Re: Splus or R
Anonymous God-fearer [EMAIL PROTECTED] wrote: Does anyone know how to generate a correlation matrix given a covariance matrix in Splus? Or could you give the details of how to do it in another language? corr[i,j] = cov[i,j]/sqrt(cov[i,i]*cov[j,j]) = Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at http://jse.stat.ncsu.edu/ =
Re: Applied analysis question
Rolf Dalin [EMAIL PROTECTED] wrote: Brad Anderson wrote: I have a continuous response variable that ranges from 0 to 750. I only have 90 observations and 26 are at the lower limit of 0, What if you treated the information collected by that variable as really two variables, one categorical variable indicating zero or non-zero value. Then the remaining numerical variable could only be analyzed conditionally on the category was non-zero. In many cases when you collect data on consumers consumption of some commodity, you would end up in a big number of them not using the product at all, while those who used the product would consume different amounts. IIRC, your example is exactly the sort of situation for which Tobit modelling was invented. = Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at http://jse.stat.ncsu.edu/ =
Re: Evaluating students
John Kane [EMAIL PROTECTED] wrote: Dennis Roberts wrote: At 08:56 AM 11/16/01 -0700, Roy St Laurent wrote: It's not clear to me whether recent posters are serious about these examples, but I will reiterate my previous post: For most mathematics / statistics examinations, the answer to a question is the *process* by which the student obtains the incidental final number or result. The result itself is most often just not that important to evaluating students' understanding or knowledge of the subject. And therefore an unsupported or lucky answer is worth nothing. the problems with the above are twofold: 1. this assumes that correct answers are NOT important ... (which believe me if you are getting change from a cashier, etc. etc. ... ARE ... we just cannot say that knowing the process but not being able to come up with a correct answer ... = good performance) No it doesn't. It doesn't assume that they're not necessary, merely that they're not sufficient. To take it to the extreme, would you say a student should get at least partial credit for a correct answer if he arrived at it by copying it from another student's paper? After all, he/she *did* answer correctly. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Kappa negative, doubt?
Ivan Balducci [EMAIL PROTECTED] wrote: I would like to understand the meaning of negative Cohen´s Kappa value (1960). I got it from the following data (from Dental Radiology): Magnetic resonance versusFacial Pain Total Yes No Yes 14 5 No 11 3041 Total 12 34 46 Kappa = - 0,0422 My concern is: are negative Kappa values less important than random agreement or they mean something else? They mean that the agreement is actually worse than chance. Note that the odds ratio is .68, which is a significant, though weak (Yule's Q=.19) *negative* association. You can't expect any agreement under those circumstances. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Boston Globe: MCAS results show weakness in teens' grasp of
Dennis Roberts [EMAIL PROTECTED] wrote: if we take the infamous #39 item ... where the options were (if i recall)... A. mean only B. median only C. range and mean D. range and median well, even if we accepted this item as fair ... a student looks at the graph ... sees that there is a bunch of stuff in the boxplot ... and then sees A and B ... without too much thought ... they could say, it can't be mean or median ONLY ... there must be more you can do with that boxplot than that ... so, without knowing anything of consequence about boxplots ... you are down to a two choice item thus, for even the uninformed ... we have essentially a T and F item but, if you did happen to KNOW that the boxplot uses the median ... then the fact that the RANGE is part of the C and D options ... means the term RANGE is irrelevant to the item ... given that A and B are just too restrictive (because of the use of the term ONLY) and C and D have RANGE as a common element ... so, if you did pick D ... then it has to be because you knew that isolated fact ... which certainly is about as low level as you can get And furthermore, not all the wrong answers are equally bad. Someone who would answer A or B must know quite a bit less than someone who would answer C (in fact, it would tend to indicate that they had no concept at all of what the boxplot represented). = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: adjusted r-square
