Re: Analysis of covariance
Bruce Weaver <[EMAIL PROTECTED]> wrote: : Paul's post reminded me of something I read in Keppel's Design and : Analysis. Here's an excerpt from my notes on ANCOVA: : the analysis of covariance is more precise with correlations greater than : .6. Since we rarely obtain correlations of this latter magnitude in the : behavioral sciences, we will not find a unique advantage in the analysis : of covariance in most research applications. I've NEVER seen a pre-post correlation less than .4 : Keppel (1982, p. 513) also prefers the Treatments X Blocks design : to ANCOVA on the grounds that the underlying assumptions are less : stringent: He's wrong in the random assignment case; the assumptions are essentially the same. The ANCOVA estimates are unbiased without any assumptions and without assumption test exactly the same hypothesis as the simple t test or the test of change scores (or treatments x blocks) 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: Analysis of covariance
Paul R. Swank <[EMAIL PROTECTED]> wrote: : Some years ago I did a simulation on the pretest-posttest control group : design lokking at three methods of analysis, ANCOVA, repeated measures : ANOVA, and treatment by block factorial ANOVA (blocking on the pretest using : a median split). I found that that with typical sample sizes, the repeated : measures ANOVA was a bit more powerful than the ANCOVA procedure when the : correlation between pretest and posttest was fairly high (say .90). As noted I tried to : publish the results at the time but aimed a bit too high and received such a : scathing review (what kind of idiot would do this kind of study?) that I : shoved it a drawer and it has never seen the light of day since. You did good. Median splits are always dumb while a test of the change scores will only be more powerful than ANCOVA if the regression coefficient is near 1. Usually the reg coeff is about the same as the corrrelation since the sds are likely to be about the same. Hence ALWAYS use ANCOVA with random assignment to groups = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Analysis of covariance
Morelli Paolo <[EMAIL PROTECTED]> wrote: : HI all, : I have to analyse some clinical data. In particular the analysis is a : comparison between two groups of the mean change baseline to endpoint of a : score. i hope that your study is randomized; if not it's not worth worrying about. If so his analysis is equivalent to ANCOVA on post covarying pre and is the only proper analysis. The true measure of change is a comparison between the two groups since the proper question is how does the experimental group compare to what it would have been without the experimental condition. Ancova simply increases power here you should test for parallelism = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
RE: Analysis of covariance
On 27 Sep 2001, Paul R. Swank wrote: > Some years ago I did a simulation on the pretest-posttest control group > design lokking at three methods of analysis, ANCOVA, repeated measures > ANOVA, and treatment by block factorial ANOVA (blocking on the pretest using > a median split). I found that that with typical sample sizes, the repeated > measures ANOVA was a bit more powerful than the ANCOVA procedure when the > correlation between pretest and posttest was fairly high (say .90). As noted > below, this is because the ANCOVA and ANOVA methods are approaching the same > solution but ANCOVA loses a degree of freedom estimating the regression > parameter when the ANOVA doesn't. Of course this effect diminshes as the > sample size gets larger because the loss of one df is diminished. On the > other hand, the treatment by block design tends to have a bit more power > when the correlation between pretest and posttest is low (< .30). I tried to > publish the results at the time but aimed a bit too high and received such a > scathing review (what kind of idiot would do this kind of study?) that I > shoved it a drawer and it has never seen the light of day since. I did the > syudy because it seemed at the time that everyone was using this design but > were unsure of the analysis and I thought a demonstration would be helpful. > SO, to make a long story even longer, the ANCOVA seems to be most powerful > in those circumstances one is likely to run into but does have somewhat > rigid assumptions about homogeneity of regression slopes. Of course the > repeated measures ANOVA indirectly makes the same assumption but at such > high correlations, this is really a homogenity of variance issue as well. > The second thought is for you reviewers out there trying to soothe your own > egos by dumping on someone else's. Remember, the researcher you squelch > today might be turned off to research and fail to solve a meaty problem > tomorrow. > > Paul R. Swank, Ph.D. > Professor > Developmental Pediatrics > UT Houston Health Science Center > Paul's post reminded me of something I read in Keppel's Design and Analysis. Here's an excerpt from my notes on ANCOVA: Keppel (1982, p. 512) says: If the choice is between blocking and the analysis of covariance, Feldt (1958) has shown that blocking is more precise when the correlation between the covariate and the dependent variable is less than .4, while the analysis of covariance is more precise with correlations greater than .6. Since we rarely obtain correlations of this latter magnitude in the behavioral sciences, we will not find a unique advantage in the analysis of covariance in most research applications. Keppel (1982, p. 513) also prefers the Treatments X Blocks design to ANCOVA on the grounds that the underlying assumptions are less stringent: Both within-subjects designs and analyses of covariance require a number of specialized statistical assumptions. With the former, homogeneity of between treatment differences and the absence of differential carryover effects are assumptions that are critical for an unambiguous interpretation of the results of an experiment. With the latter, the most stringent is the assumption of homogeneous within-group regression coefficients. Both the analysis of covariance and the analysis of within-subjects designs are sensitive only to the linear relationship between X and Y, in the first case, and between pairs of treatment conditions in the second case. In contrast, the Treatments X Blocks design is sensitive to any type of relationship between treatments and blocks--not just linear. As Winer puts it, the Treatments X Blocks design "is a function-free regression scheme" (1971, p. 754). This is a major advantage of the Treatments X Blocks design. In short, the Treatments X Blocks design does not have restrictive assumptions and, for this reason, is to be preferred for its relative freedom from statistical assumptions underlying the data analysis. -- Bruce Weaver E-mail: [EMAIL PROTECTED] Homepage: http://www.angelfire.com/wv/bwhomedir/ = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Analysis of covariance
I would have to respectfully disagree with Dennis' comment also. Having the pre values twice in the model does not hurt or change anything in interpreting the treatment effect. BUT I do not like this approach. It makes the results more difficult to interpret when you do have a variable in both places. As it is mandatory to have the pre measurement as a separate covariable at any rate, the response variable I prefer is the follow-up assessment, not the change. A good discussion is in Stephen Senn's "Statistical Issues in Drug Development" book (Wiley). -Frank Harrell Radford Neal wrote: > > In article <[EMAIL PROTECTED]>, > Dennis Roberts <[EMAIL PROTECTED]> wrote: > > >the basic idea is to be able to "explain" the post score variance in terms > >of something ELSE ... that is, for example ... we know that some of the > >variance in pain is due to one's TOLERANCE for PAIN ... thus, if we can > >remove the part of pain variance that is due to TOLERANCE FOR pain ... then > >the leftover variance on pain is a purer measure in its own right .. > > > >if you do as suggested ... remove the pre from the post ... say pre pain > >from post pain ... what is left over? it is not pain anymore but rather, > >some OTHER variable ... which is not what the purpose of the study was ... > >to investigate (i assume anyway) > > Well, the idea is that the OTHER variable is the treatment effect, > whose quantification presumably IS the purpose of the study. I think > this is a pretty standard thing to do. > > It seems that the original question was meant to address the more > technical issue of whether you can include the pre-treatment value as > an explanatory variable when the response variable is already the > CHANGE from before treatment to after treatment. As another poster > has ably explained, you can, though it's a bit strange and redundant. > >Radford > > > Radford M. Neal [EMAIL PROTECTED] > Dept. of Statistics and Dept. of Computer Science [EMAIL PROTECTED] > University of Toronto http://www.cs.utoronto.ca/~radford > -- Frank E Harrell Jr Prof. of Biostatistics & Statistics Div. of Biostatistics & Epidem. Dept. of Health Evaluation Sciences U. Virginia School of Medicine http://hesweb1.med.virginia.edu/biostat = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
RE: Analysis of covariance
Some years ago I did a simulation on the pretest-posttest control group design lokking at three methods of analysis, ANCOVA, repeated measures ANOVA, and treatment by block factorial ANOVA (blocking on the pretest using a median split). I found that that with typical sample sizes, the repeated measures ANOVA was a bit more powerful than the ANCOVA procedure when the correlation between pretest and posttest was fairly high (say .90). As noted below, this is because the ANCOVA and ANOVA methods are approaching the same solution but ANCOVA loses a degree of freedom estimating the regression parameter when the ANOVA doesn't. Of course this effect diminshes as the sample size gets larger because the loss of one df is diminished. On the other hand, the treatment by block design tends to have a bit more power when the correlation between pretest and posttest is low (< .30). I tried to publish the results at the time but aimed a bit too high and received such a scathing review (what kind of idiot would do this kind of study?) that I shoved it a drawer and it has never seen the light of day since. I did the syudy because it seemed at the time that everyone was using this design but were unsure of the analysis and I thought a demonstration would be helpful. SO, to make a long story even longer, the ANCOVA seems to be most powerful in those circumstances one is likely