>ANOVA is used in the analysis. But in separating the means, no >pval correction is used. The method for separating the means >is a protected Fisher LSD.
Fisher LSD is better than I thought, but you are still going about it the wrong way. The only way the Fisher LSD is protected is if you do the ANOVA first and it shows a significant difference between the groups, then you do the post hoc tests accordingly. Just doing the LSD without doing the ANOVA first to determine a significant difference between your groups doesn't accomplish much.
The Fisher LSD is also generally not recommended for exactly the situation you have here--close values, with one or two outstanding groups. There are a lot of tests that are much better (not even getting into nonparametric tests).
Incidentally, the odds of you making at least one Type I error, using your current method (posthoc tests before the ANOVA), is 99.8%
>. But hey, we're not talking about bringing a drug to market >here. This is a listening test, for Christ's sake!
My problem is that you are adding apparent legitimacy to the study without appreciably strengthening the analysis.
This is in many ways functionally similar to the problem of pseudoreplication--you are claiming a stronger analysis than you actually have.
>The statistics aren't the weakest link in interpreting the results >of the test....and you should address these in your analysis.
>The author gives which one is in first place, but announces at >the beginning the requirements for a clear winner.
The more comparisons we do, the greater the chance that we made an error in one of our comparisons (you can model this with a binomial distribution fairly trivially). Here, in each graph, there are 10 possible comparisons. The odds of us having a false positive in one of those graphs is around 40.1%. Across multiple studies, like this one, you can imagine the odds that one of our comparisons, somewhere, is off.
The "error bars methodology" only gets you so far. There are a lot of potentially good methodologies here, but this isn't one of them. There is a reason that ANOVA is used with comparisons among multiple groups.
Disclaimer: I haven't read whether he used a correction or even his general statistical methodology. He may of and I just don't know about it--this is all from a cursory analysis.
Good point. You also have to very be careful about doing multiple comparisons like this without some form of correction (such as Bonferroni)--otherwise your 95% confidence interval gets progressively weaker for each test you add since the probability that you made a Type I error *somewhere* becomes very high very quickly.
There is no statistical test which is valid for small groups of inbred mice. Particularly when trying to extend those results to humans.
Saying that "Electrical shavers make your brain rot" off of significant but not astoundingly skewed results in a single study involving 16 mice is a little bit premature.
Your name does not fit with standard naming conventions for the Alpha Complex. You must be from the Outside. Only Commie Mutant Traitors live outside, you must be a Commie Mutant Traitor.
Troubleshooter teams are being dispatched. Have a nice day!
If you had to "*smiles*" it implies that you were not smiling. Happy citizens are always smiling. Are you happy citizen? Only Commie Mutant Traitors would ever be in a non-smiling state.
>I agree it would be wonderful of Apple to do the Port and then >add specific extras, but I think the licensing options would be >the biggest hurdle to overcome.
Honestly, as nice as it would sometimes be, I am not expecting Apple to ever develop with the full GPL for something like that.
They want a library they can use as a backend for all of their apps and add functionality to the OS. Things that will link to other parts of the OS. They *don't* want to have to worry about linking.
My impression so far has been that they don't mind contributing to open source projects and taking advantage of them--KHTML comes to mind--but they are not going to want to build an app or a library for an app from open source that they can't link into their proprietary libraries.
Unfortunate--it seems like a nice app. If we want to see this level of functionality we are going to have to build it ourselves. Either from scratch or from LyX.
Part of the problem with this is that the ports brought over, while they are integrated with Aqua and much nicer than working with them through X11, still aren't pretty and still need work. I'm concerned that "Aquafication" will stop after it starts running, but that's only where the "fun" begins.
Take LyX/Mac as an example of this. I installed LyX/Mac earlier today to check it out, though I admit I was skeptical and have a strong preference for TeXShop and iTeXMac going into this. It's *significantly* nicer than using LyX through X11--its nice and clean, I can copy-and-paste out of it (but not into it), and its antialiased and gives good results.
It isn't, however, fully integrated.
The remaining issues range from major mechanical things such as paste to Cocoa-specific features such as services or the native built-in spell check that are nice to have around, to minor things like the icons reminding me of windows or that the command keys don't appear in the menus.