In sci.stat.consult Graeme Byrne [EMAIL PROTECTED] wrote: In short, you don't. If the number of terms in the model equals the number of observations you have much bigger problems than not being able to compute adjusted R^2. It should always be the case that the number of observations exceed the number of terms in the model otherwise you cannot calculate any of the standard regression diagnostics (F-stats, t-stats etc). My advice is get more data or remove terms from the model. If neither of these is an option you are stuck. It's worse than not being able to calculate regression diagnostics. You can't make *any* inferences beyond your observed data. Consider the degenerate case of trying to fit a bivariate regression line when you have only two observations. You'll *always* get a perfect fit because two points mathematically define a line. But that perfect fit will tell you absolutely nothing about the underlying relationship between the two variables. It's consistent with *any* possible relationship, including complete independence. You can't tell how far off your model is from the observations because there simply isn't any room for it to be off (room for the model to be off otherwise being known as degrees of freedom). A model with as many parameters as observations is equivalent to the observations themselves, and therefore testing such a model against the observations is the same thing as asking if the observations are equal to themselves, which is circular reasoning. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: independent, identically distributed
Philip Ackerman [EMAIL PROTECTED] wrote: Hello, I have a question regarding the term independent, identically distributed random variables. For example, suppose that your random variable is height, and the population of interest is adult males. You take a sample (size n) of adult males, and measure the variable for each subject. Then, if I understand correctly, you have n individual random variables, i.e. X1, X2, ..., Xn, which represent the one population random variable X. What I do not understand, is why an individual's height is a random variable (in addition to height in the population being a random variable), and henceforth, has the same distribution as the population random variable. Is this because you would get a different height for the same person if you took measurements at different times? Does the distribution of height for each individual consist of all hypothetical measurements you could take? I know that the population distribution of a random variable is also the probability distribution when you select one individual randomly from the population. Does this have anything to do with iid? Any help would be greatly appreciated. A random variable is, by definition, any function whose domain is a probability space. Both the height of an individual and the average height of a population are the values of such functions. The *potential* heights an individual could have form a probability space; the *actual* height of an individual is the value of a function defined on this space. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Do you known?
In sci.stat.edu Monica De Stefani [EMAIL PROTECTED] wrote: Hi, is there anybody known Quade, D. (1976) . Nonparametric partial correlation, Measurement in the social Sciences. Edited by H.M. Blalock, Jr. Aldine Publishing Company: Chicago, 369-398? I would known how he calcolate Kendall's partial tau (presisely), please. Kendall's partial tau is calculated exactly the same way as Pearson's partial r. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Normality in Factor Analysis
In sci.stat.consult haytham siala [EMAIL PROTECTED] wrote: I have a question regarding factor analysis: Is normality an important precondition for using factor analysis? It's necessary for testing hypotheses about factors extracted by Joreskog's maximum-likelihood method. Otherwise, no. If no, are there any books that justify this. Any book on factor analysis or multivariate statistics in general. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Correction procedure
Bekir [EMAIL PROTECTED] wrote: 2. There are apparently and exactly three groups; groups 1, 3, 5 that had the same proportions of translocation. Therefore, to compare only the group 2 and 3 with the control can be appropriate, can it be? Thus, there would be two comparisons and the p valus 0.008 (0.008 x 2 =0.016) and 0.02 (0.02 x 2 = 0.04)would be significant. Is it right? No, that's not appropriate, because the fact that groups 3 and 5 had the same proportions is a contingency of your experimental results, not something that is _a priori_ true, and therefore the (zero) difference in proportion between them is subject to as much error as any of your other results. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: The False Placebo Effect
Rich Ulrich [EMAIL PROTECTED] wrote: On 26 May 2001 03:50:32 GMT, Elliot Cramer [EMAIL PROTECTED] wrote: I don't see how RTM can explain the average change in a prepost design - explanation: whole experiment is conducted on patients who are at their *worst* because the flare-up is what sent them to a doctor. Sorry; I might have been more complete there. All the pre-post studies in psychiatric intervention (where I work) have this as something to watch for. That's why you see so many multiple sclerosis patients in the testimonials of quacks; MS is very much an exacerbation/remission disease, and it's very easy for the patient to attribute a remission to an ineffective treatment taken during an exacerbation. You never see testimonials saying that a treatment taken during a remission lowered the frequency of exacerbations. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: errors in journal articles
Warren Sarle [EMAIL PROTECTED] wrote: Telling the Truth About Damned Lies and Statistics By JOEL BEST [snip] So the prospectus began with this (carefully footnoted) quotation: Every year since 1950, the number of American children gunned down has doubled. I had been invited to serve on the student's dissertation committee. When I read the quotation, I assumed the student had made an error in copying it. I went to the library and looked up the article the student had cited. There, in the journal's 1995 volume, was exactly the same sentence: Every year since 1950, the number of American children gunned down has doubled. This quotation is my nomination for a dubious distinction: I think it may be the worst -- that is, the most inaccurate -- social statistic ever. I'm reminded of the bragging Scientologist who claimed that his particular org had at least quadrupled its membership every year since 1954. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Help me an idiot
W. D. Allen Sr. [EMAIL PROTECTED] wrote: Five different condiments, plus no condiments, means 6*5*4*3*2*1 = 720 distinct combinations. I wanted that with *fries* and *ketchup*! *Not* ketchup and fries! = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Student's t vs. z tests
Mark W. Humphries [EMAIL PROTECTED] wrote: Hi, I am attempting to self-study basic multivariate statistics using Kachigan's "Statistical Analysis" (which I find excellent btw). Perhaps someone would be kind enough to clarify a point for me: If I understand correctly the t test, since it takes into account degrees of freedom, is applicable whatever the sample size might be, and has no drawbacks that I could find compared to the z test. Have I misunderstood something? You're running into a historical artifact: in pre-computer days, using the normal distribution rather than the t distribution reduced the size of the tables you had to work with. Nowadays, a computer can compute a t probability just as easily as a z probability, so unless you're in the rare situation Karl mentioned, there's no reason not to use a t test. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: rotations and PCA
Rich Ulrich [EMAIL PROTECTED] wrote: - Intelligence, figuring what it might be, and categorizing it, and measuring it... I like the topics, so I have to post more. On Thu, 05 Apr 2001 22:09:33 +0100, Colin Cooper [EMAIL PROTECTED] wrote: (2) Gould's implication is that since Spearman found one factor (general ability) whilst Thurstone fornd about 9 identifiable factors, then factor analysis is a method of dubious use, since it seems to generate contradictory models. There are several crucial differences - I read Gould as being more subtle than that. Definitely. Gould's point was that since the various forms of factor analysis generate models that are consistent with multiple, and contradictory, explanations of intelligence, and these models all resolve the same amount of variance in the raw data, none of the models provide a basis for favoring one theory over the other. Specifically, if a model is consistent with both a theory that says that intelligence is (hereditary, immutable) *and* with a theory that says that intelligence is (environmental, mutable) than that model does not provide any support for a claim that the first theory better explains reality than the second theory (or vice versa, for that matter). In science, it's not enough to say that you have data that's consistent with your hypothesis; you also need to show a) that you don't have data that's inconsistent with your hypothesis and b) that your data is *not* consistent with competing hypotheses. And there's absolutely nothing controversial about that last sentence, except to extreme post-modernists who would claim that it's an inherently white male view of the world, i.e. people who are far, far, to the left of most of Gould's critics. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: rotations and PCA
Rich Ulrich [EMAIL PROTECTED] wrote: On Thu, 29 Mar 2001 10:17:09 +0200, "Nicolas Voirin" [EMAIL PROTECTED] wrote: OK, thanks. In fact, it's a "visual" method to see a set of points with the better view (maximum of variance). It's like to swivel a cube around to see all of its sides ... but this in more than 3D. When I show points in differents planes (F1-F2, F2-F3, F2-F4 ... for example), I make rotations, isn't it ? I think I would use the term, "projection" onto specific planes, if you are denoting x,y, and z (for instance) with F1, F2, F3 : You can look at the projection onto the x-y plane, the y-z plane, and so on. Here is an example in 2 dimensions, which suggests a simplified version of an old controversy about 'intelligence'-- tests might provide two scores of Math=110, Verbal= 90. However, the abilities can be reported, with no loss of detail, as General= 100, M-versus-V= +20. Historically, Spearman wanted us all to conclude that "Spearman's g" had to exist as a mental entity, since its statistical description could be reliably produced. And Thurstone's dissatisfaction with Spearman's theories led him to invent the technique of axial rotation, particularly the varimax criterion (rotate the axes until the variance of the items' projections on the axes is maximized). See Stephen Jay Gould's _The Mismeasure of Man_ for more details; note that Thurstone adopted varimax rotations because their results were consistent with *his* pet theories about intelligence. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Determing Best Performer (fwd)