to run into but does have somewhat rigid assumptions about homogeneity of regression slopes. Of course the repeated measures ANOVA indirectly makes the same assumption but at such high correlations, this is really a homogenity of variance issue as well. The second thought is for you reviewers out there trying to soothe your own egos by dumping on someone else's. Remember, the researcher you squelch today might be turned off to research and fail to solve a meaty problem tomorrow. Paul R. Swank, Ph.D. Professor Developmental Pediatrics UT Houston Health Science Center -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of jim clark Sent: Thursday, September 27, 2001 7:00 AM To: [EMAIL PROTECTED] Subject: Re: Analysis of covariance Hi On 26 Sep 2001, Burke Johnson wrote: > R Pretest Treatment Posttest > R PretestControl Posttest > In the social sciences (e.g., see Pedhazur's popular > regression text), the most popular analysis seems to be to > run a GLM (this version is often called an ANCOVA), where Y > is the posttest measure, X1 is the pretest measure, and X2 is > the treatment variable. Assuming that X1 and X2 do not > interact, ones' estimate of the treatment effect is given by > B2 (i.e., the partial regression coefficient for the > treatment variable which controls for adjusts for pretest > differences). > Another traditionally popular analysis for the design given > above is to compute a new, gain score variable (posttest > minus pretest) for all cases and then run a GLM (ANOVA) to > see if the difference between the gains (which is the > estimate of the treatment effect) is statistically > significant. > The third, and somewhat less popular (?) way to analyze the > above design is to do a mixed ANOVA model (which is also a > GLM but it is harder to write out), where Y is the posttest, > X1 is "time" which is a repeated measures variable (e.g., > time is 1 for pretest and 2 for posttest for all cases), and > X2 is the between group, treatment variable. In this case one > looks for treatment impact by testing the statistical > significance of the two-way interaction between the time and > the treatment variables. Usually, you ask if the difference > between the means at time two is greater than the difference > at time one (i.e., you hope that the treatment lines will not > be parallel) > Results will vary depending on which of these three > approaches you use, because each approach estimates the > counterfactual in a slightly different way. I believe it was > Reichardt and Mark (in Handbook of Applied Social Research > Methods) that suggested analyzing your data using more than > one of these three statistical methods. Methods 2 and 3 are equivalent to one another. The F for the difference between change scores will equal the F for the interaction. I believe that one way to think of the difference between methods 1 and 2/3 is that in 2/3 you "regress" t2 on t1 assuming slope=1 and intercept=0 (i.e., the "predicted" score is the t1 score), whereas in method 1 you estimate the slope and intercept from the data. Presumably it would be possible to simulate the differences between the two analyses as a function of the magnitude of the difference between means and the relationship between t1 and t2. I don't know if anyone has done that. Best wishes Jim =
Re: Analysis of covariance
Hi On 26 Sep 2001, Burke Johnson wrote: > R Pretest Treatment Posttest > R PretestControl Posttest > In the social sciences (e.g., see Pedhazur's popular > regression text), the most popular analysis seems to be to > run a GLM (this version is often called an ANCOVA), where Y > is the posttest measure, X1 is the pretest measure, and X2 is > the treatment variable. Assuming that X1 and X2 do not > interact, ones' estimate of the treatment effect is given by > B2 (i.e., the partial regression coefficient for the > treatment variable which controls for adjusts for pretest > differences). > Another traditionally popular analysis for the design given > above is to compute a new, gain score variable (posttest > minus pretest) for all cases and then run a GLM (ANOVA) to > see if the difference between the gains (which is the > estimate of the treatment effect) is statistically > significant. > The third, and somewhat less popular (?) way to analyze the > above design is to do a mixed ANOVA model (which is also a > GLM but it is harder to write out), where Y is the posttest, > X1 is "time" which is a repeated measures variable (e.g., > time is 1 for pretest and 2 for posttest for all cases), and > X2 is the between group, treatment variable. In this case one > looks for treatment impact by testing the statistical > significance of the two-way interaction between the time and > the treatment variables. Usually, you ask if the difference > between the means at time two is greater than the difference > at time one (i.e., you hope that the treatment lines will not > be parallel) > Results will vary depending on which of these three > approaches you use, because each approach estimates the > counterfactual in a slightly different way. I believe it was > Reichardt and Mark (in Handbook of Applied Social Research > Methods) that suggested analyzing your data using more than > one of these three statistical methods. Methods 2 and 3 are equivalent to one another. The F for the difference between