+ Reply to Thread
Results 1 to 5 of 5

Thread: Missing bootstrapping estimates

  1. #1
    Points: 1,711, Level: 24
    Level completed: 11%, Points required for next Level: 89

    Location
    Switzerland
    Posts
    1
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Question Missing bootstrapping estimates




    Hi everyone

    I am a newcomer to SEM and I have encountered a problem with AMOS. Given that my data is not multivariate normally distributed and the sample for some of my analyses is small (n=190) I want to use bootstrapping in AMOS. So, I've taken one of the models I tested by means of maximum likelihood estimation as a baseline, have selected a new SPSS file containing raw data where I had eliminated all the cases with missing values on any of the variables in the model as the data file and then ran the analysis with this sample.

    First I tested for multivariate non-normality by looking at the assessment of normality. As expected the data turned out not to be normally distributed. Next, I used the Bollen-Stine Bootstrap to test the overall fit of the model. This worked fine.

    Then I wanted to have a look at the bootstrap parameter estimates. In the Anylysis Properties -> Bootstrap tab I deselected the Bollen-Stine bootstrap and selected perform bootstrat, percentile confidence intervals and Bias-corrected intervals.

    When I view the text output after running this,
    only the non-bootstrapped estimate output is available. Or at least I
    think so, because the output tables do not contain the columns SE ...
    SE/SE ... Mean ... Bias ... SE Bias.

    The directory for the bootstrap output is shown in the second row of
    the left-hand column on the output window. This says Estimates /
    Bootstrap > Estimates > Bootstrap standard error > Bootstrap
    Confidence. The second line ("Estimates") is in grey, suggesting it is
    inactive. I can't click on any of these links, and hence can't see any
    of the bootstrapped estimates.

    I assume that the bootstrapping worked, because the following
    information is displayed under the Summary of Bootstrap link:
    0 bootstrap samples were unused because of a singular covariance matrix.
    2 bootstrap samples were unused because a solution was not found.
    200 usable bootstrap samples were obtained.

    Does anyone have an idea why I can't view the bootstrap estimates?
    Is there something I might have forgotten, like fixing certain parameters? But I don't get any AMOS warnings that something is amiss. Furthermore, the Bollen-Stine bootstrap worked just fine. So why doesn't this work?

    Thank you so much for your help!
    Sybille

  2. #2
    Points: 982, Level: 16
    Level completed: 82%, Points required for next Level: 18

    Posts
    1
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Missing bootstrapping estimates

    Hi Sybille,

    I was wondering if you ever found an answer to your question. I have encountered the exact same problem, and I have been unable to find a solution.

    Thanks!
    Catie

  3. #3
    Points: 430, Level: 8
    Level completed: 60%, Points required for next Level: 20

    Posts
    1
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Missing bootstrapping estimates

    Hi Catie and Sybille.

    I'm dealing with the exact same problem on my side. Any solutions?

  4. #4
    Points: 416, Level: 8
    Level completed: 32%, Points required for next Level: 34

    Posts
    1
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Missing bootstrapping estimates

    Okay, This is really finicky, but you have to click in the estimates portion of the above analysis. After you click on the estimates portion of regular analysis, the bootstrapping estimates field becomes active allowing you to see and click on this portion of the analysis.

  5. #5
    Points: 7, Level: 1
    Level completed: 13%, Points required for next Level: 43

    Posts
    1
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Missing bootstrapping estimates


    I had this problem too. On "Analysis Properties" under "Output" you have to check "Indirect, direct and total Effects" (even if you're just doing a CFA and have no indirect effects). Then, in the text output, when you click on "Estimates-->Scalars-->Regression Weights" the "Estimates/Bootstrap" will be ungreyed and you can access the information you want (i.e. SE, SE/SE...etc).
    Cheers.

+ Reply to Thread

           




Similar Threads

  1. Bootstrapping in SAS
    By alkesta in forum SAS
    Replies: 23
    Last Post: 05-30-2011, 10:13 AM
  2. Replies: 1
    Last Post: 12-11-2010, 02:58 AM
  3. Bootstrapping in SPSS
    By carmen in forum SPSS
    Replies: 0
    Last Post: 11-25-2010, 06:00 PM
  4. bootstrapping
    By karthikeya_ivb in forum Statistics
    Replies: 2
    Last Post: 08-31-2009, 09:02 AM
  5. Replies: 6
    Last Post: 12-02-2008, 10:25 AM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats