SPSS - ANOVA - Comparing more than 2 means | Ton Van Voorden | Skillshare

SPSS - ANOVA - Comparing more than 2 means

Ton Van Voorden, MSc.

Play Speed
  • 0.5x
  • 1x (Normal)
  • 1.25x
  • 1.5x
  • 2x
3 Lessons (22m)
    • 1. 12

      7:55
    • 2. 12

      6:09
    • 3. 12

      8:15

About This Class

Wooooowwww! Anova! Exciting!

Compare 4 different countries in people happiness! That's it! Easy peasy! If you don't get it, please let me know and I can always add another videos ;)

Transcripts

1. 12: Oh, this is Stone of Order, Your statistics teacher and this video I'm going to explain to you the UN over otherwise there mattered to compare means within SPS s. What you see over here is the question we're gonna ask now is the ever off Greek people different? Is the efforts off the Netherlands efforts of to Spain or have different as their efforts off Sweden and each one will be otherwise and our dependent variable will be to this country has an effect off wanting, wanting to eat a bigger we're really gonna check whether there is just one difference or more than one. The difference instead off. No differences at all. So it zero is everything is the same. It's one is otherwise which actually means if there's like, see exactly the same and one is different, then you would conclude age one because not all the means are exactly the same. So what we're gonna do, you go to S. P S s. We click on analyze compare means one way and over. So before we did a T tests T tests are to compare two means on over is to compare more into needs of course you can still use on over for also to means over here you put wanting to eat a burger in the deep end and what has influence on it? You call that a factor? In this case, it's the country and noticed the factor is ah, nominal. Maybe order. No available. But normally it's in Norman available. No, this kill. Over here, you click on options and within the options, you click on the descriptive homogeneity off variants and that Vail statist press. Okay, get your output. What you gonna do first over years, he that this kept it stable. So for the Greek people that have a mean off around 5.5 the Dutch, it's quite higher. 7.66 And Spain and Sweden are quiet to get it in the middle. And if you're running your teases, you can. All this means to get it without standard deviations. So, for example, I mean for the great people equals 5.4941 In between brackets, standard deviation equals 1.2879 It's like a p a cell. The other part, you don't really use. That's of homogeneity. Offerings on over this like in an independent T test. There is also to kind off a no over analysis. And there's one for equal variances and one for not able variances. So what is the test off homogeneity of variances? A Levin statistic like we saw before in the T test. If the Y down f you put this a baseless first of three dot 216 equals value 2.31 You see, Levin statistic, you are down f and most importantly, B equals 0.78 The 80 in this test will be all that variances are the same off a Greek Netherlands, Spain and Sweden and AIDS one. Otherwise, in this case, we don't have enough proof to reject age. No. So which is actually good? Because one of the assumptions off this on over table will be that, um oh, variances will be equal. So if it would have had rejected 81 they would have used the blue states of equality of means. But in this case, we can just use the unova table how to interpret the on over table. You. Why don't as first the F f equals you get this tool The degrees of freedom 3216 equals 16 points. 914 Just write down his numbness. Most important is to say with you right down this be a smaller as the open is the only one . Of course, this is lower than often off the open 05 So we would to go up and reject it. Zero. So now we conclude otherwise, not all averages are the same. So we can be quite sure that there's a difference between the efforts of Greek and efforts off Netherlands. But we cannot yet be sure if there is that difference between Spain and Sweden. Separatist has to be conduced for it, and I will explain this post hoc or post on the test contrast to you in later videos. Let's go through the what if scenario is in the example? You see a both here. The p value off this test would not have bean 0.78 But point is, there are There are three, for example. In that case, we would half rejected a zero. If this wouldn't 0.3 it would have rejected 80 And because of that, we were not used on over table because this will not be really to his body. But instead we will do a boost Test off. Equality off means How do you interpret this? Quite the same. It's also leftist. Why don f three and this numbers t 115.984 and then you know I don't be smaller than 0.1 So what you see, if you use the Unova test, other boosters of equality of means in this case is resolved will be exactly the same. But there are cases where the result will be slightly different and then used her boosters off. Equality of mean called the Wells test. Thank you so much for watching this video. 2. 12: Hello. This is Stone from Florida. Your statistics teacher and this video. I'm going to tell you something about post hoc test belonging. Toa Unova Rico from last video First, once a normal and over video. To understand this video competitively, let's recap it shortly over. You see the output off their hypothesis. Test efforts off Grieg is the same. Its efforts off Netherlands is the same as the efforts of Spain is the same Mr Evans off Sweden? It's one otherwise so other differences. We have seen that the means are indeed a bit different. First as we had to do was homogeneous homogeneity off variants. It's not significant. So we can use the UN over output with a significant. So there's four numbers are significantly different. At least two numbers are different, so it can be quite sure that this to are different because they like the most far away from its other. But now, in this city, I'm going to explain to you we want to investigate into this data further and see where are actually differences. Forties, we start. SPS s analysed compare means and we click on one day on over Over here. You still see the your dependent list of wanting to eat a burger, and the factor is still country as possess. We remember these things for you, and we click now on post hoc. You see a lot of options to play Gone. Actually, all these things are quite the same. Um, there's a lot of different people were invited methods to compare like the differences. The most mostly used is Tookie Orban phony, this one for equal fans assumed and equal Vance another shooting games. Howell, I'm going to explain to you to Turkey and the games Hubble, with the output from, for example, born phony, you can read it and exactly the same way. So just practice with it yourself, if you need to. For your studies. Otherwise took is one of the best matters. I press continue, I press okay and I get my output. What you see in your output will be like again all the on over things. Since the variances ah assumed to be equal. We use that tricky test. If events were not assumed to be equal used games, how will test and I made a table with both. So then you see, I can explain both to you. What you see over here is like just a comparison is great, different from