Dennis,
"Dennis Roberts" <[EMAIL PROTECTED]> wrote in message [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > > > if the circumstances are such that the assumptions and conditions are good > for CR ... then by all means use it This is a brave admission, if you mean use CR. I would not say that data of no use to CR should not be used for other purposes, however. > > if the circumstances are such that the assumptions and conditions are NOT > good for CR ... don't use it Do not use the data with CR, but it may be of use in other methods.. > > don't force data onto a model ... and don't force a model, onto the data True, but realize you are free to collect and to modify data. If you have a data set and you know every other value is bogus. Then delete the bad data. If you have data that poorly reflects the variable in the extremes, delete the bad sections of the data. Data is a collection of observations. They are not a unit which must be accepted all or none. Accept the true, reject the false. > > now, if you want to play around with data and models just to see what > happens, that's great and, it can be fun Yes, but it can also be very useful scientifically. > > BUT, if you do force a model onto the data AND, there is substantial misfit > ... don't blame the data ... Forcing a model means to lie about the nature of the data or model. The data is just what it is. We should not blame data, only the person who tries to misuse or lie about its nature. When I say the data is bad, I only mean that it is being treated as though it were something that it is not. For example, using normally sampled variables in Steve's so called tests of CR was bad because he knew the assumption was that the causes were uniform/factorial. He hide the normal nature of his causes. Similarly, Gus will not reveal the nature of his data. This is deceptive. The data in all casses are simply what they are. The misuse of the data make them "bad" for one or another purpose, for example, as measns of testing CR. Trimming the data and the residual, might have been a good way of transformingSteve's inappropriate data to some thing that would be of use in testing CR. I have no idea what Gus' data is. Just about everyone who does the simulations agrees that CR works with such perfect data. This is more than we can say with TETRAD or LISREL. People do not have to buy into the idea that causation is combination, nor should they block the communication of this idea. Certainly no one should defraud anyone to sell more books and software and inferior methods. But they do and I am peerless in my fight against this fraud. My motivation comes from my heart and mind and from the memory of sweet child who lived for a while because good doctors and scientists worked hard and told the truth. She was the first to survive her congenital heart condition. There were NO previous incidences of survivors with her condition. Theory said it was possible in the mind of her doctors. Theory and hardwork saved her life. There is nothing any of the rotten boys on semnet or on this newsgroup can say or do to convince me that science is not a worthy pursuit. They can rot in hell for stealing the resources that could be helping children like my Gracie. I will never give in to their kind of abuse. NEVER. But I can be convinced that CR is wrong. Just show me as a scientist. Best, Bill > > look for a better model > No, look for the answer to your scientific question. Data is not God. Nor is CR. Neiother data or method are the reasons for science. KLnowledge is the reason for science. The construct of causation is one valid and important means of gaining knowledge. With regard to causal inference, we need better data than what is typically collected in survey type research. Many multivariate statisticians have seen the advantages of simple structure. I am not new in this regard. Many experimentalists have seen the value of factorial designs. I am not new in this regard. I am new in offering a valid means of inferring causation without experimentation. Suppressing CR by requesting, accepting and holding a paper for three years... so your books on an alternative and inferior method (TETRAD) can sell, and then rejecting that paper without a scientific reason, is fraud. That fraud was the origin of this thread. Fraud was attempted on this newlist. Petty little egoes and moral depravity allow criminals to market themselves as experts and to control the works of true scientists. They tried to reject corresponding regressions without bothering to test it as scientists. Many have done this before this thread. All have failed. But maybe you an I at least are making progress in this exchange. Bill > . > . > ================================================================= > Instructions for joining and leaving this list, remarks about the > problem of INAPPROPRIATE MESSAGES, and archives are available at: > . http://jse.stat.ncsu.edu/ . > ================================================================= . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
