0@Loc: Person/c28.cha1@PID: 11312/t-00015821-12@Begin3@Languages: eng4@Participants: TS Teacher, SS Student5@ID: eng|Person|TS|||||Teacher|||6@ID: eng|Person|SS|||||Student|||7@Media: c28, video8@Comment: C-28 STATISTICS9*SS: xxx. ▶10*TS: well, just talk about anything you want to talk about. ▶11*SS: alright. ▶12*SS: uh, (.) xxx. ▶13*TS: oh really? ▶14*SS: right. ▶15*TS: well, that's an astute observation. ▶16*TS: we'll keep that on tape. ▶17*TS: just read it out loud for me [= indicates card]. ▶18*SS: topics, statistics, type one and type two errors, decision matrix,19and probability at point zero five, and a t@l test. ▶20*TS: ok, great. ▶21*TS: thanks. ▶22*TS: I whipped that out from under you. ▶23*TS: &=laughs ok, um, just to start out with did you have any particular24questions I want to make sure we time to cover any special concerns?25▶26*SS: well, I don't know if I really (.) got the basic (.) grip on all of27it. ▶28*SS: 0 [= looking through book]. ▶29*TS: ok. ▶30*TS: well, we can just go through it and see what happens. ▶31*SS: ok. ▶32*TS: if you, interrupt if you have any specific concerns you want to33cover. ▶34*TS: ok. ▶35*TS: you remember coming from last week, and just sort of the general36idea of what are inferential statistics are? ▶37*SS: they allow us to make inferences or general, generalizations. ▶38*TS: 0 [= nods head] right. ▶39*SS: about (.) what the sample (.) compared to the population. ▶40*TS: in other words. ▶41*SS: by using the sample. ▶42*TS: we can, we can. ▶43*SS: make an inference. ▶44*TS: yeah. ▶45*TS: and it allows us to determine what, to what degree we are sure the46results of our sample pertain to the population. ▶47*SS: uh huh. ▶48*TS: what you said [= gestures at S] just in different terms. ▶49*SS: ok. ▶50*TS: alright, but even if a researcher has made sure that he or she has51equivalent groups, like experiment controling all extraneous52variables or by randomization or whatever, does this mean that any53differences in the dependent measures must necessarily be due to the54independent variable? ▶55*SS: (.) no. ▶56*TS: ok. ▶57*TS: why not? ▶58*SS: um (.) it could just be, um, by chance. ▶59*TS: uh huh. ▶60*SS: factor xxx. ▶61*TS: that's fine. ▶62*TS: in other words, you're right. ▶63*TS: it could be error, it could be sampling error or measurement error,64or whatever. ▶65*TS: that's the whole point, that's whole reason why we need inferential66statistics. ▶67*TS: it is, is a way, kind of assessing how much error we have in our68data and how that affects what we can then say about the population69based on the sample. ▶70*TS: (o)kay? ▶71*TS: so that's really the whole point. ▶72*TS: if you remember that, you know, even if you've done, very precise73procedures and chosen reliable, valid measures, etc. ▶74*TS: there's still error. ▶75*TS: there's going to be error. ▶76*TS: this is not a perfect world, or whatever. ▶77*TS: at least not our measure of it. ▶78*TS: and that's the whole point of using inferential statistics. ▶79*TS: xxx what happens and how much of that is xxx. ▶80*TS: ok. ▶81*TS: so, you've got xxx descriptive statistics. ▶82*TS: but before we do anything in experimentation we've got to have our83hypothesis, right? ▶84*SS: right. ▶85*TS: ok. ▶86*TS: so, what two types of hypotheses do we usually deal with? ▶87*SS: the null and the alternate xxx. ▶88*TS: ok. ▶89*TS: so, what is the null symbolized as? ▶90*SS: h@l o@l. ▶91%act: TS writes symbols for null hypothesis on board while talking92@Comment: pic002 (c28/image002.jpg)93*TS: ok. ▶94*TS: and what does it state? ▶95*SS: it states that there is no difference between the means of the96population. ▶97*TS: ok. ▶98@Comment: pic004 (c28/image004.jpg)99*TS: in other words [= writes symbols on board while talking]. ▶100*SS: equals. ▶101*TS: our population, our sample mean equals our population mean, or our102sample means of two groups is equal. ▶103%act: finishes writing equation on board104@Comment: pic006 (c28/image006.jpg)105*SS: right. ▶106*TS: to each other. ▶107*SS: right. ▶108*TS: if you drew two samples from the population then they'd be equal. ▶109*SS: 0 [= nods head] yeah. ▶110*TS: in other words, you usually speak of testing, testing means, so you111speak of testing means, you're looking at whether or not they're112equal or different. ▶113*TS: so, xxx testing means, the null hypothesis means the means are114equal. ▶115*SS: 0 [= nods head]. ▶116*TS: right. ▶117@Comment: pic008 (c28/image008.jpg)118*TS: ok. ▶119*TS: what else does it mean then in terms of independent and dependent120variable relationship? ▶121*TS: (.) our means are equal, that means our sample, our two samples122came from the same population xxx other population equal the mean123population xxx population mean. ▶124*TS: in other words, our independent variable had no effect. ▶125%act: writes on board while talking126*TS: or significant effect. ▶127*SS: oh, I see what you're saying. ▶128*SS: so. ▶129*TS: see? ▶130*TS: what the null means is the population means are equal, I mean our131means are equal, our group means are equal, and our independent132variable, therefore, had no effect. ▶133*TS: if it had an effect then you would expect them to be different. ▶134*SS: ok. ▶135*TS: that's the whole point of testing groups. ▶136*SS: alright. ▶137*SS: so no matter what you xxx independent variable. ▶138*TS: well, you're usually only testing it as per one research endeavor.139▶140*SS: right. ▶141*TS: in other words, um, just using the example, xxx what might be an142infinite variable relationship, maybe effecting amount of study on143test performance. ▶144*SS: 0 [= nods head]. ▶145*TS: if you had x@l one equals five hours of study; x@l two equals ten146hours of study. ▶147%act: writes on board out of camera range148*TS: ok. ▶149*TS: in terms of this. ▶150*TS: ok. ▶151*TS: and our dependent measure xxx. ▶152*SS: so these. ▶153*TS: in terms of the null then, the null hypothesis is gong to say the154means of these two groups. ▶155*TS: the mean performance, the mean test score these two groups is going156to be the same. ▶157*TS: in other words, xxx. ▶158*SS: 0 [= nods head] ok. ▶159*SS: I understand it. ▶160*TS: ok. ▶161*TS: so that's all it, I mean, it all says several things. ▶162*TS: I mean, when you look, when you, when you talk about coming from163the same population that's what it means. ▶164*TS: in other words, if the i_v had an effects, then, in other words,165you're saying that these two groups came from the same population. ▶166*TS: they don't differ. ▶167*TS: they do not differ according to this characteristic of hours of168study. ▶169*TS: you know. ▶170*SS: 0 [= nods head]. ▶171*TS: that