change scores will equal the F for the interaction. I believe that one way to think of the difference between methods 1 and 2/3 is that in 2/3 you "regress" t2 on t1 assuming slope=1 and intercept=0 (i.e., the "predicted" score is the t1 score), whereas in method 1 you estimate the slope and intercept from the data. Presumably it would be possible to simulate the differences between the two analyses as a function of the magnitude of the difference between means and the relationship between t1 and t2. I don't know if anyone has done that. Best wishes Jim James M. Clark (204) 786-9757 Department of Psychology(204) 774-4134 Fax University of Winnipeg 4L05D Winnipeg, Manitoba R3B 2E9 [EMAIL PROTECTED] CANADA http://www.uwinnipeg.ca/~clark = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Analysis of covariance
At 02:26 PM 9/26/01 -0500, Burke Johnson wrote: > >From my understanding, there are three popular ways to analyze the > following design (let's call it the pretest-posttest control-group design): > >R Pretest Treatment Posttest >R PretestControl Posttest if random assignment has occurred ... then, we assume and we had better find that the means on the pretest are close to being the same ... if we don't, then we wonder about random assignment (which creates a mess) anyway, i digress ... what i would do is to do a simple t test on the difference in posttest means and, if you find something here ... then that means that treatment "changed" differentially compared to control if that happens, why do anything more complicated? has not the answer to your main question been found? now, what if you don't ... then, maybe something a bit more complex is appropriate IMHO _ dennis roberts, educational psychology, penn state university 208 cedar, AC 8148632401, mailto:[EMAIL PROTECTED] http://roberts.ed.psu.edu/users/droberts/drober~1.htm = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Analysis of covariance
>From my understanding, there are three popular ways to analyze the following design >(let's call it the pretest-posttest control-group design): R Pretest Treatment Posttest R PretestControl Posttest In the social sciences (e.g., see Pedhazur's popular regression text), the most popular analysis seems to be to run a GLM (this version is often called an ANCOVA), where Y is the posttest measure, X1 is the pretest measure, and X2 is the treatment variable. Assuming that X1 and X2 do not interact, ones' estimate of the treatment effect is given by B2 (i.e., the partial regression coefficient for the treatment variable which controls for adjusts for pretest differences). Another traditionally popular analysis for the design given above is to compute a new, gain score variable (posttest minus pretest) for all cases and then run a GLM (ANOVA) to see if the difference between the gains (which is the estimate of the treatment effect) is statistically significant. The third, and somewhat less popular (?) way to analyze the above design is to do a mixed ANOVA model (which is also a GLM but it is harder to write out), where Y is the posttest, X1 is "time" which is a repeated measures variable (e.g., time is 1 for pretest and 2 for posttest for all cases), and X2 is the between group, treatment variable. In this case one looks for treatment impact by testing the statistical significance of the two-way interaction between the time and the treatment variables. Usually, you ask if the difference between the means at time two is greater than the difference at time one (i.e., you hope that the treatment lines will not be parallel) Results will vary depending on which of these three approaches you use, because each approach estimates the counterfactual in a slightly different way. I believe it was Reichardt and Mark (in Handbook of Applied Social Research Methods) that suggested analyzing your data using more than one of these three statistical methods. I'd be interested in any thoughs you have about these three approaches. Take care, Burke Johnson http://www.coe.usouthal.edu/bset/Faculty/BJohnson/Burke.html = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Analysis of covariance
In article <[EMAIL PROTECTED]>, Dennis Roberts <[EMAIL PROTECTED]> wrote: >the basic idea is to be able to "explain" the post score variance in terms >of something ELSE ... that is, for example ... we know that some of the >variance in pain is due to one's TOLERANCE for PAIN ... thus, if we can >remove the part of pain variance that is due to TOLERANCE FOR pain ... then >the leftover variance on pain is a purer measure in its own right .. > >if you do as suggested ... remove the pre from the post ... say pre pain >from post pain ... what is left over? it is not pain anymore but rather, >some OTHER variable ... which is not what the purpose of the study was ... >to investigate (i assume anyway) Well, the idea is that the OTHER variable is the treatment effect, whose quantification presumably IS the purpose of the study. I think this is a pretty standard thing to do. It seems that the original question was meant to address the more technical issue of whether you can include the pre-treatment value as an explanatory variable when the response variable is already the CHANGE from before treatment to after treatment. As another poster has ably explained, you can, though it's a bit strange and redundant. Radford Radford M. Neal [EMAIL PROTECTED] Dept. of Statistics and Dept. of Computer Science [EMAIL PROTECTED] University of Toronto http://www.cs.utoronto.ca/~radford = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Analysis of covariance
