The Parameter Estimates table is reporting p-values for a test where the null hypothesis is that one regression coefficient is zero. In the Type III tests, the null hypothesis is that a certain linear function of the regression coefficients is zero. Sometimes these two tests overlap, i.e. the linear function being tested is that just one coefficient is zero, in which case the Wald test statistic is calculated similarly.    The "Four Types of Estimable Functions" chapter in the SAS/STAT documentation provides some relevant information about the Type III tests. In particular, the very last line in the "Estimability" section covers the form of the Wald test statistic and the "Type III Estimable Functions" section covers how a hypothesis for the test is constructed. I will admit, whenever I have to refer to the material about constructing the tests I have to read it (at least) two or three times before it starts to sink in.    As to why the tests using the Wald statistics versus the generalized score statistic differ in your example, I'm not sure there's really much I can say. The number of clusters is not small but not so large to make differences between the tests surprising.    ... View more

First I want to note that if you are using the same data to both select the model by backwards selection and then fit the model with the selected predictors, that introduces a post-selection inference issue that violates the assumptions underpinning the inference that PROC GENMOD performs for the selected model. Appropriate methods for performing post-selection inference is an active area of statistical research. Data splitting is always a valid approach if there are enough data.   To your original question, for GEE models by default the Type III tests are conducted by using the generalized score statistic. As described in the “Generalized Score Statistics” section of the PROC GENMOD documentation, this test statistic involves solving a restricted GEE model (restricted to match the null hypothesis of the test) and evaluating the generalized estimating equations at the restricted solution. This test statistic is typically preferred over a Wald based test statistic. You can request that PROC GENMOD conduct the Type III tests with the Wald test statistic by using the WALD option. Using the Wald statistic might be preferable when the data have a very large number of clusters. In that situation computing the generalized score statistics can be computationally intensive and the differences between it and the Wald test statistic should be minimal.   The LSMEANS statement computes predicted population margins, which is a different task then then testing the Type III contrasts. For the predicted population margins computation there is not an analogue to the generalized score statistic used to test the Type III contrasts. ... View more

I think it might be worth looking more into Rick's comment about checking if the data sets are exactly the same in the two runs. In the output you attached the two "Node Information" tables report different "Pred" values for the Training and Validation data for the Root Node (0.00191 and 0.00195 versus 0.00200 and 0.00182). I believe that would indicate that the two runs of PROC HPSPLIT are working with different training and validation partitions leading to the different results.    In terms of why the data sets might be different, in addition to (or in conjunction with) possible "bitness" and host differences, the ordering of the input data provided to PROC SURVEYSELECT will effect the simple random sample even if the same seed is provided (see below).     data exData; do i=1 to 1000; x = rand("uniform"); output; end; run; /* Sampling for data ordered by i */ proc surveyselect data=exData out=exData_byi seed=123 samprate=0.6 method=srs outall; run; proc sort data=exData; by x; run; /* Sampling for data ordered by x */ proc surveyselect data=exData out=exData_byx seed=123 samprate=0.6 method=srs outall; run; /* Different Samples */ proc compare data=exData_byi c=exData_byx; run; proc sort data=exData; by i; run; /* Sampling for data ordered by i again */ proc surveyselect data=exData out=exData_byi2 seed=123 samprate=0.6 method=srs outall; run; /* Same Samples */ proc compare data=exData_byi c=exData_byi2; run;       ... View more

You can use the ORDER= option in the PROC GLM statement to control how the levels of classification variables are ordered. The default ordering is by the formatted values, or internal values if no formats are provided. In this case that does put April before January. If your data are already sorted in the order of the variable Month that you want used, you can specify ORDER=DATA, in which case the variable levels are ordered how their appearance in the input data set.  ... View more

Yes, here is an example in a previous communities post that creates footnotes and titles both in and outside of the graph walls. The example uses the SGANNO= option in PROC SGPLOT to add a title and a footnote within the plot window using an annotation data set.   For more examples of using annotation, there is this SAS Global Forum paper or Chapter 4 in this book.      ... View more

Looking at the documentation the ODS WORD statement, the title specified in the TITLE= option appears in the document properties. Using the example you provided I see      I think instead you might want to use the TITLE and FOOTNOTE statements along with the NOGTITLE and NOGFOOTNOTE options in the ODS WORD statement to move the title and footnote out of the graph borders.    So for example something like: ods word file="H:\Boxplots.docx" style=mystyle nogtitle nogfootnote; title 'Boxplots of clinical chemistry'; footnote 'footnote text'; proc sgplot data=cc8; vbox ch /group=arm; label ch="Change from baseline (unit)"; refline 0; run; ods _all_ close;   ... View more

You might want to follow up with Technical Support at SAS regarding this issue. If I recall this was a bug in SAS 9.4M3 (SAS/STAT 14.1). The information in the Parameter Estimates table is correct, there was an error in the response profile message and the REF= option as a response variable option was not supported. These issues were fixed starting in SAS 9.4M4 (SAS/STAT 14.2). ... View more