Netherlands? Yes, because it's significant. It's great. Different from Spain? Yes, and create different from Sweden. Yes, it is no different from Greg. Yes, but look, we already got it here. Captured is now on different from Spain. Also, because all this numbers are below 0.5 significance different. But this number Netherlands and Sweden todo not significantly different. So between Netherlands and Sweden there's no effect. But there's a effect between a lot off the others. I would just press for you significantly different results. Sweden. It's quite in the middle because it's only different from Greek notes from gallons, not from Spain. If the livin tas was significant, so like, this number was below 0.5 In that case, you would not have used these numbers, which would have used that games Howell numbers instead. Ah, you know, least the tiki Eight as dig is a nice table over here and what you can see for often off 0.5 you can see that Greek has a separate subset on what it means is that a Greek Because here is not anything, anything, anything at all. No numbers. Greek is different. Then Spain, Sweden and Netherlands. You can see it back here. A great is different form. Netherlands, Spain and Sweden. If you look to Spain, then Spain is different from Greek and different from Netherlands, but not different from Sweden, because he is also a number. These numbers, actually Adam means so the mean off crickets 5.49 the mean of civilians, 6.79 in the main of Sweden and 6.91 are quite close together. So not significantly different. You see the same as Sweden for Netherlands. So if you look to Sweden, for example, you can see like here, Sweden is the same Spain that's reading is also the same as Netherlands. Like they're not a significant there's not. A significant difference is over. But Sweden is still different form, I think. Thank you so much for what sink this video 3. 12: Hello. There's a storm of order, your statistics teacher and in the last videos explained to you how to do it on over to compare multiple means and afterwards in the video afterwards, explain to you how to do a post hoc test. If the Unova significant, where our differences and this video I'm going to tell you how to do a contrast in this case, you are not really interested. If there's like some differences between means bird in this case, your interest in a specific difference. For example, you're curious if the efforts and wanting to eat a burger will be different. But if there's a difference between Dutch together with three, these people vs Spain. So you have, like some people from the Netherlands, some people from Sweden and you see them as the same. They're like from the northern of Europe and Spain is a sound of you. So you're curious. Is there a difference between the people from the North and people from the South? How to do It is a lot of steps, and I work out already some steps before you because it's needs a lot off mathematical calculation. SPS is not that good program to do this, but you can do it within SPS s and a mathematical calculation that difficult. But it's a bit advanced. So first step will be analyze. Compare means one way and over and over here. Is he putting again wanting to eat the country You press now not opposed hope but contests . And over here you get like a strange menu and you have to put in coefficients already. Put the golf sense and for you 0.57 miners 10.43 I'm going to explain in second why I put exactly this numbers in here. But just to show you how this works, for example can click here, remove put 0.43 And actually And if you put in for coffee sensi of four different averages , Um, because as four different counties in my day, That said it's putting for now, but you press continue and how to get to this numbers. So what's important to know first is what we want to compare. You want to compare the efforts off Netherlands and Sweden, First of Spain. So I put like dot, dot, dot, immune, Alan, splits placed under the muse reading equals dotted off New Spain. Spain has 100% because it's just one group. So all the people from Spain they called us 100%. But a Netherlands and Sweden. There is not equal amount of people in my data, so I have to make use of percentage off my data and how I will do that. First. I need to know how many people there are in total, like Greek has 46 Netherlands 66 paying 59 sweet and 49 then have to calculate how many are in Netherlands and Sweden together, that is 115. So the proportion off the people from the Netherlands 66 divided by 150 the proportion of people from Sweden are 49 divided about 115. So therefore Netherlands, the efforts of the Netherlands will count a bit happier than the efforts of Sweden. And this way we I just for them Spain. I just put in 59 or 59 with, of course, that is 100%. What do you see more? I put it very labels and Netherlands is in my FDA, says data said. The to Sweden is the four. Spain is the three. So I re gold Netherlands into a to Sweden into a four Spain into a three. Then I put like this divisions and then I calculate him on my calculator. Escape says only allows you to put in to two numbers behind behind the dot So therefore, I put it here in in a two number digit say upon 57 Mewtwo 60 65 about one of the 15 years 0.57 bliss is open for three equals one, and now a really important step is to put the mu Spain to the left side. So if you it's like a mathematical thing, if you put something toe other side of the equation on the right side, I do minus one new three to evaluate on the right side. Then on the left side, you also have to do minus one music and then the right side becomes zero because there's nothing left. If ice obstruct everything on the left side, the minus one music will be added. Furthermore, what's important is the Greek people are also my data set, but they're not joining in this calculation. So therefore I put zero mute one. So, like the Greek are notes joining. And now here I got my contrast. Yeah, I got my contest. Disney My have to feeling into SP Assess. What a difficult work. I think in one of the future versions off SBS s this should be this could be done automatically. But till now, till version 24 this is not possible. You have to do, like in this step for way. So what I do over your contest? I put it 0.57 minus 10.43 And the order is really important. So let's just do all the steps putting in. Let's continue press, Okay, you get your output. And if you look to your output, then you see like an extra contest test. You see consciousness, assume equal fans and does not assume equal variances. It's the same as the two sample T test compared to means. Actually, it's just comparing to means homogeneity. Orban says it's not significant. So we assume equal variances. So my results will be t 260 because of freedom equals 2.157 and my fever you will be point there three to So, actually there's a difference between Nelson Sweden vs speed in the future. Please ask me. Put like a note like I want to practice a bit more with contest. If you want to practice a bit more, I will make so exercise. If you would please ask me for it. Because otherwise that first gonna continue making other videos and practice a bit with this yourself. Important thing to know. Thank you very much for watching this video.