Morelli Paolo wrote: > > HI all, > I have to analyse some clinical data. In particular the analysis is a > comparison between two groups of the mean change baseline to endpoint of a > score. The statistician who planned the analysis used the ANCOVA on the mean > change, using as covariate the baseline values of the scores. > Do you think this analysis is correct? > I thing that in this way we are correcting twice. I think that the right > analysis is an ANOVA on the mean change. > Please let me know your opinion > thanks > Paolo It's convoluted, but not wrong. I do it sometimes because some researchers, for whatever reason, are more comfortable with that approach. The research question is usually: If two people have the same initial value, will there final value be the same except for the effect of treatment. (I'm assuming your groups are the result of random assignment to treatment. If not, these arguments does not apply and I leave it to you to read the literature to find out why. I'm quickly using up my daily allotment of keystrokes!) This gets you the ANCOVA model final = constant + b1 * initial + treatment effect Change is final - initial, so the model can be rewritten change = constant + (b1-1)* initial + treatment effect and the estimated treatment effect is the same. Since the treatment effect is the same, the analysis is okay, odd as it looks. = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Analysis of covariance
At 03:19 PM 9/25/01 +, Radford Neal wrote: >Neither the question nor the response are all that clearly phrased, but >when I interpret them according to my reading, I don't agree. For instance, >if you're measuring pain levels, I don't see anything wrong with measuring >pain before treatment, randomly assigning patients to treatment and control >groups, doing a regression for pain level afterwards with the pain level >before and a treatment/control indicator as explanatory variables, and >judging the effectiveness of the treatment by looking at the coefficient for >the treatment/control variable. Or is the actual proposal something else? IMHO seems like to remove the variance from post pain ... using pre pain variance ... is a no brainer ... since the r between the two pain readings will necessarily be high (unless there is something really screwy about the data like severe restriction of range on the post measure) ... what has been explained in the post pain variance? pain? the basic idea is to be able to "explain" the post score variance in terms of something ELSE ... that is, for example ... we know that some of the variance in pain is due to one's TOLERANCE for PAIN ... thus, if we can remove the part of pain variance that is due to TOLERANCE FOR pain ... then the leftover variance on pain is a purer measure in its own right .. if you do as suggested ... remove the pre from the post ... say pre pain from post pain ... what is left over? it is not pain anymore but rather, some OTHER variable ... which is not what the purpose of the study was ... to investigate (i assume anyway) i do most certainly agree with radford that ... random assignment is still essential in this design ... unfortunately, far too many folks use ANCOVA to somehow makeup for the fact that NON random assignment happened and, they think ANCOVA will solve that problem ... it won't >Radford Neal > > >Radford M. Neal [EMAIL PROTECTED] >Dept. of Statistics and Dept. of Computer Science [EMAIL PROTECTED] >University of Toronto http://www.cs.utoronto.ca/~radford > > > >= >Instructions for joining and leaving this list and remarks about >the problem of INAPPROPRIATE MESSAGES are available at > http://jse.stat.ncsu.edu/ >= _ dennis roberts, educational psychology, penn state university 208 cedar, AC 8148632401, mailto:[EMAIL PROTECTED] http://roberts.ed.psu.edu/users/droberts/drober~1.htm = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Analysis of covariance
Morelli Paolo wrote: >>I have to analyse some clinical data. In particular the analysis is a >>comparison between two groups of the mean change baseline to endpoint of a >>score. The statistician who planned the analysis used the ANCOVA on the mean >>change, using as covariate the baseline values of the scores. >>Do you think this analysis is correct? Dennis Roberts <[EMAIL PROTECTED]> wrote: >NO! ... this is not a legitimate covariate ... a pre measure of the same >thing you are measuring latter as evidence of effectiveness Neither the question nor the response are all that clearly phrased, but when I interpret them according to my reading, I don't agree. For instance, if you're measuring pain levels, I don't see anything wrong with measuring pain before treatment, randomly assigning patients to treatment and control groups, doing a regression for pain level afterwards with the pain level before and a treatment/control indicator as explanatory variables, and judging the effectiveness of the treatment by looking at the coefficient for the treatment/control variable. Or is the actual proposal something else? Radford Neal Radford M. Neal [EMAIL PROTECTED] Dept. of Statistics and Dept. of Computer Science [EMAIL PROTECTED] University of Toronto http://www.cs.utoronto.ca/~radford = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Analysis of covariance