Bob Hayden [EMAIL PROTECTED] wrote: The most cost-effective method is to roll a die. (See Deming.) If the representative's performances all lie within the control limits for the process that's being measured, yes. If one or more representatives' performances lie outside the system on the "good" side, then Deming would have had no problem with rewarding them (though he'd have emphasized that the most important thing for management to do would be to find out why they were doing so well and whether they were using any techniques that could be adopted by the others). = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: MA MCAS statistical fallacy
Robert J. MacG. Dawson [EMAIL PROTECTED] wrote: Therefore, I would not expect regression to the mean to be sufficient to explain the observed outcome (in which "practically no" top schools met expectations); and I conclude that the goals may well have been otherwise unreasonable. Indeed, requiring every school to improve its average by at least two points every year is not maintainable in the long run, and only justifiable in the short term if there is reason to believe that *all* schools are underperforming. This is the second time in the last two months that I've said to myself "I wish W. Edwards Deming were still alive." He always railed against arbitrary numerical performance goals that were set with no understanding of the system that produced the results and no specific plan for changing the system to produce better results. He'd probably be quoting Lloyd Nelson's quip about such goals, to the effect that if you say that you want to increase, say, revenues by 5% this year and you don't plan to change the system, then you're admitting that you've been slacking because if the current system allowed for it, you should already have done it. (The other context in which I thought of Deming was the election results. Deming insisted that it was meaningless to talk about the "true value" of some quantity in the absence of an operational definition of how to measure it. In this context, I'm sure he would have insisted that it was nonsense to assert that this method or that method of counting disputed votes gave a closer answer to the "true vote count" because the latter doesn't even exist until you specify a particular method of evaluating ambiguous ballots.) = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: NY Times on statisticians' view of election
In sci.stat.edu Ronald Bloom [EMAIL PROTECTED] wrote: So far, NOT ONE person here has responded to my point that the likelihood of getting into a tangle of some sort with a machine or mechanical procedure of some kind does not necessarily have anything to do with one's level of literacy! (And that illiterate persons may function Quite Well with mechanical devices, c) OK, one person has :) IMHO, the efforts to describe difficulty with the Palm Beach ballot as the results of "illiteracy" or "stupidity" are mostly simple propaganda tricks: in American culture, illiteracy and lack of intelligence carry a high degree of social stigma, whereas difficulty with machines, outside of geek circles, doesn't (and in some cases is treated as a matter of perverse pride). If you're a strong partisan, you want to attach socially stigmatized labels to your opponent's supporters. OTOH, it's also possible to use such labels almost unconsciously. Most Usenet users are probably far more adept with human-machine interfaces than most people, and it's easy for us to take a lot for granted. We tend to surround ourselves with people of similar adeptitude and fall into the assumption that we're typical and that there must be something wrong with people who aren't as adept as we are. In a way, this is just a form of fundamental attribution error. Finally, I think there's a cognitive dissonance phenomenon going on here: most of us *want* to believe that our system of elections is fair, and when presented with evidence that *might* indicate that this isn't so, we're inclined to interpret it in such a way as to confirm that things are the way we want them. And American culture strongly favors explaining outcomes in terms of individual, rather than systemic or structural, phenomena; at first glance, the idea that Gore voters were stupid or that Jeb Bush directed a campaign of fraud in favor of his brother "make more sense" to us than the idea that there are inherent inaccuracies in the system of vote-counting. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Fwd: Butterfly ballots (fwd)
Ronald Bloom [EMAIL PROTECTED] wrote: Would the group of kids doing a post-hoc experiment be biased inasmuch as the nature of the problem at hand may have become common-knowledge by now; even among kids; and so one would be forewarned of the error-mode in question, and be much less likely to fall into that mode of error? Quite likely. At any rate, what inference am I being prompted to draw here? That the people who claimed to have been confused were either (a) ignoramuses or (b) changing their tune after the fact? Is there some more generous interpretation, (c), say? I suspect the average visual acuity of the fourth-graders was better than that of the Palm Beach voters. There are certain cognitive tasks that children do better on than adults; could this be one of them? Or could it be one of those tasks that younger adults do better on than older ones? Valid research questions here, and if the latter turns out to be true, it would certainly have legal implications for future ballots. Did the exercise actually involve punching a hole using the apparatus found in an actual voting booth, or did it involve filling in a circle? If, as I suspect, it was the latter, than I doubt the results are transferrable. There are plenty of user interfaces that seem to work in mockups, but fail in production because of seemingly minor differences (one of my pet examples is that a simulated thumbwheel control as often seen on MP3 player software or other programs with a "skin" interface requires an entirely different set of motor movements (and skills) than a real-life thumbwheel; to me at least, the motor movements involved in operating a simulated thumbwheel rather close resemble those involved in removing a child-resistant cap from a medicine bottle ("press and turn simultaneously")). = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Palm Beach Stats
Reg Jordan [EMAIL PROTECTED] wrote: It's interesting that no Republicans have claimed that the ballot was misleading -- all the complaints seem to come from Democrats. Wouldn't the "misleading, confusing" nature of the ballot apply equally across the voting spectrum? Bush's name and hole were both in the same position (the first position) which makes it quite a bit less likely that anyone intending to vote for him would be confused by the ballot. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: probability questions
[EMAIL PROTECTED] wrote: Two probability questions... If X has chi-square distribution with 5 degrees of freedom 1. what is the probability of X 3 Look that value up in a chi-square table and find out. 2. what is the probability of X 3 given that X 1.1 Look both values up in a table and then use your knowledge of conditional probability to compute the desired probability. If you're not in posession of such knowledge, acquire it (it's explained in any elementary treatment of probability). = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: .05 level of significance
dennis roberts [EMAIL PROTECTED] wrote: [regarding the "point biserial correlation"] and it certainly has nothing to do with a "shortcut" formula for calculating r ... it MAY have decades ago but it has not for the past 20 years ... While I certainly agree that many textbooks convey the absolutely misleading impression that the "PBC" is some special form of measure, I think that the usual formula presented for it is pedagogically useful in a few ways (not that the typical textbook makes use of them): 1) It demonstrates that a correlation problem in which one variable is dichotomous is equivalent to a two-group mean-difference problem. 2) It shows that in such a case, the correlation coefficient is a function of both a standard effect-size measure (Cohen's d) and the relative sizes of the two groups. 2a) It demonstrates that variations in the relative sizes of the group will result in variations in the magnitude of the correlation, even if the effect size is held constant. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: point biserial formula
dennis roberts [EMAIL PROTECTED] wrote: At 06:14 AM 10/21/00 +, Eric Bohlman wrote: 1) It demonstrates that a correlation problem in which one variable is dichotomous is equivalent to a two-group mean-difference problem. maybe you can make this point but, to a typical student ... i would say this equivalence would be lost As Karl pointed out, that represents a serious problem and it would be worth investigating *why* that is. Joe Ward, if you're reading this, would you care to comment on how your model-building approach could be used to help clear up this confusion? 2) It shows that in such a case, the correlation coefficient is a function of both a standard effect-size measure (Cohen's d) and the relative sizes of the two groups. great ... if you have talked about these things prior to a simple correlation coefficient ... how likely is that? I was thinking that this relationship would be introduced later on. 2a) It demonstrates that variations in the relative sizes of the group will result in variations in the magnitude of the correlation, even if the effect size is held constant. how is this the case? i don't see n's for the groups in this formula But I see p's and q's, which have implicit n's. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: How to select a distribution?
Herman Rubin [EMAIL PROTECTED] wrote: As we get more complex situations, like those happening in biology, and especially in the social sciences, it is necessary to consider that models may have substantial errors and still be "accepted", as one can only get some understanding by using models. "All models are wrong. Some models are useful." -- George Box I think what a lot of people forget (or never realized in the first place) is that a model is by definition an oversimplification of the state of nature. A model that fit perfectly would be of no use, as it would be just as complicated as the state of nature itself. As Stephen Jay Gould pointed out in his discussion of factor analysis in _The Mismeasure of Man_, when we build models we are *deliberately* throwing out *information* (not just "noise") in the hopes that we can deal conceptually with what remains. We really can't do otherwise simply because our brains aren't infinitely powerful. But we have to remember that that's what we're doing, and (again a major point of Gould's) disabuse ourselves of the notion that we're discovering something that's more real than the real world. Models are not Platonic ideals. They are conceptual shortcuts, heuristics if you will. They help us cope with uncertainty, but do not make it magically disappear. (I find phraseology like "this data was generated by that model" extremely offensive, as it subtly plays in to both the Platonic ideal notion and the postmodern notion that reality is purely a social or linguistic construct.) = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: memorizing formulas
Karl L. Wuensch [EMAIL PROTECTED] wrote: I think that Bob Hayden is on to something essential here ("I noted that Karl presented all the understandings he sought verbally on the list. Why not do the same in class?"). I think of the "definitional formulae" just as a convenient shorthand for the verbal definition of a construct. But it may be the case that most of our students assume that something that looks like a formula is just for use with mindless computations. They may have learned this in their first 12 years of schooling, where formulas may indeed be presented as nothing more than mindless recipes for getting some quantity not really well understood. How can we break our students of that bad habit? I do frequently verbalize the 'formula' after writing it on the board -- for variance, saying something like "look at this, we just take the sum of the squared deviations of scores from their mean, which measures how much the scores differ from one another, and then divide that sum by N, to get a measure of how much scores differ from one another, on average." The shorthand is really convenient, I don't know how I would get along without it. I have always thought that success in stats courses was much more a function of a student's verbal aptitude and ability to think analytically, rather than mathematical aptitude. Has anybody actually tested this hypothesis? I think the problem here is that, at least in the US K-12 system, "mathematical aptitude" really means "computational proficiency." In "back to basics" math (and the overwhelming majority of K-12 math curricula *are* "back to basics"; it's only in a few elite schools with high-performing students that any of those educational innovations that right-wingers claim are destroying the minds of our students have actually been used, but I digress; see Alfie Kohn's writings), mathematical notation is taught as a series of Taylorized (as in Frederick Winslow) job instructions rather than as a language for precisely describing certain types of relationships. IOW, it's taught as a bunch of hoops to be jumped through without really understanding what you're doing; all that matters is going through the prescribed motions. A student who is "good at math" is one who can jump through those hoops quickly. The object of most K-12 math is to turn out human calculators (something that actually made economic sense in the Old Days when a machine to do calculations cost a lot more than an employee doing calculations longhand). As John Allen Paulos pointed out in _Innumeracy_, the majority of American K-12 students know *how* to add, subtract, multiply, and divide, but a much smaller number of them know *when* to add, subtract, multiply, and divide. They do just fine when given a spelled-out list of calculations to perform, but they panic when confronted with word problems. Over in comp.infosystems.www.authoring.html, there's a long-running and acrimonious debate between those who see HTML markup as a *description* of the structure of a document and those who see it as a bunch of commands interspersed with text. The notion that formal notation can be a description or explanation rather than a sequence of tasks to perform is alien to a lot of people. It may very well be that some people simply don't have the built-in cognitive ability to grasp the former, but I suspect that a much larger number of people *could* develop that ability if only they were taught how to think that way (and it is indeed a way of thinking rather than a set of "skills"). Right now I suspect that most who developed that ability picked it up through intellectual exploration outside the formal system of schooling. I'm pretty sure I did (I was a math major but am not involved in teaching). = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: interaction effects
Mike Hewitt ([EMAIL PROTECTED]) wrote: : I am looking for assistance in interpreting results of a study. It : involved the testing of three different music practicing conditions. I : performed a GLM-repeated measures with three factors (modeling, : self-listening, self-evaluation) in addition to a repeated measure : (test). There was a significant interaction for test x modeling x : self-evaluation. There was also a significant result for test x : modeling. Does the higher-order interaction negate the results the : "main effects" or lower-order interaction? : : Specifically, musicians who listened to "model" performance improved : their performance more than those that did not listen to a model. : Great. For the interaction (test x modeling x self-evaluation), the : modeling/self-evaluation group improved more than did the no : modeling/self-evaluation group (reinforcing the results for modeling : only). HOWEVER, the same result did not occur for the groups that did : not self-evaluate. They improved similarly to each other. : So...listening to a model is more effective than not listening to one : when there is no-self-evaluation. : : What then does this mean as far as the results for the test x modeling : result? I guess my question is does a higher-level interaction : "overrule" a lower-level interaction? It depends on how you define "overrule." I'd say that your results indicate that there is indeed an interaction between test and modelling, but the amount of the interaction varies with self-evaluation. IOW, you've demonstrated that the test/modelling interaction *exists*, but any *estimate* of it is going to be poor because it will amount to an average of the results of two or more incomparable scenarios (i.e. it need not be close to the value of *any* observed two-way interaction for any given level of self-evaluation). = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: to frame or not frame
Robert Dawson ([EMAIL PROTECTED]) wrote: : Well, yes, there are; there is no easy way to pass on a reference : any more. It is aggravating when you want to send somebody the URL for : one page in a big site and it is a frame on a huge page, so that the : URL gets you only to the "home frame". Also, the same thing happens : with bookmarking. Further disadvantages of frames: 1) It's not possible for a browser to look at an individual content page and determine what frameset it belongs to. This means that if a user comes to a content page via a search engine that's indexed it, and all the site navigation is kept in frames of its own, there will be no way for the user to access the site navigation. This is commonly called the "orphaned frame" problem. Promotional sites often try to solve this problem by discouraging search engines from indexing anything but their home page, but this isn't appropriate for informational sites, as it will render the whole site opaque to most actual searches. At the very least, individual content frames need to have links back to the opening frameset. 2) Reduction of available screen real estate. The typical use of frames sacrifices content area in order to keep navigational information constantly in view, yet the user generally only needs to see navigation information at the very beginning of a content page (if a quick glance tells him he wants to go somewhere else) or at the end (when he's done reading). For a commercial promotional site, there may be some value in constantly keeping the company's name and logo within the user's visual field in order to build brand identity, but this is just annoying with an informational site. 3) Inaccessible content. If you make frames non-resizable, non-scrollable and of fixed width, some text or image content may get cut off if the viewer's font sizes are different from the designer's (remember that on different platforms, the same point size may correspond to different pixel sizes). 4) The usage of frames makes it difficult to print documents. Most of the problems with frames occur on browsers that *do* support them. === This list is open to everyone. Occasionally, less thoughtful people send inappropriate messages. Please DO NOT COMPLAIN TO THE POSTMASTER about these messages because the postmaster has no way of controlling them, and excessive complaints will result in termination of the list. For information about this list, including information about the problem of inappropriate messages and information about how to unsubscribe, please see the web page at http://jse.stat.ncsu.edu/ ===
Re: Sample size: way tooo big?