confusing? ▶172*SS: no, that's, that's clear. ▶173*TS: I xxx confuse. ▶174*TS: ok. ▶175*TS: alright. ▶176*TS: in other words, the null hypothesis states that the independent177variable has no effect, it doesn't make a difference in our178population, whether they have it or not, I mean which level they179have. ▶180*SS: right. ▶181*TS: the mean for them is going to be the same no matter which condition182xxx. ▶183@Comment: pic010 (c28/image010.jpg)184*TS: that's the null. ▶185*TS: ok. ▶186*TS: and then the alternative hypothesis. ▶187*TS: xxx h@l one. ▶188%act: writes on board while talking189*TS: says what? ▶190*SS: the means are different. ▶191*TS: right. ▶192%act: writes on board193@Comment: pic012 (c28/image012.jpg)194*SS: it means that your independent variable had an effect. ▶195*TS: right. ▶196%act: writes on board197*SS: when we, when we say “had no significant effect, effect” that means198we had, that means using the point o@l five? ▶199*TS: by convention what's usually (.) um, considered significant is the200same thing as at the point o@l five level. ▶201*TS: in other words, ninety five percent confidence that our results are202(.) representative of xxx. ▶203*TS: in other words, you're ninety five percent sure that, there's204ninety five percent chance that our results couldn't have occured205under the null hypothesis (.) there is a five percent chance that206they could have but [= gestures]. ▶207*SS: uh huh. ▶208*SS: well, in this case209[= points at alternate hypothesis in picture six], does that mean210we're saying there's a greater than five percent chance that xxx211that won't be xxx? ▶212*TS: ok. ▶213*TS: the way all this works &=laughs . ▶214*SS: ok. ▶215*TS: let's get into it. ▶216*TS: alright. ▶217*TS: when we speak about (.) ok, let's probably in other words the218probability level of our sampling distripution ect . ▶219*TS: ok, first of all, the logic behind using the null and alternative220hypothesis is that if we can determine that the null is incorrect,221then we can research, we can accept the research hypothesis. ▶222*TS: in other words, we speak of rejecting or failing to reject the null223[= points at alternative hypothesis in picture six] . ▶224*TS: ok? ▶225*SS: uh huh. ▶226*TS: alright. ▶227*TS: the independent, if you reject the null then you say that the228independent variable probably has an effect on the dependent229variable, Ok? ▶230*TS: the null is an exact statement. ▶231*TS: it says that the population means are exactly equal, Ok? ▶232*TS: this allows us to calculate precisely the probability of the233observed outcome occurring when the null hypothesis is true. ▶234*TS: ok? ▶235*TS: if you calculated exactly it'd be the exact same because both sides236are equal, Ok? ▶237*TS: and you do that with our sampling distribution, Ok? ▶238*SS: ok. ▶239*TS: you have a sampling distribution of the means, Ok? ▶240@Comment: pic014 (c28/image014.jpg)241*TS: of all possible means. ▶242%act: draws normal curve on board243*TS: ok? ▶244@Comment: pic016 (c28/image016.jpg)245*TS: xxx distribution numbers xxx. ▶246%act: marks on normal curve247*TS: xxx people say it's a theoretical xxx distribution. ▶248*TS: not a particular number of xxx, Ok? ▶249*SS: 0 [= nods head]. ▶250*TS: if the null hypothesis is true, that is if the population means are251exactly equal, then we can figure out what is the probability of252each possible outcome when the null is true. ▶253*SS: (.) ok. ▶254*TS: ok? ▶255*TS: in other words, if we have xxx you know, kind of out there256hypothetically in the universe, we could take a different member of257the xxx of the population, sample from the population, get their258means, distribute the, distribute them, it's the same thing as259distribution, Ok? ▶260*TS: and it's from that sampling distribution that xxx standard normal261curve, and stuff like that? ▶262*TS: to get probability out of? ▶263*SS: yes. ▶264*TS: that's where that comes from. ▶265*SS: ok. ▶266*TS: this sampling distribution. ▶267*TS: ok? ▶268*TS: um, well, when we talk about probabilities, we are looking at (.)269the chances of that occurring when the null is true or not. ▶270*TS: ok? ▶271*TS: alright, when the null is true, that means the population is equal.272▶273*SS: alright. ▶274*TS: the population is equal, Ok? ▶275*TS: so, if we draw a sample from that population so there's, it's not,276it's rarely, if ever, equal exactly. ▶277*TS: your [= gestures]. ▶278*SS: 0 [= nods head]. ▶279*TS: alright? ▶280*TS: we're going to get some variance, in other words, from error281because of, of it being a sample and not representing the whole282population. ▶283*SS: yeah [= nods head]. ▶284*TS: you're not just xxx the sample, Ok? ▶285*TS: um, we can look at the probability of xxx to the chances that that,286that we could have attained a certain result. ▶287*TS: say if we got, uh, you know, the ten hour study people had288[= pointing at board] you know, made ninety five. ▶289*TS: I mean a mean ninety five. ▶290*TS: the five hours had a mean seventy five. ▶291*TS: there's a certain, um +... ▶292*TS: when you test the mean, there's a certain, we can look at the293probability that those kinds of results +... ▶294*TS: have occurred even while the null is still true. ▶295*TS: in other words, due to sampling error we got the result. ▶296*TS: but, if the probability of the chances that that result could have297occurred and the null still be true get so far away, in other words298that maybe there's such a big difference here [= pointing at board]299that it looks like it's so unlikely that the null really is true,300then we accept it. ▶301*TS: but that's, I mean we accept the alternative there and reject the302null. ▶303*TS: and, ah +... ▶304*TS: what we talked about the confidence level at point o@l five is that305we're saying that we're ninety five percent sure with our results in306other words there is a five percent chance that they could have307still occured under the null. ▶308*TS: in other words there is a five percent change that there really is309no difference but we are ninty five percent sure there is a310difference then we are gonna reject the null. ▶311*SS: right. ▶312*TS: ok so that's what that deals with. ▶313*TS: and as far as the (.) you know, the five percent I mean, that's314just a convention xxx. ▶315*SS: ok. ▶316*TS: in fact, a lot more xxx ninety nine percent point o@l one alpha317level is usually [= gestures]. ▶318*TS: these days, point o@l one is a little more, xxx it's often319considered more (.) reliable (.) no, it's not reliable. ▶320*TS: I don't mean reliable xxx aaaah. ▶321*SS: &=laughs . ▶322*TS: it's so hart to talk about statistics and use a word that