Paolo -- Here comes my usual response to messages similar to yours: Following the use of Regression/Linear Models: 1. State your research question in "NATURAL LANGUAGE" not in terms of a "canned statistical name" that may or may not be relevant to your question. 2. Create an ASSUMED MODEL that allows you to translate your "NATURAL LANGUAGE" questions into RESTRICTIONS on your ASSUMED MODEL. 3. Impose the restrictions on your ASSUMED MODEL to obtain your RESTRICTED MODEL and then you have the essentials to test your hypotheses. If this procedure is IDENTICAL to someone's COVARIANCE ANALYSIS then you might want to call yours a COVARIANCE ANALYSIS. -- Joe *** Joe H. Ward, Jr. *** 167 East Arrowhead Dr. *** San Antonio, TX 78228-2402 *** Phone: 210-433-6575 *** Fax: 210-433-2828 *** Email: [EMAIL PROTECTED] *** http://www.northside.isd.tenet.edu/healthww/biostatistics/wardindex.html *** --- *** Health Careers High School *** 4646 Hamilton-Wolfe *** San Antonio, TX 78229 * - Original Message - From: "Morelli Paolo" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Tuesday, September 25, 2001 5:26 AM Subject: Analysis of covariance > HI all, > I have to analyse some clinical data. In particular the analysis is a > comparison between two groups of the mean change baseline to endpoint of a > score. The statistician who planned the analysis used the ANCOVA on the mean > change, using as covariate the baseline values of the scores. > Do you think this analysis is correct? > I thing that in this way we are correcting twice. I think that the right > analysis is an ANOVA on the mean change. > Please let me know your opinion > thanks > Paolo > > > > > = > Instructions for joining and leaving this list and remarks about > the problem of INAPPROPRIATE MESSAGES are available at > http://jse.stat.ncsu.edu/ > = > = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Analysis of covariance
If you are using ANCOVA then the base score is the covariate and the final score the criterion. ANCOVA is generally preferred to ANOVA on gain scores. John Ambrose University of the Virgin Islands St. Thomas VI 00802 At 10:26 AM 9/25/01 +, Morelli Paolo wrote: >HI all, >I have to analyse some clinical data. In particular the analysis is a >comparison between two groups of the mean change baseline to endpoint of a >score. The statistician who planned the analysis used the ANCOVA on the mean >change, using as covariate the baseline values of the scores. >Do you think this analysis is correct? >I thing that in this way we are correcting twice. I think that the right >analysis is an ANOVA on the mean change. >Please let me know your opinion >thanks >Paolo > > > > >= >Instructions for joining and leaving this list and remarks about >the problem of INAPPROPRIATE MESSAGES are available at > http://jse.stat.ncsu.edu/ >= = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Re: Analysis of covariance
At 10:26 AM 9/25/01 +, Morelli Paolo wrote: >HI all, >I have to analyse some clinical data. In particular the analysis is a >comparison between two groups of the mean change baseline to endpoint of a >score. The statistician who planned the analysis used the ANCOVA on the mean >change, using as covariate the baseline values of the scores. >Do you think this analysis is correct? NO! ... this is not a legitimate covariate ... a pre measure of the same thing you are measuring latter as evidence of effectiveness the notion of a covariate is to have previously collected data ... on a variable that rationally should explain some of the variance in the criterion ... and the idea is to "remove" that part of the criterion variance that can be accounted for by the co-linearity with the covariate in situations where the treatment effect is likely to be small ... especially if error variance is large ... using an appropriate covariate (assuming of course that Ss were randomly assigned to the different conditions) is a good way to reduce the error term and hence, increase your chances for finding "significance" (if that is your goal) >I thing that in this way we are correcting twice. I think that the right >analysis is an ANOVA on the mean change. >Please let me know your opinion >thanks >Paolo > > > > >= >Instructions for joining and leaving this list and remarks about >the problem of INAPPROPRIATE MESSAGES are available at > http://jse.stat.ncsu.edu/ >= _ dennis roberts, educational psychology, penn state university 208 cedar, AC 8148632401, mailto:[EMAIL PROTECTED] http://roberts.ed.psu.edu/users/droberts/drober~1.htm = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
Analysis of covariance
HI all, I have to analyse some clinical data. In particular the analysis is a comparison between two groups of the mean change baseline to endpoint of a score. The statistician who planned the analysis used the ANCOVA on the mean change, using as covariate the baseline values of the scores. Do you think this analysis is correct? I thing that in this way we are correcting twice. I think that the right analysis is an ANOVA on the mean change. Please let me know your opinion thanks Paolo = Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =
