I was reading some discussions about Bonferroni correction on the web site below: http://www.researchgate.net/post/Do_we_need_Bonferroni_corrections_for_a_series_of_T-tests_that_are_conducted_on_different_tasks2
I have some questions about Bonferroni correction and was wondering if anyone of you could kindly help me with them as an expert of Statistics. I really appreciate it! I found a succinct summary of Bonferroni correction here: http://www.fon.hum.uva.nl/praat/manual/Bonferroni_correction.html
Here is my situation: I am currently writing my dissertation in which I used a survey to collect data from two groups of participants and used SPSS to analyze the data. In order to answer Research Question Two “How do the competent and less competent university students differ in terms of their degree of metacognitive knowledge about English writing?”, two independent samples T-tests (or independent t-test, for short) were conducted on the two groups of participants (group A and group B). One was an independent T-test performed on the two groups regarding mean scores of the three sub-scales (person, task, and strategy variables), and the other was an independent T-test conducted on the two groups to compare them item by item (there are 50 items in the survey). My questions are as follows: Question 1. Since there are only two groups of participants involved in this study, is it necessary to perform Bonferroni correction to reduce Type I error ? I thought Bonferroni correction is meant for 3 groups or more, according to the video clip below: https://www.youtube.com/watch?v=BK7Ay49jM_8
Question 2. If an independent T-test is conducted on the two groups to compare the means of the two groups item by item with 50 items in the survey, is that considered one T-test or 50 T-test? In SPSS, I only needed to run one T-test to compare the mean of the two groups item by item for the 50 items.
Thank you so much! Best regards, Mike
https://www.youtube.com/watch?v=BK7Ay49jM_8
http://www.fon.hum.uva.nl/praat/manual/Bonferroni_correction.html
As a rule of thumb, in basic inferencial statistics, Bonferonni's correction (or Bonf. test) is gemneraly used for 3 or more officially independant and normally distributed populations (sub-samples in relaity, created on the basis of the same measurement. Practically, under the (mostly always violated) assumption of normality of the distributions such as what is occuring when utilizing one of the kind test of the ANOVA's family, the question is a bit more complex. There is a good "fast" answer for the logics of your question: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2548561/pdf/bmj00576-0038.pdf
Good luck with your dissertation. Sincerly, François L.
As a rule of thumb, in basic inferencial statistics, Bonferonni's correction (or Bonf. test) is gemneraly used for 3 or more officially independant and normally distributed populations (sub-samples in relaity, created on the basis of the same measurement. Practically, under the (mostly always violated) assumption of normality of the distributions such as what is occuring when utilizing one of the kind test of the ANOVA's family, the question is a bit more complex. There is a good "fast" answer for the logics of your question: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2548561/pdf/bmj00576-0038.pdf
Good luck with your dissertation. Sincerly, François L.
Thanks for the extra James. Next step woul be to verify the item structures using IRT...
http://mres.gmu.edu/readings/PSYC557/Drasgow_Hulin_Item_Response_Theory.pdf
http://harvey.psyc.vt.edu/Documents/TCP_IRT98.pdf
Or the basics... Hambelton (1991) or more recent http://books.google.ca/books?hl=fr&lr=&id=VqJ1yhgP1sMC&oi=fnd&pg=PA148&dq=Item+response+theory&ots=PkmoY12QLg&sig=uU5_PTTXSJyLghmF8X755fPMV9s
If we have two Groups. each group with 3 meaurements. How we can use post hoc test?
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis