Three Different t tests using SPSS with Dr Ami Gates

hey there welcome to learning statistics with dr. Amy gates today our video is going to focus on examples of the three types of t-tests that you can use and our examples are going to be done in SPSS our first type of t-test that we're going to look at is called an independent sample means t-test in this particular case we're going to compare two sample means to see if they are significantly different in our example that we're going to see we're actually going to look at the mean height of the mail people in our sample and we're going to see if that is significantly different than the mean height of the female people in our sample so our null hypothesis is that our mean heights are not significantly different and our alternative hypothesis is that our mean heights between males and females are significantly different this is going to be a two-tailed test and we're going to use an alpha of 0.05 in this particular example we're going to be grouping by gender and we're going to be comparing two independent sample means let's see how this looks in SPSS in this particular data set that I'm going to be using for all these examples we have all kinds of different data listed we have the person's ID their gender their age their height and so on all the way till the end of our data set where we even have pretest and post-test scores for each of these people now in this particular first example we're going to be looking at the height of our people and we can even see that our data set might have some errors in it but that is a possibility so we're going to be looking at our height and we're going to be grouping by gender and we're going to be wondering is there a significant difference in height between our genders in order to test this we're going to click on analyze and then we're looking to compare means and because we have two samples one male and one female that we're comparing and these samples are independent we're going to choose the independent samples t-test once we select this we need to tell SPSS what variable we're going to be grouping by in this case I want a group using gender in other words I want one of my groups to be my males and I want the other group to be my females I have to define those groups for SPSS and I'm going to scooch this over a little too so we can actually see how they're named all datasets are different in this data set the words male and female are actually used to describe male and female but in some data sets they'll use a1 and a2 or any other categorical naming system so it's always good to check out your data set before starting to do analysis on it now when I define my groups I know that my group 1 is actually called male with a capital M and my group 2 is going to be my females the capital F so I'm going to click continue and this tells SPSS that I want to group by gender I want my first group to be male my second group to be female just the way it reads here on my data set and one of my testing testing height so I'm going to move that over as well this is all that SPSS needs to know I'll click OK and then it will bring up the results for this particular test as we expected our males are taller than our females but the question is is the difference statistically significant and our results are listed over here when we're comparing sample means one of the first things we have to do is determine if we're going to assume equal variances or if we're not going to assume equal variances SPSS performs an F test for us and gives us the p value of that F test let's look at the results on the PowerPoint slides we can see them better and determine what our results are for this particular example so in this case our first step was to determine if we're going to use equal variances whenever the p value for the F test is greater than our alpha value of 0.05 we can use equal variances and so I've decided to use the top row of information so my t-test value is 1.5 three four and my p-value for this t-test is 0.12 six whenever my p-value which is two-tailed in spss is larger than 0.05 we cannot reject the null because we did not fall into the rejection region so that's how we find the results for each of our SPSS tests all right let's look at our second example in our second example we're actually looking at what's called a one sample t-test in this case we're going to compare a single sample mean from our data to a known mean of a population so we're kind of wondering does our sample is it different from the population mean that we know or is it the same as the population mean that we know so this is called a single sample t-test because we're only using one sample and we're comparing it to a known population mean so our null hypothesis is that the mean of the population that we know is actually the same as are not significantly different than the mean of our sample the alternative is that the mean of the population and the mean of our sample are actually significantly different again we're going to use a two-tailed test and alpha is going to be 0.05 now in this example that I'm going to show you an SPSS we're actually given first that our population mean height of males and females overall is 62 inches so we just went all over the world and grabbed all the heights of adult males and females our population is known to be 62 inches so the question is given that we know our population mean is the sample that I've taken the same or different and that's what we're going to look at here in the next example let's see how that works in SPSS so we're back to the same data set and in this case we want to just compare all of the heights of all of our males and females which is our sample to that known population me of 62 inches when we click analyze we're going to again compare means but this time we're going to use the one sample t-test because we just have one sample that we want to compare to a known population mean so let's choose that notice that it asks us for the test value the test value is the known population value that we want to compare to in our case that 62 inches because that's what we were given in the problem the test variable from our sample or our data set is the height that's what we're actually looking at here to compare so we're wondering is our sample group of heights significantly different than our population height that we know once we tell SPSS these two values we click OK and it runs our test in this case we don't have to determine whether our samples come from the same variance or have the same variance because we have just the one sample that we're comparing to our population mean so here's our results and if you'll notice our sample mean is 61 point 95 that's really close to 62 so we can almost guess that there's going to be no significant difference between our sample mean and the population sample mean let's look at the results these results listed by SPSS give us the t-test value which is actually quite small it's very close to zero and it gives us the p-value which is quite large our p-value of 0.975 is well above our alpha value so we most certainly cannot reject the null hypothesis and that makes sense if we think about it because the mean of our sample is almost identical to our population mean so it certainly wouldn't be significantly different and that's why we're not rejecting the null here all right let's look at our last example in our last example we're looking at paired sample t-test paired samples are samples that actually come from the same individual but have more than one collected value for example we might have pretest and post-test scores so each individual in our sample will take a pretest and then they'll later take a post-test so we'll have a pair of scores for each person in our sample we could do the same thing with blood tests we could have everyone take a blood test and then they could do maybe an exercise program to lower their cholesterol and then take another blood test and again those would be paired scores in this case we're trying to figure out and that's a typo that should say H a we're trying to figure out if the sample mean of our pretest group is not really different from the sample mean of our post-test and our research hypothesis of course that should be an H a is that the sample mean of our pretest is actually significantly different than the sample mean of our post-test again two-tailed and alpha is 0.05 let's see what this looks like in SPSS when we're working with paired data okay so in this case we're actually looking at pretest and post-test scores so each one of our people or our students in this case took a pretest and then later took a post-test and then the second person same thing they took a pretest later took a post-test so these data elements are paired because these two are for one person these two are for one person and so on but we still want to know overall did people do better on the post-test than they did on the pretest for example so let's analyze again we want to compare the means but in this case we want to use paired samples because we know that our samples are paired up so we'll click this option and then it's really straightforward we have to just tell SPSS what are the two variables we're pairing up what are we looking at in this case we want to look at the pretest that's our first variable in the pair and our second variable in the pair is the post-test we're only looking at one pair so we don't have to go to number two or do anything here these are the two variables we're looking at and that's it that's all SPSS needs to know we click ok and again it generates our results for us we can look at the means quickly and see that the pretest is a 60 1.47 the post-test is a 100 point 65 those numbers seem pretty different we're going to guess that this is going to be a significant difference but let's look at the results here our results tell us that our T value is is really hugely negative if you imagine the rejection region this T value is all the way over on that left tail it's very much rejected and look at the p value it's zero zero is definitely less than 0.05 so we can absolutely reject the null and we can determine that our pretest and post-test scores were significantly different so this was a quick example of the three different T tests that you can run how to run them in SPSS and how to interpret the results thanks for joining me

Channel: ProfessorAmiGates

Views: 90,967

Rating: 4.9030023 out of 5

Keywords: hypothesis, testing, t-test, means, sample, SPSS, hypothesis testing, comparing means, paired data, dependent

Id: dkBQHmWjNjs

Length: 12min 27sec (747 seconds)

Published: Wed May 30 2012