0@Loc: Person/b14.cha1@PID: 11312/t-00015816-12@Begin3@Languages: eng4@Participants: TS Student, SS Student5@ID: eng|Person|TS|||||Student|||6@ID: eng|Person|SS|||||Student|||7@Media: b14, video8@Comment: B-14 STATISTICS9*TS: xxx there are a lot of confusing things in here. ▶10*TS: I'm sure you have questions, make sense. ▶11*TS: you can just start where the book starts and just move through12orderly. ▶13*TS: xxx it starts on one hundred and thirty four, uh, one hundred and14thirty five (.) one hundred and thirty five it starts. ▶15*TS: you can start by defining inferential statistics for me. ▶16*SS: ok. ▶17*SS: inferential statistics is taking, uh, two means from a couple of,18two means from the same sample and then do an operation on them. ▶19*TS: so, our focus in doing that is, when you have a sample, um,20hopefully our sample is going to be indicative of our population. ▶21*SS: yeah. ▶22*SS: that's right. ▶23*TS: but, sometimes we have a sample that not necessarily that24indicative. ▶25*TS: that's what these statistics allow you to do. ▶26*TS: to apply information about these samples to your larger population.27▶28*TS: um, equivalency between groups. ▶29*TS: there's two ways that, to maintain equivalency in groups. ▶30*SS: uh, two ways? ▶31*TS: uh huh. ▶32*SS: one is, uh, random. ▶33*TS: right. ▶34*TS: randomization. ▶35*TS: xxx. ▶36*SS: and number of subjects. ▶37*TS: you just, um, control for all variables. ▶38*TS: which is pretty much impossible. ▶39*TS: that's why randomization is the preferred method .40*TS: and if you do have these equivalent groups, you're going to assume41that there's, they're going to be the same within a certain margin42of error on any particular treatment or any particular test or43observation. ▶44*TS: and therefore, if one group differs from another and you can say45it's due to the treatment and not due to the content of the group. ▶46*SS: uh huh. ▶47*TS: ok? ▶48*TS: let's move on to the null hypothesis. ▶49*TS: do you want to explain that to me? ▶50*TS: the null hypothesis? ▶51*SS: it's, uh, +... ▶52*SS: I guess it's, uh, you're setting a hypothesis and you either reject53it or accept it and you have to set a, a statement whether, uh, you54reject it or not. ▶55*SS: I really can't explain it that much. ▶56*TS: you have, uh, two hypotheses. ▶57@Comment: pic002 (b14/image002.jpg)58*TS: uh, when you set up an experiment, you basically have what's called59the null hypothesis. ▶60*TS: and I've seen it written like that. ▶61*TS: and the experimental hypothesis. ▶62*TS: and, what the null hypothesis almost always states is that there63is, uh, no difference between the groups. ▶64%act: draws Ho and H1 on the board65*TS: the experimental hypothesis is there is a difference. ▶66*TS: and sometimes you'll you'll want to say, if it's bigger or smaller.67▶68*TS: so, let's say, uh, we're doing an experiment on Head Start. ▶69*TS: are you familiar with that program? ▶70*SS: xxx. ▶71*TS: ok, well, you don't have to be that familiar with it. ▶72*TS: um, let's say our experimental hypothesis is: kids who start in73Head Start, um, end up doing better in elementary school than kids74who, who never started. ▶75*TS: so, that was our experimental hypothesis. ▶76*TS: kids given Head Start will do better than kids not given Head77Start. ▶78*TS: our null hypothesis then would be that it didn't, they didn't. ▶79*TS: it would just be that equal groups: the ones that did have it and80the ones that didn't have it are equal. ▶81*TS: understand? ▶82*TS: there's no difference because of Head Start. ▶83*TS: ok, um, so this is the difference, is due to the independent84variable. ▶85*TS: the variable that we're manipulating. ▶86*TS: and, uh, this doesn't mean that the groups are identically the same87necessarily. ▶88*TS: or they could be identical &=mumbles means. ▶89*TS: but it does mean that, uh, the difference between the groups is due90only to chance. ▶91*SS: ok. ▶92*TS: so, it deviates, it deviates from chance. ▶93*TS: and that's &=mumbles difference there. ▶94*TS: um, the reason we use these particular statements is this is95relatively easy to validate or, or throw out. ▶96*TS: you can't really do it with deal with this one97[= pointing to null hypothesis on the board]. ▶98*TS: that's why we either reject or accept this one. ▶99*TS: and if you reject this, they use the term, “accept the experimental100hypothesis” and I don't necessarily agree with that. ▶101*TS: um, because you can reject this [= pointing to the board] and this102might not necessarily be true because say that, the, they do worse.103▶104*SS: uh huh. ▶105*TS: the kids who are given Head Start do worse which means that106&=mumbles kids &=mumbles or something. ▶107*TS: so, we even have, this is obviously, you reject this108[= points to H1]. ▶109*TS: this is still not correct. ▶110*TS: so, all we can say is we either, um, accept or reject the null111hypothesis. ▶112*TS: and by doing that we can never prove the experimental hypothesis. ▶113*TS: it's kind of unfortunate, but that's the purpose of the replication114and other kind of research in the same area is to, is to get around115that. ▶116*SS: why can't you get that? ▶117*TS: um, (.) because it's not, it's not, it doesn't lend itself that118there is a difference, that there is a difference doesn't lend119itself to statistical tests very well. ▶120*TS: because if there is no difference it's easier to &=mumbles. ▶121*SS: yeah. ▶122*SS: ok. ▶123*TS: and you reject this, you reject the null hypothesis, there is a low124probability that differences obtained are due to to random error. ▶125*SS: &=nods . ▶126*TS: is that, I guess that's the most official way you can say it. ▶127*TS: differences, um, that the differences are due to random error is a128very low probability. ▶129*SS: so, you reject it? ▶130*TS: uh huh. ▶131*TS: (.) ok. ▶132*TS: do you have any questions about either the null or experimental133hypothesis? ▶134*SS: no, no, no. ▶135*TS: ok. ▶136@Comment: pic004 (b14/image004.jpg)137*TS: that's going to be the basis of most of what we talk about today. ▶138*TS: um, some of what we talk about today anyway. ▶139*TS: ok, probability and sampling distributions. ▶140*TS: (.) um (.) this is more attaching, putting this particular thing141[= points to board] into concrete terms. ▶142*TS: so, this is an example of a sampling distribution. ▶143*TS: it'll be our sampling distribution. ▶144*TS: I'll pretend it's bell shaped. ▶145%act: draws a bell shaped curve146*TS: ok, um, so, we administer an IQ test. ▶147*TS: let's just say, we'll go back to our Head Start. ▶148*TS: and this is our range of results, for this particular test149[= pointing to the bell- shaped curve]. ▶150*TS: ok. ▶151*TS: we need to set what is called your probability level which I've152always been told is called your alpha level. ▶153*TS: you've had statistics? ▶154*SS: yeah. ▶155*TS: ok. ▶156@Comment: pic006 (b14/image006.jpg)157*TS: that's gonna make this a lot easier. ▶158%act: draws line indicating 0.05 probability on figure159*TS: let's just say point o@l five because that's the most commonly used160one. ▶161@Comment: pic008 (b14/image008.jpg)162*TS: (.) so, if this is the poison distribution of scores, it's163possible, unlikely, but it's possible that our results can fall in164here [= pointing to inner region]. ▶165*TS: and what that would do, well, what this says is that if there is a166difference and we say “ok (.) there is a difference” with a ninety167five percent chance of being right. ▶168*TS: there's a ninety five percent chance of sureness in that and only a169five percent chance of not necessarily being sure. ▶170*TS: um, sometimes what experimenters do is they'll say, “All right,171that's supposed to be real specific I don't want anything left to172chance, we're going to set our levels to point o@l o@l one”. ▶173%act: draws line indicating 0.001 on figure174*TS: that gives you a much smaller probability of making a mistake. ▶175*SS: &=nods uh huh. ▶176*TS: and we'll see this a little bit more when we come back and talk177about, uh, type one and two errors. ▶178*TS: because that's related, interrelated with your, uh, probability179level. ▶180*SS: what do you, what do you do? ▶181*SS: do you reject the null hypothesis if it's, like if it's point o@l182one for your alpha. ▶183*SS: if it's less than that do you reject it or when it's greater? ▶184*TS: greater. ▶185*SS: ok. ▶186*TS: we're actually going to run through and do a manual t_test before187you, it's actually, if you just understand the formula and why we do188things it's a lot, it's not hard at all. ▶189*TS: I'm not gonna, you're not expected to memorize the formula. ▶190*TS: everyone does t_test by computer nowadays anyway. ▶191%act: erases the board192*TS: ok, how about sample size? ▶193*TS: (.) all right, sample size often refers to simply an n@l. ▶194*TS: and it's very important in relation to this, uh, accept and reject195null hypothesis. ▶196*TS: and you'll see it more when we plug it into the t, the t_test197formula. ▶198*TS: but the higher your N, the more likely you are to obtain an199accurate measure of the true population. ▶200*SS: &=nods head yes . ▶201*TS: that pretty much xxx. ▶202*TS: but what's interesting is if you have an extremely high n@l, a one203thousdand subjects participating, and you have a very small204difference, because your N is high, your difference can be smaller205and still be significant. ▶206*SS: &=nods yes . ▶207*TS: I'm going to ask you to accept that until we get to the formula,208then it'll be real clear. ▶209*TS: so, when they say in statistics, some, you know, can lie. ▶210*TS: if I have a point that I want to prove, I'm just going to go out an211get tons, and tons of subjects and even a small difference, which212might even, you know, we'd just almost expect it. ▶213*SS: oh! ▶214*TS: then I can end up, uh, rejecting the null hypothesis. ▶215*TS: ok, t_test and the f_test. ▶216*TS: um, they're both similar in their function. ▶217*TS: um, the F_test is an expansion of the t_test really. ▶218*TS: and what do these two t_tests do? ▶219*TS: they, uh, essentially are statistical tests that tell you whether220you should accept or reject the null hypothesis. ▶221*TS: and, um, a score is obtained and you look up the score on a table222and it, and using, using you're, uh, alpha level, your significance223level, um, you can determine whether or not you're, um, you reject224the null hypothesis. ▶225*TS: xxx . ▶226*SS: ok. ▶227*SS: they're both the same right? ▶228*SS: in function? ▶229*SS: xxx. ▶230*TS: yeah. ▶231*TS: in function they're the same, yeah. ▶232*SS: although you do, you use them in a different situation. ▶233*TS: yeah. ▶234*TS: the t_test is ideally suited to compare two groups and only two235groups and look at whether or not there's a difference between the236two. ▶237*SS: &=nods yes. ▶238*TS: and xxx it's a good test for that. ▶239*TS: um, the F_test is a more general statistic test and you can use it240to compare, you could use it to compare two, two different, uh,241groups, but you can also use it to compare three or more groups. ▶242*TS: and, you don't need to worry about this, but you can use it to243determine the results of factorial designs. ▶244*TS: I don't know if that's the next tutoring session or a couple down245the road, but it's, it's only two xxx. ▶246*TS: ok +... ▶247*TS: so, um, if the t_test yields a critical score and this critical248score (.) is less than your alpha level, then you reject the null249hypothesis. ▶250%act: writes something off screen251*SS: you get the critical from the book? ▶252*TS: hmm? ▶253*SS: you get the critical from back of the book? ▶254*TS: from the table, uh huh, and I'll show you how to use the table. ▶255*SS: good. ▶256*TS: you probably remember from statistics. ▶257@Comment: pic010 (b14/image010.jpg)258*TS: ok, what the t_test really is, remember you have two groups (.) the259group difference over the within group (.) xxx. ▶260%act: draws t-test formula261*TS: and formula (.) ok, the group difference is the difference between262the means of the two different groups. ▶263*TS: so, you've got two group means. ▶264*TS: the within group variability is the amount of variability of the265scores about the mean. ▶266*TS: and, essentially you can think of it this way. ▶267%act: draws something that can't be seen268*TS: the group means over the random error. ▶269*TS: and you can see then that t would get larger if you had a larger270number as the group differences increased. ▶271*SS: uh huh. ▶272*TS: ok? ▶273*TS: so, the formula, now, the formula the book gave, I'm not sure274exactly how it was derived. ▶275*TS: I haven't really looked at it. ▶276*TS: but I use simpler formula from another book that suits us just as277fine. ▶278%act: draws second formula, can't see it279*TS: it's the sample mean minus the population mean (.) xxx standard280deviation times the square root of the sample size. ▶281*TS: now, you would use this formula if you knew, you have to know the282population. ▶283*TS: so, if it's something like a census, you're seeing how you're group284compares to most Americans xxx. ▶285*TS: so +... ▶286*TS: I lied to you a little while ago when I said that if it's smaller287(.) if the critical value is larger (.) wait, the value, if this is,288if t@l is larger than the critical value in the table, then you say289there's a significant difference, not if it's smaller. ▶290*SS: ok. ▶291*TS: so, say you have a critical value of four point six nine or292something like that. ▶293*TS: and this [= points to formula on board] yields something higher. ▶294*TS: then you would say it's significant, your difference is significant295and you'd reject the null hypothesis. ▶296*TS: I'm sorry [= rest of sentence mumbled]. ▶297*TS: all right, um, when you need to use the table, you have to be able298to determine the degrees of freedom and that's the amount that the,299uh, values of the observation are free to vary. ▶300*TS: I know that's kind of a vague description. ▶301*TS: um, but do you understand how to use it statistically? ▶302%act: writes the formula for computing degrees of freedom on the board303*TS: um, what it is is, uh, the number in group one plus the number in304group two plus the number in group three minus the number of groups.305▶306@Comment: pic012 (b14/image012.jpg)307*TS: so, if we have one group with ten subjects, the degrees of freedom308would simply be. ▶309*TS: if you have three groups with ten subjects each, xxx. ▶310*SS: &=nods yes. ▶311*TS: if you had different numbers you'd still be able to think about it.312▶313*TS: and if the number of scores that are free to vary. ▶314*SS: what's that mean, free to vary? ▶315*TS: free to, um +... ▶316*TS: Let's say you have a frequency distribution like this, the degrees317of freedom oughta be the amount which is the central point tendency318xxx free to vary and still be significant. ▶319*SS: ok. ▶320*TS: ok? ▶321*TS: um, xxx. ▶322*TS: there's one_tailed and two_tailed designs. ▶323*TS: and that's pretty much straightforward. ▶324*TS: tell me the difference between them. ▶325*SS: uh, two_tailed and one_tailed? ▶326*SS: in case you don't know the results and you're trying to prove327either or. ▶328*TS: it's if you say that, “You know, I don't know if there's going to329be an increase or a decrease, but I think my treatment is going to330show a difference”. ▶331*TS: so, what you do is you would have, you set your significance level332at point five and set it at both ends, xxx at each end. ▶333*TS: so it's kind of a (.) the movement restricts it a little more. ▶334*TS: and you also use the table differently. ▶335*TS: and of course a one_tailed test is when you make a prediction that336it's going to be greater or it's going to be smaller. ▶337*TS: now we're going to talk about type one and type two errors. ▶338*TS: xxx. ▶339%act: erases the board340*TS: ok. ▶341*TS: um, a type_one error is, uh? ▶342*SS: rejecting the null hypothesis when it's really true. ▶343*TS: uh huh. ▶344*TS: and type_two ? ▶345*SS: is, uh, accepting it when it's true, false, when it's false. ▶346*TS: the best way to think of this, I have not seen any, a decision347measure before. ▶348%act: draws a table depicting type_one and type_two errors349@Comment: pic014 (b14/image014.jpg)350*TS: so, let's see (.) make a true statement xxx (.) null hypothesis is351true (.) null hypothesis is xxx false. ▶352*TS: you accept &=achoo ! ▶353*TS: excuse me. ▶354*TS: so, if we accept it when it's true, that's good. ▶355*TS: and if we reject it when it's false, that's good. ▶356*TS: but if we accept it when it's actually false, we've made a type_two357error. ▶358*TS: if we reject it when it's true, we've made a type_one. ▶359*TS: the book goes on to say that this is in psychology360[= points to box depicting type_one error] probably a more dangerous361mistake to make. ▶362*TS: because what that means is that, uh, we're rejecting, when there363really isn't the difference we're saying that there is. ▶364*TS: and just the way political journal system works is that they're365not, they don't want to print articles that show no difference, even366though it's just as good science. ▶367*TS: they don't want it. ▶368*TS: the want to show that this treatment has a, makes a difference. ▶369*TS: so this has a good chance of being funded. ▶370*TS: if you print it, then it's going to be latched on to by people and371before you know it there's an investment in this particular idea and372it's just as damaging especially when you find out it's false. ▶373*TS: on the other hand, this one right here374[= points to box depicting type_two error], um, we say “Well, there375is no difference” when actually there is a difference. ▶376*TS: well, somebody following will come along and will show you xxx. ▶377*TS: in psychology, we may think of this one as a more dangerous type of378mistake. ▶379*TS: however, if, uh, we're in a court case (.) and this is also380type_one and type_two error. ▶381%act: writes something off screen382*TS: um, with a type_one error we reject the null hypothesis but the383null hypothesis is true and we have an F@l or t@l value that says384we're ninety five percent sure that we made the right decision. ▶385*TS: but that still gives us a five percent chance of making a mistake.386▶387*TS: and, uh, you're probability of making a type_one error is always388your significance level. ▶389*TS: so, the lower your significance level, the lower the chance of390making a type_one error. ▶391*TS: now there's not really that, ok, so this point o@l five (.) five392out of one hundred times we run the experiment, we're going to get a393type_one error. ▶394*TS: so (.) xxx that's pretty good chance. ▶395*TS: there's probably (.) all right, you can think of it this way. ▶396*TS: if there's a five percent chance of making a mistake, saying397there's a difference when there isn't a difference and nobody prints398articles that are the other way, that show no difference, so this399takes up one hundred percent of the market, then you can say that400ten percent of all research in the journals is the result of401type_one error xxx. ▶402*TS: that's a pretty scary concept. ▶403*TS: just a little aside there &=laughs . ▶404*TS: xxx cynical, but xxx. ▶405*TS: a type_two error, um, you can't really assign it the same, so, same406was as the significance level. ▶407*TS: it's hard to, uh, um, say what your chance of making a type_one408error you increase the chance of making a type_two error. ▶409*TS: so, if you say, “hey, yeah, I'm going to set my significance level410I'm definitely going to get good results”. ▶411*TS: but that's going to, uh, really hinder my chances of, um, that's412going to increase my chances of making a type_two error, of saying413there's no difference when there really is. ▶414*SS: why's that? ▶415*TS: because this is so restrictive, this particular number is so416restrictive that the data has to fall within a tight. ▶417*SS: oh, Ok, I see. ▶418*TS: difference. ▶419*TS: so, if we make it hard to reach that null hypothesis, which is what420we're doing here, we're making it hard to reject the null421hypothesis, so if you can, if we're real sure of our results, then422it makes it hard to reject our null hypothesis. ▶423*TS: that increases our chances of making a type_two error. ▶424*TS: Um, choosing an alpha level. ▶425%act: flips through the book426*TS: choosing significance level, um (.) what I usually see is point427zero five or point zero one. ▶428*TS: and that's really for no reason. ▶429*TS: I've seen, sometimes I've seen one. ▶430*SS: &=nods yes and mumbles. ▶431*TS: uh huh. ▶432*TS: sometimes, you'll see in different types of studies, if you're433doing a preliminary type of study where you're trying a brand new434treatment and you're really not sure, you know, what's going to435happen, you might set your alpha level as high as point two five. ▶436*TS: and that's because you're looking for maybe subtle changes and437subtle differences. ▶438*TS: because your technique isn't refined and you may have one part of439your technique that works and parts that don't. ▶440*TS: so, if you're doing like, almost a preliminary type of research,441sometimes you will set this very high. ▶442*TS: uh, on the other hand, if you're doing replications of somebody443else's studies, you think to yourself +"/. ▶444*TS: +" man, I read this study but I don't, I don't see how that is445possible, knowing what we know, how that could be (.) I'm going to446try a study, but I'm going to restrict the significance level a447little bit and see, and see if he didn't create a type_one error so448I'm going to make mine point zero one while his was point zero five449. ▶450*TS: but generally you see point zero one or point zero five as the451significance level. ▶452%act: writes Statistical Vs. Clinical Significance on board453*TS: and that's just because with a clinical background, I think of it454that way. ▶455*TS: but, say we have a particular therapy. ▶456*TS: we're working with clients now and we have a particular therapy and457one hundred subjects and we administer this therapy and, low and458behold, they're scoring higher on their, uh, their little written459tests we're giving them. ▶460*TS: but we don't really see much of a marked improvement in how they're461getting along and things like that. ▶462*TS: so, we'd say “yes, it's statistically significant, but it doesn't463help them, so it's not clinically significant”. ▶464*TS: and I think that's an important argument. ▶465*TS: that's, that's one of the big philosophy questions in psychology,466is this, uh, distinction that you have between the two types of467significance. ▶468*TS: and now, there are people that say, “now, I don't really care about469this [= referring to statistical significance] what's important”. ▶470*TS: and in clinical psychology I would, I would agree that it's a471little different goal. ▶472*TS: but in experimental psychology you'd see a little less of that. ▶473*TS: so, uh, say we, we've set our alpha level real low, real low and we474say “well, there's no difference” . ▶475*TS: xxx well, we've increased our chances of making a type_two error476and there actually maybe a difference and we just didn't detect it.477▶478*TS: so, when we accept the null hypothesis. ▶479*TS: that not necessarily meaning that you xxx, and another thing I said480earlier was that if you reject the null hypothesis then you're not481necessarily proving your experimental hypothesis by doing that. ▶482*SS: &=nods head yes . ▶483*TS: &=erases part of the board . ▶484*TS: ok, (.) any questions so far? ▶485*TS: what I want to do now is ask you questions. ▶486*TS: (.) I have a bunch of them for you. ▶487*TS: and they're just basically over what we covered. ▶488*SS: xxx. ▶489*TS: um, what's the difference between the null and experimental490hypothesis? ▶491*SS: ok, the null is, uh, showing that there's no difference while the492alternative is that there is a difference. ▶493*TS: um, when does a researcher reject the null hypothesis? ▶494*SS: when the, when the probability, uh, is the relationship is, I495didn't get that question. ▶496*TS: ignore that question. ▶497*TS: we'll do a t_test and look on the chart and do all that. ▶498*TS: I didn't go over that with you. ▶499*TS: um, when we say something is statistically significant, what are500we, what are we saying? ▶501*SS: that, uh +/. ▶502*TS: not really in terms of looking at the table, but just in general503terms. ▶504*SS: that, uh, that there's a difference in the independent variable505xxx. ▶506*TS: uh huh. ▶507*TS: so, the difference is due to the? ▶508*SS: xxx. ▶509*TS: and not due to? ▶510*SS: random chance. ▶511*TS: right. ▶512*TS: very good. ▶513*TS: um, differentiate between type_one and type_two error. ▶514*SS: ok. ▶515*SS: type I is, uh, rejecting the null when it's true and type_two is,516uh, accepting the null when it is false. ▶517*TS: uh huh (.) Ok. ▶518*TS: (.) umm, describe two hypothetical experiments, one being a519one_tailed test ant two, being at two_tailed test. ▶520*SS: a two_tailed test would probably be, uh, (.) showing whether a521particular drug had a negative or positive effects on a sample or522population. ▶523*TS: ok. ▶524*SS: and a one_tailed test being that if a drug does indeed have a525positive, a positive. ▶526*TS: ok. ▶527*SS: effect. ▶528*SS: ok? ▶529*TS: so, xxx you look at, uh, in early research on some unknown530relationship, you might want to use a two_tailed test xxx. ▶531*TS: ok, what are the two ways we make groups look alike? ▶532*SS: uh, random sampling. ▶533*TS: uh huh. ▶534*SS: and, uh, manipulation of control. ▶535*TS: you use control for every variable? ▶536*TS: it's not impossible! ▶537*TS: but, you will if you're not matching. ▶538*TS: xxx control and that's within the, find a group subjects and then539get the control group so you can, so you've got a lot of people at540this age and from this background. ▶541*TS: and that's called matching. ▶542*SS: xxx. ▶543*TS: at the end of the semester. ▶544*TS: um, when is a t_test appropriate? ▶545*SS: uh, when is it? ▶546*TS: uh huh. ▶547*SS: uh (.) I guess when you want to, uh, try to find a (.) significant548difference between two populations? ▶549*TS: right. ▶550*TS: if you're looking at two different groups and you want to see if551there's a difference between the two groups. ▶552*TS: that's also the importance of being equivalent. ▶553*TS: and when is an F_test appropriate? ▶554*SS: xxx. ▶555*TS: it's just the more general effect. ▶556@Comment: pic016 (b14/image016.jpg)557*TS: and you can also use it in factorial designs xxx. ▶558*TS: ok. ▶559*TS: um. ▶560%act: erases the board and writes "group" and "?" then writes A, B, B, D561under group and 6, 4, 4, 5 under "?"562*TS: ok. ▶563*TS: I'm doing a study with four different groups. ▶564*TS: um, what would be my degrees of freedom? ▶565*SS: uh, fifteen, wait xxx six, four, four. ▶566*TS: nineteen minus. ▶567*SS: fourteen. ▶568*TS: four, yeah, right. ▶569*TS: degrees of freedom. ▶570%act: writes something on the board, obstructed by digital numbers on571screen572*TS: oh, ok. ▶573*TS: what's the relationship between significance level and, uh,574probability of making type_one and type_two errors? ▶575*SS: uh, the lower the significance level, the more, uh, the more576accurate, I guess, your hyp, type_one will be? ▶577*SS: xxx. ▶578*TS: xxx. ▶579*SS: yeah. ▶580*TS: right. ▶581*TS: and it's if your significance level is too high? ▶582*TS: (.) hmm, a point twenty five significance level. ▶583*SS: you need a more chance, have a better chance of making a type_one?584▶585*TS: uh huh. ▶586*TS: because you're saying, I don't know, even if it varies even a587little bit, we'll call it significant. ▶588*TS: ok. ▶589*TS: what I'm going to do now is give you a hypothetical experiment590+... ▶591%act: erases the board592*TS: and I'll let you figure out, um, (.) a few things about it. ▶593*TS: (.) uh, let's go back to (.) uh, t_tests. ▶594*TS: you know, that's population mean and group mean. ▶595*TS: right? ▶596*TS: and your deviation mean is like that. ▶597*TS: ok. ▶598%act: writes something on the board, obstructed by digital numbers on599screen600*TS: a researcher is examining fetal alcohol syndrome. ▶601*TS: can you xxx that? ▶602*TS: ok, and believes that alcohol consumed during pregnancy lowers603birth weight. ▶604*TS: so, what would be our experimental hypothesis and our null605hypothesis with that? ▶606*SS: well, would you repeat that? ▶607*TS: a researcher is examining fetal alcohol syndrome and believes that608alcohol consumed during pregnancy lowers birth weight. ▶609*SS: ok, so your, uh, experimental hypothesis would be, uh, (.)610increased alcohol causes, uh, decreased birth weight. ▶611*TS: uh huh. ▶612*SS: and the null would be, uh, that, that the increase of alcohol613doesn't effect birth weight. ▶614*TS: right. ▶615*TS: and would you call this a one_or two_tailed test? ▶616*SS: uh, I'd say two_tailed. ▶617*TS: because? ▶618*SS: because you really don't know the xxx you're trying to xxx. ▶619*TS: but we're saying that, uh, a researcher believes that alcohol620lowers. ▶621*SS: ok, then I would change it to one_tailed. ▶622*TS: right. ▶623@Comment: pic018 (b14/image018.jpg)624*TS: because he's got a specific belief about what's going on. ▶625*TS: ok. ▶626*TS: um, pregnant female rats are injected with alcohol and the weight627of their pups is measured. ▶628*TS: so, our independent variable is alcohol. ▶629%act: writes IV equals alcohol and DV equals birth weight underneath the630stuff previously obstructed631*TS: and our dependent variable (.) the variable we're measuring (.) is632birth weight of the baby rats. ▶633*TS: ok? ▶634*TS: um, extensive prior research has shown that this particular strain635of rats has, um, an average birth weight of eight point four grams.636▶637*TS: so, looking at this formula . ▶638*SS: xxx. ▶639*TS: our population mean is eight point four. ▶640%act: underneath all that, writes u@l equals eight point four641@Comment: pic020 (b14/image020.jpg)642*TS: that's from all the research done with this particular kind of643frat. ▶644*SS: would that be xxx, too? ▶645*TS: um, no. ▶646*TS: it's just that it's convenient. ▶647*SS: oh. ▶648*TS: I could use a more complicated version of this. ▶649*TS: if we didn't know the population mean and we were just comparing650two different groups. ▶651*SS: &=shakes head yes. ▶652*TS: but in this case, we're just comparing one group that we don't know653xxx. ▶654@Comment: pic022 (b14/image022.jpg)655*TS: the researcher's sample of one hundred pups. ▶656%act: writes x@l equal one hundred657*TS: shows that they weight an average of seven point nine eight grams.658▶659%act: writes blank equals seven point eight660@Comment: pic024 (b14/image024.jpg)661*TS: and a standard deviation of one point three. ▶662*TS: (.) so, you want to set up the equation? ▶663*TS: xxx. ▶664*SS: xxx. ▶665*TS: no, I, I want you to write it up there first666[= pointing to the board]. ▶667@Comment: pic026668*TS: ok. ▶669@Comment: pic028 (b14/image028.jpg)670*TS: (.) um (.) do you have a calculator? ▶671*SS: how many decimal places? ▶672*TS: um, just take it to three places. ▶673%act: adds a two onto negative point zero three674@Comment: pic030 (b14/image030.jpg)675*TS: ok. ▶676*TS: now, what do we do, find out if, if it is actually significant. ▶677*TS: um, first of all, what are the degrees of freedom? ▶678*TS: of the particular experiment. ▶679*SS: how many in each group? ▶680*TS: xxx well, actually there's a hundred and (.) there's act Hmm! ▶681*TS: it's actually only one group. ▶682*SS: one group of a hundred? ▶683*TS: uh huh. ▶684*SS: ok, I would say ninety nine. ▶685*TS: yes. ▶686*SS: ok +... ▶687*TS: ok, we go to page two hundred and seventy five (.) and uh, uh,688let's say we're using an alpha level of, uh, you would look where it689says, “point zero five”. ▶690%act: writes point zero five next to everything691@Comment: pic032 (b14/image032.jpg)692*SS: uh huh. ▶693*TS: because uh, where we've indicated it's a one_tailed test. ▶694*TS: if it was a two tailed test, we'd use this column695[= pointing to something in the book] and the bottom number but we696use this number right here. ▶697*TS: it's a one_tailed test and you find their degrees of freedom, let's698say it's somewhere between sixty and one hundred and twenty,699alright. ▶700*TS: xxx +... ▶701*TS: and if our number is greater than that number, then we would say we702have +... ▶703*TS: if, if it's, uh, greater than that number, we would have704significance. ▶705*TS: in this case, t@l equals point zero three two which is not less, so706you'd say it's not significant. ▶707*SS: &=shakes head yes. ▶708*TS: xxx different data which probably just brought back a rush of709memories from statistics. ▶710*SS: &=laughs that was a couple of years ago. ▶711*TS: xxx do you have any, uh, questions on any of this material? ▶712*SS: you've just about answered them all. ▶713*TS: ok. ▶714*TS: that's, that's not too, uh, it's not too difficult. ▶715*TS: the hard part of the concepts behind the statistics. ▶716*TS: it's always easy to plug in numbers. ▶717*TS: but the important thing is you understand, um, why you are plugging718in which numbers and what, what everything means. ▶719*SS: &=shakes head yes. ▶720*TS: when we say something's significant, what does that mean? ▶721*TS: xxx ok, well, we're out of time. ▶722*TS: xxx. ▶723*SS: alright. ▶724*TS: did you have class today? ▶725@End