I saw a number of points referencing the apples to apples comparision. What I did not see is discussion on particpant(I believe, the Rosenthal effect), bias-most would assume there would be none but there was not discussion in the methods on how to control for this. Experimenter bias (I believe the Hawthorn Effect-could of mixed up with Rosenthal effect)- experimenters are aware of what they looking for inadvertently skew the test results. Also, it would make for a better test of the OS support and administration if both systems ran Apache, mysql, and tomcat, since these will run on windows systems (IIS will only run on Window), and both used PHP. In experinments, the only difference should be what experimentally is being controlled for, in this case it would OS support. It can still represent the real world since one may use open source software with Microsoft OS. Also, it is a bit odd that the 3rd party vendor products were not provided. Irregardless of what the researches point out (protecting the vendor), you can't replicate the study with out that exact methods used in the study and that includes 3rd party apps (poor research practice which brings in to question the researches motives). Also, the results as some point out really don't provide much in regards to proof of differences other than some odd numbers, the size of the subject pool is to small to run statistical analysis so what number they did capture cannot indicate causality not even a relational association. All conclusions are subject and have not statistically backed. So how do we know if 34% is really difference for each administration. I would think the bias alone would need to be adressed if we are to believe the data let along the weak experimental controls used in the study that would need to be rectified as well...
How does the Doctor defend against these (what some would consider catastrophic) failures in research protocol?
If I am off base here, sorry.
I was under the impression that BMW's series 5 and up have a form of Windows (I thought CE) in them already. I was also under the impression this has been going on for 3 years. It is one of the reasons the 7 series has had problems, such as locking drivers in thier own car. They have a tendency to shutdown for no reason. The problems relate to the computer chip and software used in the cars...
Slashdot Mirror
Sorry, that was Partipant Bias and Hawthorn Effect. While the experimenter Bias is the Rosenthal effect. Sorry about that...
I saw a number of points referencing the apples to apples comparision. What I did not see is discussion on particpant(I believe, the Rosenthal effect), bias-most would assume there would be none but there was not discussion in the methods on how to control for this. Experimenter bias (I believe the Hawthorn Effect-could of mixed up with Rosenthal effect)- experimenters are aware of what they looking for inadvertently skew the test results. Also, it would make for a better test of the OS support and administration if both systems ran Apache, mysql, and tomcat, since these will run on windows systems (IIS will only run on Window), and both used PHP. In experinments, the only difference should be what experimentally is being controlled for, in this case it would OS support. It can still represent the real world since one may use open source software with Microsoft OS. Also, it is a bit odd that the 3rd party vendor products were not provided. Irregardless of what the researches point out (protecting the vendor), you can't replicate the study with out that exact methods used in the study and that includes 3rd party apps (poor research practice which brings in to question the researches motives). Also, the results as some point out really don't provide much in regards to proof of differences other than some odd numbers, the size of the subject pool is to small to run statistical analysis so what number they did capture cannot indicate causality not even a relational association. All conclusions are subject and have not statistically backed. So how do we know if 34% is really difference for each administration. I would think the bias alone would need to be adressed if we are to believe the data let along the weak experimental controls used in the study that would need to be rectified as well... How does the Doctor defend against these (what some would consider catastrophic) failures in research protocol? If I am off base here, sorry.
I was under the impression that BMW's series 5 and up have a form of Windows (I thought CE) in them already. I was also under the impression this has been going on for 3 years. It is one of the reasons the 7 series has had problems, such as locking drivers in thier own car. They have a tendency to shutdown for no reason. The problems relate to the computer chip and software used in the cars...