Ref:  Deddens JA, Petersen MR, Lei X.  Estimation of prevalence ratios when PROC GENMOD does not converge.  SAS Users Group International 28, paper 270 at http://www2.sas.com/proceedings/sugi28/270-28.pdf. This reference provides a method to estimate prevalence ratios and predicted probabilities within proper bounds with PROC GENMOD.  A later citation to this reference suggested that one could achieve the same results by using FREQ statement instead of copying the observations multiple times. ... View more

Binomial regression (LINK=LOG) differs from logistic regression (LINK=LOGIT) in that binomial regression uses as its measure of effect the relative risk, the ratio of two probabilities, and that logistic regression uses as its measure of effect the odds ratio, the ratio of two odds.  When the reference probability [=denominator of the relative risk] is larger (say > 0.20), then the odds ratio can be much larger or much smaller than the relative risk (deviating away from the null value of 1.00).  For comparison of prevalences, this exaggeration of the odds ratio is somewhat disconcerting.  For example, if the prevalence in one group is 0.7, and the prevalence in another group is 0.4,     the relative risk = 0.7/0.4 = 1.75, and     the odds ratio = (0.7/0.3) / (0.4/0.6) = 3.50. For the general public (who are not gamblers), it is also easier to explain probabilities and relative risks than odds and odds ratios. The disadvantage of binomial regression is that, sometimes, it may estimate probabilities for observations that exceed 1.00, but methods have been developed to constrain these estimated probabilities to be less than 1.00.  With logistic regression, estimated probabilities can never exceed 1.00. ... View more

Yes. You can include the same MODEL statement option, OFFSET, and change the MODEL statement option, DIST, from DIST=POISSON to DIST=NEGBIN:   model missed = x / dist=negbin link=log offset=ln; However, your original denominator, TOTAL, is relatively small compared to your numerator, MISSED.  I would also try binomial regression, without an offset:    model missed/total = x / dist=binomial link=log; Why have you omitted the classification variable, YEAR, from your model? ... View more

Modelling the Poisson distribution in PROC GENMOD requires both the numerator and the denominator of the proportion. In your example, the variable, Y, already has been calculated as a proportion [=# of missed appointments/ (# of total appointments)].  You should take the natural logarithm of the denominator, the number of total appointments, as your offset and the number of missed appointments as your dependent variable. However, because Poisson regression usually assumes a small numerator relative to a large denominator [and thus a small proportion close to zero], you might instead consider binomial regression with MODEL statement options of DIST=BINOMIAL and LINK=LOG without the need for an offset. ... View more

My suggestions about your earlier questions are in the attached .TXT document. The LSMEANS estimates for treatment are negative because of your use of the logit link, where probabilities between 0 and 1 are specified as logits, ln(p/(1-p)), which can be less than zero when the estimated probability is less than 0.50 [for example, if p=0.40, ln(0.4/0.6) ~= - 0.41]. ... View more

Read the documentation for the PARMS statement in PROC MCMC (available at the SAS Internet site).  To specify multiple initial parameters using the PARMS statement, precede this statement with an ARRAY statement that specifies the number of parameters you want to initialize.  The documentation shows the following example:     ARRAY mu [3];     PARMS mu {1 2 3}; Note that brackets are used in the ARRAY statement and that braces are used in the PARMS statement. ... View more

For random effects like PROCEDURE, use the following version of the RANDOM statement in PROC GLIMMIX:     random PROCEDURE / subject=ID; For a repeated measure effect like LOCATION, use the following version of the RANDOM statement in PROC GLIMMIX:      random LOCATION / subject=ID residual; The RESIDUAL option of the RANDOM statement in PROC GLIMMIX converts the effect from a "G-sided" random effect to an "R-sided" repeated measure. Fixed effects like TREATMENT should be specified as independent variables in the MODEL statement (to the "right" of the equals sign). The first MODEL statement you are puzzled about above,       model Y=a b(a); specifies as independent variables a main effect, a, and a nested effect, b(a), in which the factor, b, is nested within a.  Note that this SAS model syntax implies that b does not have a main effect term, independent of its association with a.  This MODEL statement also has the equivalent SAS syntax of       model Y=a b*a; The second and the third MODEL statements you are puzzled about above,       model Y=a b a*b; and       model Y=a | b; are equivalent in SAS, with the independent variables in the third statement, a | b, being an abbreviated notation for those in the second statement,  a  b  a*b.  Both these MODEL statements specify as independent variables two main effects, a and b, and their interaction, a*b. ... View more

Read the following article:    Prentice RL, Breslow NE.  Retrospective studies and failure time models.  Biometrika 1978 Apr;65(1):153-158. Matched case-control samples correspond to risk sets in proportional hazards models. ... View more

Look at the SAS documentation for PROC IMPORT.  Specifying a larger value in its new option, GUESSINGROWS, may solve this problem for you. ... View more