you use323in just a general English sense. ▶324*SS: right. ▶325*SS: &=laughing . ▶326*TS: you know? ▶327*TS: it has some sort of meaning. ▶328*TS: but, you know, Ok. ▶329*TS: so did, did that answer what you were talking about? ▶330*TS: in other words, the probability is just what reference is used to331specify the, the chances of an event. ▶332*TS: in other words, the differences between these, the samples, will333occur if there's no difference in the population. ▶334*TS: in other words, it's the question “what is the probability of335obtaining this result, if only random error is operating”? ▶336%act: reading from book337*TS: If the chances of obtaining that result are very small if only338random error were operating we can think +"/. ▶339*TS: +" Well <hey> [!] something else is operating. ▶340*TS: in other words xxx. ▶341*SS: 0 [= nods head slowly]. ▶342*TS: 0 [= both laugh]. ▶343*SS: yeah, I, I +/. ▶344*TS: see if you can explain it back to me a little bit. ▶345*SS: ok. ▶346*TS: so I can get a handle on what you think. ▶347*SS: um (.) proba, probability we're saying that there's a five percent348chance that (.) um (.) not random error (.) we're five, we're ninety349five percent sure that we're right. ▶350*TS: we're ninety five percent confident. ▶351*SS: percent confident. ▶352*SS: that the, um (.) it turns. ▶353*TS: the results couldn't have occurred by chance. ▶354*SS: by chance. ▶355*TS: there's a five percent chance that it still might be random error,356i_e (.) that the null might still be true. ▶357*TS: 0 [= pointing at board] but we're ninety five percent sure that (.)358it failed. ▶359*TS: that we can reject the null xxx. ▶360*SS: right. ▶361*SS: now (.) can we, can we ninety five percent (.) confident accept the362null? ▶363*TS: ok. ▶364*TS: you usually (.) &=laughs . ▶365*SS: &=laughs . ▶366*TS: sure. ▶367*TS: well, the way it works is that you look at (.) your probability is368spoken in the negative case. ▶369*TS: you say that you are, we're ninety five percent confident of our370results. ▶371*TS: well, that could, I mean the confidence, I mean, what you're saying372when you talk about your significance level is you have a373significance level of, say, the point o@l five or less, then your374saying that that that alpha level is the chance that you're wrong. ▶375*TS: in other words, the chance that +... ▶376*TS: 0 [= laughs and mumbles] +... ▶377*TS: you're ninety five percent confident means that you're ninety five378percent confident that your results are correct. ▶379*SS: ok [= nods head]. ▶380*TS: ok? ▶381*TS: so what you're saying is that you're ninety five percent sure that382these results, whatever they are, xxx they are. ▶383*TS: if they are significant at the, at the point o@l five level, then384you're ninety five percent confident that your results reflect the385true difference. ▶386*TS: that means they're not due to random error. ▶387*SS: ok, in that, if they have a true difference, then you accept the388alternative? ▶389*TS: right. ▶390*SS: ok. ▶391*TS: in other words we reject the null. ▶392*SS: ok. ▶393*TS: alright? ▶394*SS: &=chuckles alright. ▶395*TS: ok. ▶396*TS: alright. ▶397*TS: when the sampling distribution xxx distribution xxx what you use to398determine the best probability level. ▶399*TS: ok. ▶400*TS: in other words, we use the same conditions to figure out what are401the chances of any particular outcome occurring [= clears throat]402under the null hypothesis. ▶403*TS: ok. ▶404*TS: sample size. ▶405*TS: just to talk a little bit more about samples. ▶406*TS: what happens as our sample size increases? ▶407*SS: the probability of random (.) guessing decreases. ▶408*TS: well, Ok, sure. ▶409*TS: in other words, we're more likely to obtain an accurate estimate. ▶410*TS: in other words. ▶411*SS: right. ▶412*TS: if we had a population, sometimes we have a population that413consists twenty people and we chose two out of those, and we did all414kinds of descriptive statistics xxx tried to predict about the415population from the sample, we'd be much more likely to paint an416accurate picture than if we got fifteen xxx. ▶417*SS: 0 [= nods head]. ▶418*TS: you know, there's much less, much greater precision just by virtue419of larger number of the sample. ▶420*TS: in other words (.) I mean obviously you're going to have to xxx to421get a chance xxx. ▶422*SS: right [= nods head]. ▶423*TS: ok. ▶424*TS: ok [= moves papers around]. ▶425*TS: ok. ▶426*TS: um, the t_tests and the f_tests xxx. ▶427*SS: uh huh. ▶428*TS: primary inferential statistics talked about. ▶429*TS: when do we want to use that t_test? ▶430*SS: when you have xxx between two groups. ▶431*TS: yeah, or xxx, right? ▶432*SS: right. ▶433*TS: um (.) the F_test then, helps xxx there's more. ▶434*SS: more. ▶435*SS: more than two. ▶436*TS: right. ▶437*TS: ok. ▶438*TS: alright, so we need to either, either xxx statistical test we need439to do a few things before we run our analysis, Ok? ▶440*TS: xxx [= pointing to the board] our hypothesis xxx. ▶441*TS: our null hypothesis in terms of this experiment are what? ▶442*SS: that (.) uh (.) &=gestures +... ▶443*TS: study time xxx. ▶444*SS: oh. ▶445*SS: that five hours of study will equal ten hours of study on the test446score. ▶447*TS: in other words, the means are not going to differ. ▶448*SS: right. ▶449*TS: significantly. ▶450*TS: in other words, those, changing the amount of studying make any451difference according to the null xxx. ▶452*SS: right. ▶453*TS: according to the alternative hypothesis [= pauses and looks at SS].454▶455*SS: that there will be a difference. ▶456*TS: in other words, studying different amounts will change the mean457test score. ▶458*SS: right. ▶459*TS: in other words, xxx. ▶460*TS: ok? ▶461*SS: (.) independent variables, that's another question. ▶462*SS: independent variables xxx? ▶463*TS: ok, in this study independent variables, yes, xxx. ▶464*TS: independent variable is the variable that you manipulate? ▶465*SS: right. ▶466*TS: or, or if it's something like gender, xxx “hmm we're going to make467you male” without. ▶468*SS: right. ▶469*TS: complex surgical procedures anyway, it can be the observed, you470know, the observed variable, you know. ▶471*TS: xxx. ▶472*TS: but, you know, you're supposed to have xxx groups there's different473levels of it xxx. ▶474*TS: ok, and xxx. ▶475*SS: 0 [= nods head]. ▶476*TS: xxx ok. ▶477*TS: alright. ▶478*TS: so, we've specified our hypothesis. ▶479*TS: ok. ▶480*TS: what else do we have to do before we xxx or a xxx test? ▶481*TS: (.). ▶482*TS: have you thought about that xxx? ▶483*SS: xxx. ▶484*SS: determine the probability, or significance level. ▶485*TS: significance