Errors are in Y Collinearity is in X > so then I take it that the computation of vif does > not need the assumption of uncorrelated errors? VIF does not depend on Y in any way ... View more

Collinearity is collinearity, regardless of what SAS PROC you will be using to estimate your model. You can obtain VIF from PROC REG using the fixed terms in your PROC MIXED model. ... View more

> In the manufacturing world, I too often see engineers > use regression to understand engineering. It's a bit > like making soup where one just keeps adding > ingredients until the taste goes bad. A cook who > knows his/her stuff chooses ingredients that work and > then adjusts the quantities to maximize the benefit. > Similarly, variables should be selected on the basis > of (engineering) knowledge. Then use the regression > model to determine the correct magnitude/effect for > the variables chosen. Ah, the exact opposite of the empirical approach. And what would you do when you are in a situation where there isn't a lot of engineering understanding of the situation? What would you do when the process moves in ways you have never seen before, and your engineering knowledge can't explain why it did that, but you have a lot of data? What would the cook do in your analogy, when presented with ingredients that he has never seen before? ... View more

You can do this entirely with macros. You need a macro variable containing the original data set number, and another macro variable containing the number its supposed to change into. Side note: Seems kind of pointless to me to change data1 to data3, maybe its just a greatly simplified example. However, > Suppose I have 3 data sets: data1, data2, data3 and I > want to rename then data3, data5, data7. I can do > this with proc datasets: > > proc datasets; > change data1=data3; > run; This code fails, you cannot change data1 to data3 because there is an existing data3. ... View more

Since different statisticians have different ideas, and do not always agree, let me point out that I would not recommend stepwise regression, as it has known problems. Furthermore, deleting the insignificant terms may have a minor impact on the quality of the model or a major impact; you just don't know. You may get a significant decrease in the R-squared, or an insignificant decrease. You may get a significant increase in the Adjusted R-squared, or an insignificant increase. But in the end, I would not go about improving the model by deleting terms. I would examine the residuals, to see if there are indications of curvature or non-linearity. Should there be such indication, I would ADD curvature terms to the model to improve it. ... View more

These variables should not be in the CLASS statement in PROC GLM. Then they will be treated as continuous and your degrees of freedom problem should go away. ... View more

You have an overdetermined model. In other words, too many terms in the model for the number of data points available, which results in all of these issues, 0 degrees of freedom for error, no P-values, no F-values and 100% r-squared. I don't know what GDP is to you, but are there really 93 discrete levels (and is it a categorical variable)? Same question applies to POP. ... View more

Can you show me an example of the output? ... View more

Generally, you decide on the appropriate model without looking at the data. You use your knowledge of the system that you are studying to say. For example: youd decide you want main effects for B, C and D, and the B*C interaction and the B*D interaction, but not the C*D interaction. This is not something I can help with, you or someone you work with should be able to specify this. Once that is done, you fit the model, all desired terms. At this point, statisticians disagree; however my advice is to leave all terms in the model. Evaluate the model by examining the residuals. If there are problems in the residuals, you may need to modify the model. As far as your problem where there is no output, you'd have to tell us more, or show us the listing and/or log file. ... View more

And after you follow polingjw's advice, keep your fingers crossed that the predicted values mean anything. If you really have 202 dummy variables, I'm skeptical that any such prediction of out of sample values will have meaning. More than likely, you will be extrapolating, in other words, predicting in a region of X-space where you have no data. ... View more

which can be shortened to data nomiss; set weight1; if cmiss(of _all_) then delete; run; ... View more

I don't really have a simple example in that case. Probably you need to use ARRAYs. But I have a question for you. Why make an effort to delete rows with all missing values? Leaving them in the data set doesn't hurt anything, does it? ... View more

It would help if you give us an example. All I can tell you is that you are trying to use DATEPART on a character variable, and that causes an error. ... View more

look at the CHANGE command of PROC DATASETS ... View more

You can use the SUM function to add all the numeric variables in a row; if the result is missing, then all variables in the row are missing. data a; set sashelp.class; rowsum=sum(of _numeric_); run; You can add a new row with the OUTPUT statement in a SAS data step data a; set sashelp.class end=eof; output; if eof then do; age=14; height=56; weight=99; output; end; run; ... View more