other points to work would be;
    1) No.of people who have given you negative ratings(or not satisfactory ratings) – these can be treated as defects and work towards reduction of these defects.
    2) meet all those who have given negative ratings find out why ?
    3) meet some of those who are happy with your service and find out why?
     
    regards
    sathish

    0
    #112841

    ALEK DE
    Participant

    Hi Dan ,
    Lot of analysis can be done , but it all depends on what questions we have asked . First , 49 questions sounds high . One suggestion , have you asked or ensured ” what are you going to do with answer ” for each queations you have asked ? Have you set the objective of survey before actually designing the questions ?
    To waht extent you got the anwers ? Poor response may turn fatal if you want to decide something based on data you got. Generally in a survey sample size is selected based on some calculations. Hence response is an important aspect in the survey.
    For example , as you have indicated , your basic asumption is 10 segments. Now with the answers you got , you should validate that first.
    If you can share the questions you wanted to answer through this survey , possibly we can have further discussion on this.
    Thanks
    Alek
     

    0
Viewing 5 posts - 1 through 5 (of 5 total)

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