level. ▶486*TS: ok. ▶487*TS: we set our significance level, ok? ▶488*TS: and by convention in social sciences xxx use the point o@l five489level [= writes ".5" on board]. ▶490*TS: ok? ▶491*TS: in other words, you're not going to, you're not going to believe492that there's a difference, (.) that our independent variable has an493effect, unless there is a five chance or less +... ▶494*TS: probability chance that our results occurred randomly under the495null. ▶496*TS: ok? ▶497*TS: in other words, only if you're ninety five percent sure, only if498there's a five percent chance that they didn't occur by chance499&=laughs ok, then are we going to xxx. ▶500*TS: and say “oh, xxx rejecting our null and excepting our alternative501hypothesis”. ▶502*TS: ok? ▶503*TS: point o@l five means that there's still a five percent chance xxx504that we're wrong. ▶505*TS: in other words, the statistical term that our results, there's a506five percent chance that our results still could have occurred even507though the null were true. ▶508*TS: ok? ▶509*SS: 0 [= nods head]. ▶510*TS: but, there's a ninety five percent chance that we're ninety five511percent confident xxx. ▶512*SS: so, do we always try to prove the alternative? ▶513*TS: ok, we don't, that's the tough part. ▶514*TS: usually, what I get out of this book is that they're not as515precise. ▶516*TS: xxx usually you speak in terms of accepting and rejecting the null,517ok? ▶518*TS: you could, I mean if you think about it logically, how could you519ever prove that there is no difference? ▶520*SS: right. ▶521*TS: I mean, 'cause you can't with the null hypothesis. ▶522*TS: that's not logistically even really xxx, you know? ▶523*TS: xxx what kind of population xxx, whole population. ▶524*TS: um (.) so you usually don't see xxx. ▶525*TS: you know? ▶526*TS: I mean, how could you ever prove the null? ▶527*TS: you know, if you think about our type one and type two errors you528realize you can't specify, you know, a chance to make a type two529error. ▶530*SS: oh. ▶531*TS: in other words, so, you know, but, um, I think (.) different532statisticians will allow you to xxx rules, but I think if you're533always talking about it in terms of accepting or rejecting the null,534you're Ok. ▶535*TS: you don't usually talk in terms of “proving” anything xxx. ▶536*TS: I mean, like we were talking about there's always going to be some537error which we will, can never determine exactly how much exactly538xxx estimated error. ▶539*SS: uh huh. ▶540*TS: but, it's just that statistics and a whole lot of social science541research is so probabilistic you never, ever xxx. ▶542*SS: ok. ▶543*TS: ok. ▶544*TS: we've specified our hypothesis, we've specified our significance545level, and as far as out t_test we need to decide beforehand whether546it's going to be one_tailed or two tailed, Ok? ▶547*TS: what might you think this one would be? ▶548*SS: (.) well (.) two_tailed. ▶549*SS: (be)cause you're specifying that (.) there's going to be a550difference between number of hours studied. ▶551*TS: ok. ▶552*TS: well, but that's really not what two tailed and one tailed is553though, Ok? ▶554*TS: you choose a two tailed test when you, Ok, for all this555hypothesizing that there's going to be a difference, if you thought556there wasn't going to be a difference, why would we run this cell557then. ▶558*TS: do you see? ▶559*SS: no. ▶560*TS: I mean, we don't go around in science trying to prove561Unrelationships. ▶562*TS: we're trying to look at the relationships, Ok? ▶563*TS: ok, so a two_tailed test means that we, oh Ok, our alternative564hypothesis says that the difference exist, but we're not sure in565what direction. ▶566*TS: in other words, we're not sure which one is going to be greater567than the other. ▶568*TS: but for, you know, intuitive reasons hopefully based on, you know,569your scientific knowledge and expertise, you have some kind of570reason xxx to suspect that one of these variables is going to cause571greater xxx than the other one the dependent measure. ▶572*TS: then, it's a directional test, or a one tailed. ▶573*TS: now one more time, what do you think? ▶574*SS: (.) I would say one_tailed. ▶575*TS: ok. ▶576*TS: sure. ▶577*TS: why? ▶578*SS: ten hours study would be positive direction over five hours. ▶579*TS: right. ▶580*TS: in other words, Ok, one_tailed means that one side is just greater581than the other. ▶582*SS: right. ▶583*TS: obviously xxx. ▶584@Comment: pic018 (c28/image018.jpg)585*TS: ok? ▶586*TS: so, on the graph then, the one_tail is when we pile all of our587significance level (.) in this position. ▶588%act: blackens one end or normal curve.589*TS: the two_tailed, on the other hand, is when xxx. ▶590%act: writes on both ends of normal curve591@Comment: pic020 (c28/image020.jpg)592*TS: our interval xxx distribution593[= points at both ends of normal curve in picture ten]. ▶594*TS: that would mean, let's say if you knew absolutely nothing about595hours of study and test scores and xxx when you think there might be596a difference but we don't know if ten hours will have higher test597scores or not, we're just going to xxx. ▶598*SS: uh huh. ▶599*TS: ok, you'd use a two_tail and it would look like that600[= points at picture ten]. ▶601*SS: do you always divide your significance level by two in a602two_tailed? ▶603*TS: well, that's what it is. ▶604*TS: it's whatever confidence level you choose. ▶605*TS: if you tested at point o@l one for a one_tailed test, then all of606our confidence interval would be, or not our confidence interval,607but all of our, um, when looking at our critical value table, the608critical values xxx distribution. ▶609*TS: we want them to be, um, all at one end for a one_tailed. ▶610*SS: uh huh. ▶611*TS: so, that's point o@l one. ▶612*TS: if it was a two_tailed it would be point o@l one divided by two. ▶613%act: writes .01/2 on board614*TS: that would be . ▶615*TS: o@l o@l (.) point o@l two, point o@l five or whatever it is. ▶616*SS: xxx. ▶617*TS: whatever. ▶618*SS: yeah. ▶619*TS: anyway, you'd have half of it over here and half of it over here620[= points to each end of normal curve in picture 10]. ▶621*TS: ok, so you xxx the critical t@l in any situation in which your622research hypothesis specifies that there's a difference. ▶623*TS: in other words, one group is greater than the other. ▶624*TS: or you can run a two_tail in a situation in which you do not625specify xxx. ▶626*TS: and the reason you do that is, because like I said, the critical627value xxx distribution xxx. ▶628*SS: 0 [= nods head]. ▶629*TS: alright if you look on page [= turns page] (.) one hundred and630forty two see. ▶631*SS: 0 [= turns in his book]. ▶632*TS: the critical value for a t_test, you have to exceed a critical633value of plus or minus two point one hundred and one. ▶634*TS: see that? ▶635*SS: Uh huh [= nods head]. ▶636*TS: ok, whereas on the one tailed, with all this clustered on one side,637you would only have to get up to something that's greater than point638o@l seven four. ▶639*TS: so the advantage of using one_tailed is that you don't have to get640as high of a t_value to be significant. ▶641*TS: the disadvantage is if you choose the wrong direction you're out of642luck. ▶643*SS: so let me ask you one. ▶644*SS: if something fell within this region [= pointing in book]. ▶645*TS: uh huh. ▶646*SS: (.) I would, I would say that is not significant? ▶647*TS: no, that's when it is. ▶648*TS: if your critical value, ok, your t_test, we haven't really talked649about the logic behind the t_test yet. ▶650*TS: your critical value of t, in other words, let me rephrase that. ▶651*TS: if you're xxx. ▶652*TS: you do all the calculations and get a value. ▶653*TS: it's called your t_value, Ok? ▶654*SS: Uh huh [= nods head]. ▶655*TS: you compare that to a critical value of t@l which comes from one of656these hypothetical distributions657[= points at normal curve on board]. ▶658*TS: and if your critical value of t, like in this example xxx659[= pointing in book] we're talking about, is greater than two point660one hundred and one either way, then your results, you're, you're661gonna reject the null. ▶662*TS: xxx, ok? ▶663*SS: ok. ▶664*TS: let's look at what goes on when we calculate a t_test. ▶665%act: erases board666*SS: o [= holds marker board steady]. ▶667*TS: thanks. ▶668*TS: it could fall down on top xxx. ▶669*TS: ok. ▶670*TS: a t_test is a radio of two types (.) of (.) variance, Ok? ▶671*TS: right? ▶672*SS: 0 [= doesn't answer, looking at notes]. ▶673*TS: are you with me? ▶674*SS: xxx looking at my notes xxx. ▶675*TS: ok. ▶676*TS: so, in other word, what two types of variance? ▶677*SS: um (.) the within group is on the denominator. ▶678%act: TS writes equation on board679@Comment: pic022 (c28/image022.jpg)680*TS: 0 [= points to numerator of equation and looks at S]. ▶681*SS: that'd be the top part. ▶682@Comment: pic024 (c28/image024.jpg)683*SS: um +... ▶684%act: TS writes numerator of equation in picture twelve685*TS: ok. ▶686*TS: xxx ok, what this means is xxx one dependent measure, Ok. ▶687*TS: this is an estimate [= pointing at numerator] of how much these two688groups differ. ▶689*TS: xxx in other words, how much the mean of this group differs from690the mean of this group. ▶691*TS: 0 [= pointing on board out of camera range]. ▶692*SS: 0 [= nods head]. ▶693*TS: ok. ▶694*TS: the within groups measures what695[= pointing at denominator of equation]? ▶696*SS: (.) the difference among them? ▶697*SS: xxx would that be the grand mean? ▶698*TS: no. ▶699*TS: ok. ▶700*TS: grand mean refers to another xxx. ▶701*TS: that's an f_test. ▶702*TS: we're talking about a t_test here. ▶703*TS: ok. ▶704*TS: within groups variability is to determine how much variability705there is within the group, individual groups. ▶706*TS: in other words, xxx within the x@l one only how much variability is707there? ▶708*SS: ok. ▶709*SS: so your response to xxx study five hours the difference between710them. ▶711*TS: right. ▶712*TS: how much do they differ from each other. ▶713*TS: ok? ▶714*TS: right. ▶715*SS: right. ▶716*TS: got that? ▶717*SS: yes. ▶718*TS: ok. ▶719*TS: it's expressed in terms, in other words, the x@l one people are720going to have a mean. ▶721*TS: you know, there's going to be a, an average of each of those five722people who studied five hours. ▶723*SS: right. ▶724*TS: how much does one of the people that studied five hours differ from725the mean of all of the people that studied five hours. ▶726*SS: that would just, like, x@l one minus x@l? ▶727*TS: well, in that instance it would be a sums of squares or a, I mean,728if you, I mean, it would be a xxx or whatever xxx divided by some729number and all that. ▶730*SS: ok. ▶731*TS: in other words, xxx answer that for sure go back to your statistics732xxx. ▶733*SS: yeah. ▶734*TS: standard deviation and then the variance and all that. ▶735*TS: ok, in other words xxx. ▶736*SS: 0 [= nods head]. ▶737*TS: and xxx. ▶738*SS: right. ▶739*TS: xxx how much is xxx? ▶740*SS: twenty two. ▶741*TS: xxx ok. ▶742*TS: alright. ▶743*TS: so a value of t, just looking at it logically as well as744mathematically, a value of t@l is going to increase as the745difference between the two groups increases and increase as the746difference within your groups decreases. ▶747*SS: ok, the difference within the groups decreases, the estimate of748t@l. ▶749*TS: t_value. ▶750*SS: will increase? ▶751*TS: because see, I mean, like, alright, the difference between groups,752say, is, um, like five. ▶753*TS: say the difference between them xxx, Ok? ▶754*TS: the difference between them is five. ▶755*TS: ok? ▶756*TS: and this is just looking at it algebraically, Ok? ▶757*SS: 0 [= nods head]. ▶758*TS: and our within groups, say we had xxx two. ▶759%act: TS writes fraction on board760@Comment: pic026 (c28/image026.jpg)761*TS: well, if you, well let me xxx [= changes 2 to 3 in equation]. ▶762@Comment: pic028 (c28/image028.jpg)763*TS: well, five divided by three is going to be less than five divided764by xxx. ▶765%act: TS writes equation on board, partially illegible766*SS: right. ▶767*TS: ok? ▶768@Comment: pic028 (c28/image028.jpg)769*TS: I mean, it's going to be greater, excuse me, than five divided by770three. ▶771*TS: by the same token certainly, ten fifteenths or something is less772than thirteen fifteenths or something. ▶773%act: TS writes equation on board774*TS: as the top part gets bigger, then your ratio gets bigger. ▶775*TS: as the bottom part gets smaller, your ration gets bigger. ▶776*SS: ok. ▶777*TS: so, in other words, uh, the difference between, uh, you know, in778terms of this example here, obviously, what you're trying to do is779look at the chances, the probability, that your results could have780occurred by chance or if they reflect the true difference. ▶781*TS: right? ▶782*SS: right. ▶783*TS: ok. ▶784*TS: well, if the difference between the means is so large that the785chances of that result just, just having occurred out of, by chance,786out of random error, that's a pretty small probability. ▶787*TS: then, that's when we're glad to reject the null. ▶788*TS: and so that, if I'm interested, let's say our, our, our mean here,789our mean is for five hours of study is seventy five and our mean for790ten hours of study is seventy eight, you know, we're not going to791all that sure that that's a real difference. ▶792*SS: uh huh. ▶793*TS: sure this is a little bit greater. ▶794*SS: 0 [= nods head]. ▶795*SS: right. ▶796*TS: but that could just be due to random error. ▶797*SS: 0 [= nods head]. ▶798*SS: right. ▶799*TS: but, if say, these people did, like, fifty and these people were800ninety nine, in other words, what happens to our between groups801difference? ▶802*TS: bigger and bigger. ▶803*TS: then we're going to be “hey, you know, that looks a lot less like804that occurred by chance”. ▶805*SS: right. ▶806*TS: than if they were, you know, seventy five and seventy eight. ▶807*TS: you see? ▶808*SS: uh huh. ▶809*TS: so our probability, our chances, of looking xxx, so when we xxx810we're talking about how likely is it the obtained result occurred,811you know, how, how, likely is it that that could occur by chance. ▶812*TS: I mean, if you look at that, you think, well that is such a big813difference. ▶814*TS: that doesn't look like it could have occur by chance. ▶815*SS: uh huh. ▶816*TS: so when the chances, the probability, that something could have817occurred by chance get so low, in other words, this doesn't look818like xxx occurred by chance. ▶819*SS: uh huh. ▶820*TS: that is xxx reject the null xxx. ▶821*TS: ok? ▶822*SS: 0 [= nods head]. ▶823*TS: so, when our between groups ratio gets bigger, our t_value gets824bigger. ▶825*TS: and so, xxx much more often, in other words, this would be a t of826those values and you would compare it to a critical table. ▶827*SS: uh huh. ▶828*TS: xxx. ▶829*TS: if your obtained value equal or exceeds that tabled value. ▶830*TS: then if it did so you're saying, it's not just by chance that this831occurred by random error. ▶832*SS: xxx reject the null. ▶833*TS: right. ▶834*TS: ok. ▶835*TS: ok. ▶836*TS: and the same thing happens if your within groups get smaller or837your between groups gets larger the t_value gets larger, so. ▶838*SS: 0 [= nods head]. ▶839*TS: now, the f_value [= checks watc] ok, the f_value, do you want me to840keep going, I mean? ▶841*TS: xxx, but I would be glad to explain the f_value to you. ▶842*SS: um, xxx finish up. ▶843*TS: xxx. ▶844*SS: no. ▶845*SS: 0 [= points at book] finish up. ▶846*TS: oh. ▶847*TS: finish up. ▶848*TS: ok. ▶849*TS: sure. ▶850*SS: if you've got time. ▶851*TS: no, that's fine. ▶852*TS: I mean, I've got time. ▶853*TS: ok. ▶854*SS: xxx. ▶855*TS: ok. ▶856*TS: we'll run through the f_value real fast and xxx. ▶857*TS: 0 [= erases board] xxx t@l was the ratio of our858[= writes "t =" on board and waits for S to tell her the equation]859tell me again. ▶860*SS: oh. ▶861*SS: between groups. ▶862*TS: between and within. ▶863%act: TS writes equation on board864@Comment: pic032 (c28/image032.jpg)865*TS: ok. ▶866@Comment: pic034 (c28/image034.jpg)867*TS: that applies when you have two groups. ▶868*TS: ok, the mean of one group that had five hours of study and the mean869of another group that had ten hours of study. ▶870%act: TS writes means on board871*TS: well, that if we had a group of xxx fifteen hours of study. ▶872%act: TS adds third mean to list873@Comment: pic036 (c28/image036.jpg)874*SS: xxx. ▶875*TS: the f_test. ▶876*TS: the f_test is for two or more. ▶877*TS: xxx, ok? ▶878*TS: xxx on top. ▶879*TS: this is equivalent to xxx between groups. ▶880*SS: 0 [= nods head]. ▶881%act: TS writing F equation on board, but the words are illegible882*TS: xxx. ▶883*TS: they call it error variance I think. ▶884*TS: 0 [= looks in book] xxx. ▶885%act: finishes writing F equation on board, illegible886*TS: ok. ▶887*TS: xxx [= waits for SS to answer, but he doesn't]. ▶888*TS: x one. ▶889*SS: and X two. ▶890*TS: and X two. ▶891*TS: ok. ▶892*TS: now we have xxx for three. ▶893*TS: so what is our systematic or between groups variance going to look894at now? ▶895*SS: xxx. ▶896*TS: yeah. ▶897*TS: varialbity between all of them ok. ▶898*TS: so in order to look at the variability in two or more groups, the899only way, the best way to do that, is to get a grand mean. ▶900*TS: xxx grand mean, the mean of all the subjects. ▶901*TS: ok? ▶902*SS: 0 [= nods head]. ▶903*TS: alright. ▶904*TS: and our between groups we're interested in how much this one905differs, how much did all the five hour study groups differ from906everyone, how much the ten hours, and how much907[= points to fifteen hour mean]. ▶908*TS: it's the same thing. ▶909*TS: it's how much each, each group varies from all the groups. ▶910*SS: 0 [= nods head]. ▶911*TS: ok? ▶912*TS: in other words, it's the same thing. ▶913*TS: but before we only had two groups, now we have three. ▶914*TS: xxx. ▶915*SS: alright. ▶916*TS: within groups you can put [= waits for S to answer]. ▶917*SS: the error, the number, the difference xxx that's within a single918group. ▶919*TS: in other words, what are you referring to? ▶920*TS: all the people that study five hours. ▶921*SS: five hours. ▶922*TS: might have been different from what? ▶923*SS: from the mean xxx. ▶924*TS: ok. ▶925*TS: in other words, how much one person who studied five hours differs926from everybody else who studied five hours. ▶927*SS: five hours. ▶928*SS: and for the ten and for the fifteen. ▶929*TS: ok. ▶930*TS: so the f_test works on the same logic. ▶931*TS: xxx, get your critical values, xxx, you get your observed value and932compare it against your critical value. ▶933*TS: you have to do one more thing for both of them, and that is degrees934of freedom. ▶935*SS: 0 [= nods head]. ▶936*SS: oh. ▶937*TS: alright. ▶938*TS: do you remember what those are? ▶939*SS: that's the, um (.) definitionally it's the error, the number xxx. ▶940*SS: you have six occurrences and you have five degrees of freedom. ▶941*SS: xxx that one occurrence xxx. ▶942*TS: &=laughs ok. ▶943*SS: &=laughs is that? ▶944*TS: the example, I think you, the one in the book on page one hundred945and forty. ▶946@Comment: pic038 (c28/image038.jpg)947*TS: ok. ▶948*TS: the degrees of freedom, ok. ▶949*TS: the degrees of freedom of the number of scores for each of xxx950once. ▶951*TS: xxx the means are know, Ok? ▶952*TS: so, it's how many scores can vary once you know something about953what's going on, Ok? ▶954*TS: how it's, for the example. ▶955*TS: if the mean of the group is six, and there are five subjects in the956group, alright. ▶957*TS: so your mean is six Ok? ▶958%act: TS writes mean on board959*TS: and you have five subjects, Ok. ▶960*TS: um, xxx you're getting tutored, it's gonna be on the tape. ▶961*TS: do you have a calculator? ▶962*TS: I meant to bring one. ▶963*SS: 0 [= gets out calculator]. ▶964*TS: we can look at this mathematically, as well as logically. ▶965@Comment: pic040 (c28/image040.jpg)966*TS: ok. ▶967*TS: we have five subjects in a group. ▶968*TS: n@l equals five. ▶969%act: TS writes N value on board970*TS: ok? ▶971*SS: ok. ▶972*TS: alright. ▶973@Comment: pic042 (c28/image042.jpg)974*TS: and there, say you have four scores. ▶975*TS: xxx logically. ▶976*TS: well, if we have four scores and we know that x@l one equals four,977and x@l two equals eight, and x@l three equals, let's say, thirty978four, and x@l four equals, say, a nine, Ok? ▶979%act: TS writes values for each x@l as she lists them980*TS: whatever x@l five is going to be has, Ok, these four values could981have been anything [= points to x@l one-four]. ▶982*TS: ok? ▶983*TS: but once we know these four, x@l five has to be fixed. ▶984*TS: because it has to be whatever will add into all the rest of these985to keep the mean at six. ▶986*SS: 0 [= nods head] Oh. ▶987*SS: ok. ▶988*TS: xxx. ▶989*TS: in other words, we know that the mean is six already, we know that990we have five subjects. ▶991*TS: once we know any three, two of them could still be anything. ▶992*TS: but once you know four, there's only one left and it has to be993whatever will be added in and divided to make the mean six. ▶994*TS: ok? ▶995*TS: so that's you're degrees of freedom. ▶996*TS: ok? ▶997*TS: calculator xxx. ▶998*SS: 0 [= puts calculator away]. ▶999*TS: ok. ▶1000*TS: so [= reading from book] it's how many scores xxx. ▶1001*TS: ok. ▶1002*TS: got that? ▶1003*TS: alright (.) So we were looking at (.) the f_test xxx. ▶1004*SS: uh huh. ▶1005*TS: xxx. ▶1006*TS: um, so just kind of looking at it in general, these tests. ▶1007*TS: our goal is to determine if our obtained results are reliable1008according to xxx calculation xxx. ▶1009*TS: xxx what's used before hand, how confident we can be with our1010obtained results. ▶1011*TS: we are more likely to obtain significant results in our t@l or our1012f_test if the difference between the groups is large, xxx small. ▶1013*SS: 0 [= nods head]. ▶1014*TS: xxx. ▶1015*TS: so that's where the xxx. ▶1016*TS: but, xxx. ▶1017*TS: ok, do you have any other questions that you wan to ask about this?1018▶1019*SS: um, I have one. ▶1020*TS: ok. ▶1021*SS: um, if the obtained value of t@l is larger than the critical value,1022do you reject the null on that? ▶1023*TS: right. ▶1024*TS: that's what you do. ▶1025*TS: the obtained value needs to be equal to or greater than the1026critical value, which comes out of the table. ▶1027*SS: right. ▶1028*TS: so you choose which value to look at by looking at what. ▶1029*SS: what was the question? ▶1030*TS: when your looking up the t_table, the critical values of t@l are on1031page two hundred and seventy xxx, here we go, two hundred and1032seventy five. ▶1033*TS: how do we know which value to look at? ▶1034*TS: by looking at our [= waits for SS to finish sentence]. ▶1035*TS: over there in the left column. ▶1036*SS: oh. ▶1037*SS: degrees of freedom. ▶1038*TS: right. ▶1039*TS: in other words, you'd have to look at that and go xxx. ▶1040*TS: you have to choose your significant level, which you do beforehand1041in our hypothesis we're testing, and our confidence level. ▶1042*SS: uh huh. ▶1043*TS: xxx, degree of freedom. ▶1044@Comment: pic044 (c28/image044.jpg)1045*TS: ok. ▶1046*TS: xxx critical value xxx. ▶1047*TS: your critical value is just a xxx. ▶1048*TS: type one and type two errors. ▶1049*TS: ok. ▶1050*TS: the best way is to look at it, I like the way the book xxx. ▶1051*TS: it's a good way xxx. ▶1052*TS: xxx [= draws grid on board]. ▶1053*TS: our null hypothesis. ▶1054@Comment: pic046 (c28/image046.jpg)1055*TS: and our alternate hypothesis. ▶1056%act: writes null and alternate hypothesis on board1057*TS: what are, in terms of the null, what are the two types decisions we1058can make? ▶1059*SS: accept it or reject it. ▶1060*TS: sure. ▶1061%act: labels left side of grid, illegible1062*TS: in terms of our population, we can xxx, and they are what? ▶1063*SS: um, xxx. ▶1064*TS: right. ▶1065*TS: either the null is true; or the null is not going to be true. ▶1066*TS: xxx it's all probabilistic xxx. ▶1067*TS: alright. ▶1068*TS: in, ah, our null hypothesis is that changing the hours you study1069doesn't have any significant effect on our test score. ▶1070*TS: alright? ▶1071*SS: 0 [= nods head]. ▶1072*TS: ok. ▶1073*TS: if we do our experiment, analyze our results, and because of our1074sample we are led to reject the null, it looks like this error. ▶1075*TS: we get a really big difference between our hours of study. ▶1076*TS: say, we had xxx. ▶1077*TS: say that the ten, five hour people got a mean of fifty five, the1078ten hour people got a mean of, say, ninety eight. ▶1079*TS: well, statistically we find that that is a significant difference1080and we think “Alright”! ▶1081*TS: you know, that our independent variable does have an effect, so we1082reject the null. ▶1083*TS: however, what if in some theoretical sense that this was just1084chance. ▶1085*TS: that these results occurred by chance, and not really due the1086workings of our experiment, Ok? ▶1087*TS: we've rejected the null, but the null is actually true. ▶1088*TS: that would be a [= waits for S to respond]. ▶1089*TS: what type of error? ▶1090*TS: which one? ▶1091*SS: type I. ▶1092*TS: ok. ▶1093%act: writes in square on grid.1094*TS: ok. ▶1095*TS: however, if we had, in that case, when we talked about the null1096being true, Ok, our results were like this1097[= points to mean scores on board] and looked like there was such a1098big difference. ▶1099*TS: it could have been xxx. ▶1100*TS: xxx rejected the null with xxx. ▶1101*TS: well, what if we had gotten that and, and because of our sample1102size or because of our figures or whatever, we were led to fail to1103reject the null and the null was true, what do we have then? ▶1104*SS: 0 [= doesn't answer]. ▶1105*TS: we failed to reject the null and the null was, indeed true. ▶1106*TS: what have we then? ▶1107*SS: xxx. ▶1108*TS: no. ▶1109*TS: if the null is true and we don't reject it. ▶1110*SS: oh, Ok. ▶1111*SS: accept it. ▶1112*TS: in other words, we. ▶1113*TS: sure. ▶1114*TS: we've made a correct decision. ▶1115*SS: right. ▶1116*SS: xxx. ▶1117*TS: no, you're not. ▶1118%act: writes in square on grid, illegible1119*TS: ok. ▶1120*TS: in other words, let's say, let's jus hypothetically say, we got a1121seventy and a seventy six. ▶1122*SS: right. ▶1123*TS: and due to the statistical analysis we found that that is not a1124significant difference, in other words, we do not reject the null. ▶1125*TS: in fact, in our population, we have some way of knowing that t1126makes absolute sense, our null is true and we xxx our decision. ▶1127*TS: ok, what if we reject the null, in other words, we xxx