Paired Sample t-test

okay this is the lesson on the paired sample t-test otherwise known as the related sample or the repeated measures t-test we seem to have our data open we have four columns one called participants sex pretest and post-test there seem to be 10 different participants if we click on variable view we can see that the variable named sex refers to the participants X if it click on values we can see that there is a 1 means man and 2 means woman the pretest variable is some score on a test before individuals are trained or whatever it is they're being tested on or the post-test is the score on a test after being trained we can see the numbers look a little different let's say we want to know whether or not the training is effective you have this new training method and you want to know does training actually make people better at taking a test this might be the SATs where you take an SAT class and then you take an SAT prep class or for those of you considering graduate school or loss for med school you take the GREs the MCATs the the bar once you're in law school and then take a prep class and and we want to know if the prep class is actually effective at making you better in order to do this we're going to want to use a paired sample t-test our hypothesis is that the test does cause a difference our null is that there is no difference that effectively the average here is exactly the same as the average here so what we're going to do is we're going to go to analyze we're going to go to compare means and we're going to go down to paired sample t-tests now we've seen here before the independent sample t-test the one sample t-test and the one way ANOVA we're going to do the paired sample t-test and we're going to choose our two variables saying what pair do you want to compare what is your first variable what is your second variable well our first variable is the score on the test and our second variable is a score after the test and all we're going to do now is click OK now we're going to get a couple different bars here we're going to see our paired sample statistics these are the means the N the standard deviation the standard error of the mean for our before and after if we look at the test average before training to seventy nine point six if we look at it after it's an eighty two point four that doesn't look like a large difference but once again we need to determine whether or not it is a significant difference to be very honest I think most of us would like to be able to increase our grades on tests about three percent this box here is the parent sample correlations it's looking for the correlation between your two levels of variables that before and after generally when we're doing pair table t-test we're not going to pay much attention to this box it's not particularly useful we could in fact just get rid of it down here is the paired sample t-test in information and what we're going to do is we're going to go to pair one which is the score on the test before training the score in the test after training had we included other variables with they're not in this data set there would be more levels of pairs down here we're going to go all the way over to the very end this section right here and what we're going to be looking for is our T value our degrees of freedom and our significance value which once again is our p value is our p value so if our p value or our significance value is less than point zero five point zero five which it is we would say that the groups do in fact differ our null hypothesis which was that they don't differ would be rejected in this case we would reject the null hypothesis it looks as though with a t with nine degrees of freedom equally negative two point six three five with a p-value of 0.02 seven the groups do in fact differ and we can go back up here to the paired sample statistics to say that in fact the test is effective people who took the test did better after taking the test than they did prior to taking the test this is how you conduct a paired sample t-test you

Channel: bernstmj

Views: 371,079

Rating: 4.8385501 out of 5

Keywords: Statistics, Psychology, Academia, Math

Id: Qt-TlsY_b3k

Length: 4min 27sec (267 seconds)

Published: Tue Jun 21 2011