Bootstrapping Problems

[FBreit]

Hi,

I have problems running Bootstrapping. I've tried to run 200 samples, but it always stops at about 180 or 190, without any error message. The Bootstrapping window just disappears.

I don't know if this is a harware, software or a model problem. I've searched the forum for "Bootstrapping" and it seems like other people had similar problems before. But I couldn't find a solution.

My model consists of an exogenous fourth order LV (type IV, all formative) with a total of 66 indicators and one formative endogenous LV (six indicators). I operationalized the fourth order LV using the repeated indicators approach suggested by Wold (1982). Might this be the problem? In the forum I've read that Bootstrapping doen't work, if two variables are perfectly correlated. Is this also a problem in the repreated indicators approach, where I'm using the same variables several times (on first, second, third and fourth order LVs)? However, as I said before, no error message is shown.

I set the number of cases to 147, which equals my original sample size. Might the problem exist due to a too small sample size relative to the number of indicators? Again, in this case an error message should appear.

My laptop computer has 2 GB of RAM, so I think it souldn't be be problem of insufficient memory?

It would be great, if you could help me. Thank you!

Frank

[SvenjaSchmidt]

I'm encountering the exact same problems, bootstrapping aborts after 14 of 200 samples without any error message.
I'd very much appreciate any hints or tips to get the thing to run.
Thanks!

[elianacarraca]

It also happens to me.

I'm working with a second-order variable and i'm following Wold's recommendation as well. I'm starting to think that repeating the indicators in the second order variable might be the explanation, since I just bought a new laptop with 4GB of ram, and I still can't run the bootstraping procedure.

I have a sample of 160 and 69 indicators.

I really would appreciate if one of the experts in the algorithm could clarify this matter.

Thank you,

Eliana Carraça

[sudzina]

1) I am almost always able to run 5 000 iterations but never 10 000 iterations; the model has only 3 latent variables each consisting of only 2 independent variables

2) in one case, I got a message about singular matrix or something like that and the process stopped

3) values for standard deviation (STDEV) and standard error (STERR) are equal all the time; so either the or the other value is wrong; in case the latter is wrong, it affects t-statistics and this would be serious