difference,1128Oh fifty five and ninety six for example, but we reject the null and1129it is actually true that the independent variable does have an1130effect, then what do we have? ▶1131*SS: xxx correct decision. ▶1132*TS: another correct decision. ▶1133%act: writes in square on grid, illegible1134*TS: in other words, we've rejected the null and our hypothesis is true.1135▶1136*TS: i@l e@l our independent variable has an effect. ▶1137*TS: however, if we don't reject the null, we fail to reject the null,1138but the null, the research hypothesis is true, in other words, the1139null is false, what do we have? ▶1140*SS: xxx. ▶1141*TS: ok. ▶1142%act: writes in square on grid, illegible1143*TS: and in terms of this example, what does that mean? ▶1144*TS: what is happening? ▶1145*SS: that there is, um, there is a difference between the amount of1146hours studied and xxx there is no difference. ▶1147*TS: perfect. ▶1148*TS: right. ▶1149*TS: xxx there was no difference xxx. ▶1150*TS: that's a type two error. ▶1151*TS: ok. ▶1152*TS: the Type I error can be specified exactly. ▶1153*TS: xxx the chances xxx. ▶1154*SS: xxx probability? ▶1155*TS: right. ▶1156*TS: by an alpha level. ▶1157*TS: xxx our chances xxx type of error. ▶1158*TS: in other words, xxx point o@l five we have a five percent chance of1159it being a type one error. ▶1160*TS: alright? ▶1161*SS: 0 [= nods head]. ▶1162*TS: but you can't specify a type two precisely because it's specified1163in terms, I mean, it's kind of defined in terms of xxx. ▶1164*TS: in other words, xxx one error decreases it's going to harder and1165harder to reject the null. ▶1166*TS: xxx type two error. ▶1167*SS: xxx. ▶1168*TS: right. ▶1169*TS: they increased. ▶1170*TS: and logically you can see that because what you do with it when1171you're trying to make the chances of type I error smaller, what do1172you do? ▶1173*TS: you make it harder and harder to reject the null. ▶1174*SS: 0 [= nods head]. ▶1175*TS: well, obviously, if you've making it harder and harder to reject1176the null, then you're making it harder and harder to reject the1177null. ▶1178*TS: in other words, giving you a much greater chance that you won't1179reject it even if there was a difference xxx. ▶1180*SS: 0 [= nods head and mumbles]. ▶1181*TS: in other words, you see some difference but it's not great enough1182xxx. ▶1183*TS: type one chances increase type two chances decrease. ▶1184*TS: um, what, Ok, let's just do some, look at some more examples. ▶1185*TS: ok? ▶1186*TS: what if I study the effects of sleep deprivation on teaching1187performance, Ok? ▶1188*TS: one group of teachers has had twenty four hours of sleep1189deprivation, the other, say has had forty. ▶1190*TS: 0 [= reads from paper] analyzed my results and based on these1191concluded that the amount of deprivation does make a difference in1192teaching performance. ▶1193*TS: however, in reality the amount of deprivation xxx doesn't cause a1194difference. ▶1195*TS: what type of error is that? ▶1196*SS: um (.) you've rejected the, um, H one. ▶1197*TS: right. ▶1198*SS: xxx. ▶1199*TS: no. ▶1200*TS: look. ▶1201*TS: that failure to xxx rejected the null, but in reality. ▶1202*SS: oh. ▶1203*SS: degrees of freedom. ▶1204*TS: xxx draw a matrix and you can't go wrong. ▶1205*SS: draw a matrix. ▶1206*TS: ok. ▶1207*TS: I said there was a difference, xxx H one was true, but the null was1208actually true. ▶1209*TS: so I've rejected the null, and the null was actually true. ▶1210*TS: in other words, H one is true xxx rejecting xxx. ▶1211%act: writes on board near grid, illegible.1212*TS: xxx. ▶1213*SS: 0 [= nods head]. ▶1214*TS: in other words, what I did was to set the alpha high for this xxx1215the null was true, type one. ▶1216*TS: ok [= shrugs shoulders]. ▶1217*TS: if I, well, let's xxx, let's just consider more serious xxx which1218type of error. ▶1219*TS: type one or type two? ▶1220*SS: type one. ▶1221*TS: ok. ▶1222*TS: why? ▶1223*SS: xxx (.) xxx if you make a type one error, then you're saying that1224(.) an independent variable (.) is causing something that it's not1225causing and xxx. ▶1226*TS: 0 [= nods head]. ▶1227*TS: right. ▶1228*TS: in other words, xxx does something xxx. ▶1229*SS: xxx. ▶1230*TS: and a type one error is xxx something didn't have an effect when it1231actually did. ▶1232*SS: 0 [= nods head] Uh huh. ▶1233*TS: xxx that would be less serious because we could just run another1234study, somebody could just run another study to see if you were1235right or wrong. ▶1236*SS: 0 [= nods head]. ▶1237*TS: and, and xxx, well, the same thing could be true if you made a type1238one errors xxx, but the consequences are much less serious. ▶1239*SS: 0 [= nods head]. ▶1240*TS: you know, 'cuz you could say xxx type two error. ▶1241*TS: xxx. ▶1242*TS: xxx well gosh my study didn't show it but xxx run another study,1243xxx study or xxx accurate picture. ▶1244*SS: 0 [= nods head]. ▶1245*TS: in a xxx study xxx method xxx consistered less serious. ▶1246*TS: um, the book example on page one hundred and forty six and forty1247seven was about the surgeon, right? ▶1248*SS: 0 [= looks in book, nods head]. ▶1249*TS: it would be much xxx for him to make a type two error, in other1250words, if you did choose not to operate and it prevents xxx a lot1251more serious than if you did operate, right? ▶1252*SS: 0 [= nods head]. ▶1253*TS: &=laughs they'd both be serious, but xxx might die xxx if you did1254operate and the operation was serious. ▶1255*SS: xxx. ▶1256*TS: uh huh. ▶1257*SS: xxx class xxx type one or type two errors. ▶1258*TS: well, don't worry about it, I mean, provided that you remember to1259set your matrix up right. ▶1260*TS: you know, H@l one's true, H@l o@l is true, H@l one is true, xxx1261[= points along top and side of matrix as she talks]. ▶1262*TS: xxx the null or xxx. ▶1263*SS: 0 [= nods head] Ok. ▶1264*TS: you know, it's like, as long as you remember, memorize how to set1265it up, then you can use it xxx. ▶1266*SS: 0 [= nods head] Ok. ▶1267*TS: alright. ▶1268*TS: ok. ▶1269*TS: u, the whole reason why the type two error is xxx also xxx1270interpreting non_significant results. ▶1271*TS: they're kind of problem, they're problematic because you're looking1272for a relationships xxx relationships xxx relationship, but you can1273always try another study so you can look at a relationship. ▶1274*TS: but, that's xxx not necessarily xxx to find a relationship. ▶1275*SS: 0 [= nods head]. ▶1276*TS: you can find one. ▶1277@Comment: video ends1278*TS: xxx.1279*SS: um.1280*TS: xxx another question?1281*TS: xxx.1282*SS: no.1283*TS: ok.1284*SS: appreciate it.1285*TS: oh, no problem.1286*TS: it's what I'm here for.1287@End