Robert McGrath ([EMAIL PROTECTED]) wrote: : : If what you mean is that really large samples can lead to distorted : results of significance tests, I would disagree. The problem is not : that the sample is too big, but that Significance tests are interpreted in : inappropritae ways when readers assume statistical significance equals : clinical significance. I've come to believe that terms like "clinical significance" and "practical significance" unintentionally appeal to psychological set and should be phrased as "clinical *importance*" or "practical *importance*" instead. It takes a fair amount of "statistical maturity" to avoid conflating the two uses of "significance." The problem you describe is actually one of letting one's research hypotheses be dictated by the tools one is familiar with ("if all you have is a hammer, you put the cart before the horse"). "Is the effect of framagublification nonzero" is simply not an interesting question. "Is the effect of framagublification large enough to be meaningful according to our current understanding of Turner-Gates pathology" *is* an interesting question, and the fact that the former can be answered by merely clicking on menu items while the latter can't is about as relevant to the research as the fact that student's heights can be easily and repeatably measured is to the instructor's grading system. "This question is too hard to answer, so I'll search for the answer to a different question" is the motto of the coward. One practical problem with really large samples nobody's mentioned yet; lots more missing data, particularly data missing *not* at random. Non-response, procedural glitches that wipe out big chunks of data, compromises when you realize you're about to exceed your budget, etc. The bigger the project, the greater the Murphic exposure. === This list is open to everyone. Occasionally, less thoughtful people send inappropriate messages. Please DO NOT COMPLAIN TO THE POSTMASTER about these messages because the postmaster has no way of controlling them, and excessive complaints will result in termination of the list. For information about this list, including information about the problem of inappropriate messages and information about how to unsubscribe, please see the web page at http://jse.stat.ncsu.edu/ ===
Re: adjusting marks
Richard A. Beldin, Ph.D. ([EMAIL PROTECTED]) wrote: : When my students asked me (as a class) to grade on a curve, I suggested the : following alternative. : "Place N chips in a can. Let them marked in the following way: 10%F, 20%D, 40%C, : 20%B, 10%A. Let each student pick a chip and leave the class, certain of his/her : grade." : For some reason, nobody ever wanted to do that! :-) Thank you! The whole problem with norm-referenced, as opposed to criterion-referenced, grading or performance assessment is that it assumes that you can know how many people did a good job, mediocre job, bad job, etc. *before* any of them have done the job! This is not to say that all forms of norm-referenced measurement are inherently bad, but they're generally only useful for *diagnostic* purposes. Knowing that a student is way behind his peers may give you information on whether he has some problems that need to be dealt with. But using norm-referenced measurements inappropriately leads to creating "designated losers." The best possible position for an individual in a norm-referenced assessment scheme is to be an achiever among a bunch of slackers; it's a better position than being an achiever among a bunch of other achievers.
Re: adjusting marks
Michael Granaas ([EMAIL PROTECTED]) wrote: : While more careful admissions processes would certainly limit the : variability in students, and therefor grading, how is it any different : from grading? If you are going to be more careful with admissions you : need a ranking system of some sort to determine who will succeed and who : will fail. This is just puts the Social Darwinism issue at a different : stage of the process. There's a fundamental difference between admissions decisions and grading decisions: the former involve allocating an inherently scarce resource. There's a limit to the total number of students a school or program can admit, regardless of how certain qualities are distributed among the applicants. However imperfect the available criteria for selecting a subset of applicants are, you're going to *have* to use *some* criteria. All you can do is try to make them as "fair" as possible. There's a genuine cost associated with admitting another applicant. But evaluating performances within a class doesn't involve any inherently scarce resource. There's no particular cost that increases with the grade a student gets.