Pscyh_7_NOTES_11_17_ - probably no more than 25-30 As the size ofthe sample increases the shape ofthe underlying curve is altered and the

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
Page 3 of5 compute how far from the expected difference of 0 that our data su est. "':","_':. . ',' . y " '3.2 '-';1.1 .f" r. . . ,j"::~+!:~¥. , i1 " , = -v7iii9 .;: .14~3 .. 4.025 You look this t value up in a table and find that it exceeds your standard ofsignificance (p<.05) so that you can say with confidence that the difference between the means deviates from the null hypothesis. In class, we learned how to do this in Excel. t-tests and samples The two groups listed above are supposed to be two independent groups that received different treatments. As such, you would use an independent samples t-test. This is in contrast with a paired- samples test. This kind acknowledges that there is some implicit relationship between the groups (maybe they're students from the same class; or know each other; etc.) As such, you would want to use a paired-samples test that takes into consideration an overlapping relationship between the groups. The size ofthe samples is important for use ofthe t-test. It presumes that that size ofthe sample is
Background image of page 1
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: probably no more than 25-30. As the size ofthe sample increases, the shape ofthe underlying curve is altered and the t-statistic is less ofan appropriate standard. Ifit is larger than that, then you would probably use an F test. . ,Types of Samples for t-test Dependent samples t: , "Same basic format, but the two 'groups' are really the same subjects, measured at two different times Usually pre/post. Sometimes use matched groups instead of same subs Single sample t: 'Instead ofcomparing two groups, comparing data from one group to some number you already know (kind oflike 'simulating' data from second group) Effect Size The effect size is a way oftelling how variation in one variable is related to variation in another variable. It is usually calculated as correlation. Square the correlation and you get "percent variance accounted for". The reason you would wonder about this is that just because you get a significant...
View Full Document

This note was uploaded on 03/05/2010 for the course PSYCH psych 7 taught by Professor Revlin during the Fall '09 term at California State University , Monterey Bay.

Ask a homework question - tutors are online