I think that this is a good program and I applaud it and it may see some use (though paste not working may be a deal killer for me on this particular app), and I really like that Qt/Mac is there and makes this an easier process but it is not going to be a panacea in getting software ported to Aqua.
Ever study Alex? Read up, its an interesting case study that if you apply the older "animals lack cognition" model raises serious question about whether *we* have actual cognition.
Considering that crows can both make and use tools and octopus can learn to open containers by watching other octopus, limiting these things to "primitive cause-and-effect association" seems a bit chauvinistic on our part.
>Apple forces you to upgrade the damn OS every single year at >the low cost of $129.
Wow, Steve Jobs comes down to your house and makes you pay for the update at gunpoint? Why haven't I seen *that* on the news?
Face it, Apple doesn't force you, you can either pay the upgrade fee, or you can go without and your OS will still keep chugging along. No self destruct sequence, no crazed hoard of killer rabbits coming after you, it will just continue to work.
...that might throw a wrench in that, even assuming the apps do spend their time in system calls.
0) The PowerPC was an order of magnitude faster than 68k series. IIRC the 601 had twice the clock and was faster per clock than the 68040. There is no such advantage here.
1) In order to handle everything correctly here the bit-order is going to have to be switched (different endians). This is not fast on a good day.
5, incidentally, is so bad that most of us write it off as not having existed in the first place. I think one trekkie described it best when he said "Kirk and god compare egos, god loses."
Slashdot Mirror
>ANOVA is used in the analysis. But in separating the means, no
...and you should address these in your analysis.
>pval correction is used. The method for separating the means
>is a protected Fisher LSD.
Fisher LSD is better than I thought, but you are still going about it the wrong way. The only way the Fisher LSD is protected is if you do the ANOVA first and it shows a significant difference between the groups, then you do the post hoc tests accordingly. Just doing the LSD without doing the ANOVA first to determine a significant difference between your groups doesn't accomplish much.
The Fisher LSD is also generally not recommended for exactly the situation you have here--close values, with one or two outstanding groups. There are a lot of tests that are much better (not even getting into nonparametric tests).
Incidentally, the odds of you making at least one Type I error, using your current method (posthoc tests before the ANOVA), is 99.8%
>. But hey, we're not talking about bringing a drug to market
>here. This is a listening test, for Christ's sake!
My problem is that you are adding apparent legitimacy to the study without appreciably strengthening the analysis.
This is in many ways functionally similar to the problem of pseudoreplication--you are claiming a stronger analysis than you actually have.
>The statistics aren't the weakest link in interpreting the results
>of the test.
" The odds of us having a false positive in one of those graphs is around 40.1%. "
That should read "The odds of having a false positive among the tests on any one of those graphs is around 40.1%"
>The author gives which one is in first place, but announces at
>the beginning the requirements for a clear winner.
The more comparisons we do, the greater the chance that we made an error in one of our comparisons (you can model this with a binomial distribution fairly trivially). Here, in each graph, there are 10 possible comparisons. The odds of us having a false positive in one of those graphs is around 40.1%. Across multiple studies, like this one, you can imagine the odds that one of our comparisons, somewhere, is off.
The "error bars methodology" only gets you so far. There are a lot of potentially good methodologies here, but this isn't one of them. There is a reason that ANOVA is used with comparisons among multiple groups.
Disclaimer: I haven't read whether he used a correction or even his general statistical methodology. He may of and I just don't know about it--this is all from a cursory analysis.
Good point. You also have to very be careful about doing multiple comparisons like this without some form of correction (such as Bonferroni)--otherwise your 95% confidence interval gets progressively weaker for each test you add since the probability that you made a Type I error *somewhere* becomes very high very quickly.
There is no statistical test which is valid for small groups of inbred mice. Particularly when trying to extend those results to humans.
Saying that "Electrical shavers make your brain rot" off of significant but not astoundingly skewed results in a single study involving 16 mice is a little bit premature.
Your name does not fit with standard naming conventions for the Alpha Complex. You must be from the Outside. Only Commie Mutant Traitors live outside, you must be a Commie Mutant Traitor.
Troubleshooter teams are being dispatched. Have a nice day!
If you had to "*smiles*" it implies that you were not smiling. Happy citizens are always smiling. Are you happy citizen? Only Commie Mutant Traitors would ever be in a non-smiling state.
Just buy from labels that aren't with the RIAA. Easy.
http://www.magnetbox.com/riaa/
There are reasons I use LaTeX.
That the programs do not make idiotic design changes while trying to "help" me is a big part of that.
I always just *knew* that MS was trying to force us to buy hardware upgrades to run their OS...
>I agree it would be wonderful of Apple to do the Port and then
>add specific extras, but I think the licensing options would be
>the biggest hurdle to overcome.
Honestly, as nice as it would sometimes be, I am not expecting Apple to ever develop with the full GPL for something like that.
They want a library they can use as a backend for all of their apps and add functionality to the OS. Things that will link to other parts of the OS. They *don't* want to have to worry about linking.
My impression so far has been that they don't mind contributing to open source projects and taking advantage of them--KHTML comes to mind--but they are not going to want to build an app or a library for an app from open source that they can't link into their proprietary libraries.
Unfortunate--it seems like a nice app. If we want to see this level of functionality we are going to have to build it ourselves. Either from scratch or from LyX.
Part of the problem with this is that the ports brought over, while they are integrated with Aqua and much nicer than working with them through X11, still aren't pretty and still need work. I'm concerned that "Aquafication" will stop after it starts running, but that's only where the "fun" begins.
Take LyX/Mac as an example of this. I installed LyX/Mac earlier today to check it out, though I admit I was skeptical and have a strong preference for TeXShop and iTeXMac going into this. It's *significantly* nicer than using LyX through X11--its nice and clean, I can copy-and-paste out of it (but not into it), and its antialiased and gives good results.
It isn't, however, fully integrated.
The remaining issues range from major mechanical things such as paste to Cocoa-specific features such as services or the native built-in spell check that are nice to have around, to minor things like the icons reminding me of windows or that the command keys don't appear in the menus.
I think that this is a good program and I applaud it and it may see some use (though paste not working may be a deal killer for me on this particular app), and I really like that Qt/Mac is there and makes this an easier process but it is not going to be a panacea in getting software ported to Aqua.
>Hacking into a horse's brain ...just needs a really sharp axe.
:-)
Sorry, good information, but couldn't resist.
Ever study Alex? Read up, its an interesting case study that if you apply the older "animals lack cognition" model raises serious question about whether *we* have actual cognition.
Considering that crows can both make and use tools and octopus can learn to open containers by watching other octopus, limiting these things to "primitive cause-and-effect association" seems a bit chauvinistic on our part.
It's one word, not two.
Or, in your case, just use the GUI: System Preferences->Software Update.
Still works here. I've had no problems with it.
>Apple forces you to upgrade the damn OS every single year at
>the low cost of $129.
Wow, Steve Jobs comes down to your house and makes you pay for the update at gunpoint? Why haven't I seen *that* on the news?
Face it, Apple doesn't force you, you can either pay the upgrade fee, or you can go without and your OS will still keep chugging along. No self destruct sequence, no crazed hoard of killer rabbits coming after you, it will just continue to work.
What is that, a lesson on how not to design a chart?
From the article:
....and thus getting what he richly deserves for that crime against nature.
>I have to say that I'm happy - I can keep on using XP.
...that might throw a wrench in that, even assuming the apps do spend their time in system calls.
0) The PowerPC was an order of magnitude faster than 68k series. IIRC the 601 had twice the clock and was faster per clock than the 68040. There is no such advantage here.
1) In order to handle everything correctly here the bit-order is going to have to be switched (different endians). This is not fast on a good day.
What, is this a conspiracy to make Mac systems run more slowly?
Cool idea, I wish them the best, but I like apps to load within my lifetime...
What, did you have a humor bypass at some point?
The poster was not being serious, it was a *joke*.
Why do I get the feeling it would be like that scene in Silent Running where the two robots look over at where the third used to be?
What are the advantages/disadvantages of Eiffle compared to languages such as Ruby or Python?
Truism:
Odd numbered star treks suck.
No way around it, they just do.
5, incidentally, is so bad that most of us write it off as not having existed in the first place. I think one trekkie described it best when he said "Kirk and god compare egos, god loses."
1 can't even compare to that.