Bootstrapping in SPSS - Part 2

[@MdAbulKalamAzadBD]

Very well done sir! Thanks for your effort.

[@fabiolalemos772]

How can I know if bootstrap actually fixed the skewness? Is there a way to check out the result?

[@Luciana-wg8gb]

shouldn't we choose the Marsenne Twister option to bootstrap the same sample used on the same test? Otherwise, wouldn't the results be random each time we bootstrap?

[@how2stats]

I haven't seen any research comparing the Mersenne Twister random number generator option versus the SPSS default method (Fishman & Moore, 1981), in the context of bootstrapping. My hunch is that, in the vast majority of cases, it would make very little difference to estimating the confidence intervals for a correlation.

[@Luciana-wg8gb]

@@how2stats Got it! Thank you very much for answering. Nice work, congrats! Hope you keep on doing it

[@ZainabZainab-jr7pb]

Thank you for this video. I have not normally distributed data. So used the spearman correlation. Unfortunatelly I do thr same as you do in this video but I dont get the bootstrapp BCA interval. I only get the correlation coefficient and sig (2 tailed) and the N. What do I do wrong? Thank you in advance

[@how2stats]

My hunch is that you have a version of SPSS that has the "bootstrapping bug". There's a way to solve the problem with syntax. I'll try to make a video on that soon.

[@ZainabZainab-jr7pb]

@@how2stats thank your for replying. I have spss version 25.0. Do i need to instal some tools for bootstrapping? Buy this tools from spss??

[@Nousha0908]

I hope you are well. Thank you for all your videos, they are really helpful. I'm hoping you can help me My data isn't ND and I have some outliers which I don't want to remove. I wanted to ask about Boot Strapping Vs Transforming data. Which is a better technique to use? I'm analysing scores from the Mood and Feelings questionnaire.

[@Mahtabzamani]

Hello sir, Thank you for this video. I don't have normally distributed data and used the same Method you showed for my spearman correlations, however, the n changes when I use bootstrap vs correlations without bootstrap. From n=31 i get n=19 , why is that ? How can I fix it? It seems to me that it excludes the missing data and calculates the BCA CI only available data?

[@how2stats]

Bootstrapping can definitely only work for data available across all variables.

[@budders444]

Can bootstrapping be used when there are a number of outliers?

[@fabiolalemos772]

I think that you need to treat the outliers first, and then, see if there is an abnormal distribution of data. So you run the test with bootstrap.

[@peterkostadinov8367]

Hello sir, it is a great video. One of the least ones on the topic online. I just have a small problem that the bootstraping results are not generated in my output even though I check all the necessary boxes. What might be the reasons? Thank you

[@how2stats]

My hunch is that you're using SPSS version 25 which has a known bug in the bootstrapping. You can fix the problem by installing Fix Pack 1: www-01.ibm.com/support/docview.wss?uid=swg27050502

[@heyrdutta]

Not getting pairwise (rather than listwise deletion) with bootstrap in spss. Any suggestions what I can do? Program notes the request for pairwise but gives listwise deletion of missing values.

[@fabiolalemos772]

Perhaps, you could do a missing values imputation by MICE first.

[@re_mamoste6779]

Merci beaucoup...