About wernie

wernie · ‎01-20-2017

I'm trying to replace only part of a value for one variable, so it matches to merge with another dataset based on this Facility ID. Don't worry about the contaminant or value variables, I just left them in there for the example. So, I want to be able to change the facility name from WP001574 to WP531574, from WP001575 to WP531575, and so on. I just need to replace part of the value. How can I do this? Thanks! Have: data contam; input FacilityID FacilityName $ Contaminant $ Value; datalines; 1 WP001574 a 5 2 WP001575 d 10 3 WP001576 b 2 4 WP002475 c 8 5 WP002476 e 30 6 WP003843 a 2 7 WP003843 a 5 8 WP003844 b 8 9 WP003845 c 20 10 WP003846 d 30 11 WP003847 e 55 ; run; Want: data contam; input FacilityID FacilityName $ Contaminant $ Value; datalines; 1 WP531574 a 5 2 WP531575 d 10 3 WP531576 b 2 4 WP532475 c 8 5 WP532476 e 30 6 WP533843 a 2 7 WP533843 a 5 8 WP533844 b 8 9 WP533845 c 20 10 WP533846 d 30 11 WP533847 e 55 ; run;

wernie · ‎01-17-2017

Thanks @gamotte and @Haikuo - both sets of code seem to work for structuring the data in the way that I wanted it! 🙂

wernie · ‎01-17-2017

I was hoping to just generate a spreadsheet with one row per facility to quickly see which contaminants were tested for that facility and if there is a given value. I guess that could probably be summarised using proc report or something perhaps but wanted it in an easier to read format I could just export to Excel for those who don't use SAS. In this case, I don't think we'll be adding additional contaminants. Oh, and I remembered that I need to create a new variable based on the sum of some of those contaminants, so I also thought it would be easier if everything was 1 row per facility (although I could be wrong!).

wernie · ‎01-17-2017

Thanks. That didn't seem to work for me as each facility is still listed with multiple observations. I created a small sample dataset and then put the code below that where I separated out each contaminant into a new variable, giving it the value from the 'value' variable. I may not need that, but that was one of the first steps I took in trying to work towards collapsing the observations based on FacilityID. data contam; input FacilityID FacilityName $ Contaminant $ Value; datalines; 1 Harry a 5 1 Harry d 10 2 Samantha b 2 2 Samantha c 8 2 Samantha e 30 3 Abby a 2 4 Ned a 5 4 Ned b 8 4 Ned c 20 4 Ned d 30 4 Ned e 55 ; run; data Contam; set Contam; if Contaminant='a' and Value>0 then a_conc=Value; if Contaminant='b' and Value>0 then b_conc=Value; if Contaminant='c' and Value>0 then c_conc=Value; if Contaminant='d' and Value>0 then d_conc=Value; if Contaminant='e' and Value>0 then e_conc=Value; run;

wernie · ‎01-17-2017

Hi all, I am having a really tough time figuring this out for some reason! I have a bunch of observations, some with the same ID, that I need to collapse so there is just one observation per ID. I put an example below of what I currently have and what I want (I do have some other variables that are associated with each ID that I want to keep, like facility name, but didn't include those here). So, some facilities may only have data for one contaminant (ID=2) whilst others have data for all (ID=4). Thank you! Current: FacilityID Contaminant ContA_conc ContB_conc ContC_conc ContD_conc ContE_conc 1 a 10 . . . . 1 c . . 8 . . 2 e . . . . 50 3 b . 2 . . . 3 d . . . 12 . 4 a 75 . . . . 4 b . 5 . . . 4 c . . 1 . . 4 d . . . 25 . 4 e . . . . 40 Want this way: FacilityID Contaminant ContA_conc ContB_conc ContC_conc ContD_conc ContE_conc 1 a 10 . 8 . . 2 e . . . . 50 3 b . 2 . 12 . 4 a 75 5 1 25 40

wernie · ‎10-31-2015

When I remove those variables from the 'class' statement and only put them in the model statement, it runs and says algorithm converged, but the problem that I get is that area 1 v area 2*period gives an estimate and then the others says 'non-est'. I'm not sure what exactly makes it nonestimable or how to address that... Thanks again 🙂

wernie · ‎10-30-2015

Thanks, Reeza. So if I have gross measures of employment, which are repeated for each age group and sex (i.e., each age group and sex for the year 2000 has the same proportion employed FT number for area 1, then each age group and sex for the year 2000 has the same proportion employed FT number for area 2, and so on), then which model should I be using? Would it still provide the RRs like I have with this model and the estimate statements?

wernie · ‎10-30-2015

Hi all, Sorry if my terminology might not be right on point, but I am trying to add covariates into a negative binomial model to see if other factors might account for changes that were found in initial analyses. The overview is: I have 3 study areas with n cases for each study area for each year over a 17 year period. I am using the estimate statement to get a rate ratio to compare the slopes over time. The models are offset by the underlying population (logpop) to account for changes in pop over time. So, that code looked like this, with the area*period interaction being the outcome of interest: proc genmod data=All; class area; model nsum=area period area*period / dist=NB link=log offset=logpop type3; estimate "Area 1 v area 2*period" area*period 1 -1 0/exp e; estimate "Area 2 v area 3*period" area*period 0 -1 1/exp e; estimate "Area 1 v area 3*period" area*period 1 0 -1/exp e; run; But, now we want to add in some covariates, so I have age, sex, proportion of population employed FT (which is a % of the population for each year), proportion in specific occupations, median income, etc. So I have 17 measures for each covariate for across the time period (e.g., 17 different measures of proportion employed - 1 for each year). When I try to add these variables in the 'class' statement and in the 'model' statement, I get errors about negative of Hessian or model did not converge. I don't know if this is the proper way to do it or how I should add in the covariates to see if they affect the area*period RRs of interest. Any help would be appreciated! Thanks 🙂 PS - I just saw in another post that continuous variables should not be included in the class statement, so unsure if that's part of what I'm doing wrong.

wernie · ‎04-11-2015

Oh okay, thanks. I had come across that material before, but was confused for some reason by that table for the scale statement. I'm looking at negative binomial, poisson, or ZINB, so now it makes sense to me that I run whichever model first and then look at the dispersion parameter in the model and input that in the macro for scale1 and scale2= Thanks for your help!

wernie · ‎04-10-2015

I am trying to run the Vuong macro and I have everything set up, minus a few things. For nparm1 and nparm2 I just looked at the output for number of parameters and used '8' for both models. I am confused on the scale1= and scale2= statements though. In some of the reading I did, it says that it should be the 'variance of the outcome measure in each regression model'. My outcome measure (DV) is 'n', but I am not sure how to get the variance of the outcome measure for each of the models (poisson, NB, ZINB) and the IVs are area, period, and the interaction between area and period (which is the part I'm interested in). I figured out how to get the variance for each variable in the dataset, so I see the variance for 'n' is listed as ~2705, but I don't think this is what I should be inputting after scale1= and scale2= because the few examples I could find were all <10. Can someone please help me to sort out where to find this value? Thanks for your help!

wernie · ‎04-09-2015

Thanks for the reply. I ran the Vuong macro to see which test would be best and NB was suggested. So is this wrong to follow this or is it more just common sense that it would be better to use Poisson rather than NB for something like this? (Sorry, I'm not really familiar with knowing which model is best - I had just thought because it is modelling count data that is overdispersed that it would be NB.) I shouldn't have missing data, just some datasets that have a lot of zeros (which is also why I used the Vuong macro - to see if a zero-inflated model would be better). I will remove 'age' from the class statement then because it is not used in the model, only in a 'where' statement. I know at one point I saw the number of observations was only saying 612, but it should have been 7344 if all of the age groups were there. I think that is what you mean with aggregating the variable. So if I include age in the proc sql statement after importing the data / before running the model, then that gives me 7344 observations when I run the model (for all ages) excluding age from the class statement. I think that was one of my previous errors. Can you clarify something for me in terms of interpreting the mean estimate with the p-value? I know traditionally if the IRR is something like 1.3 it would mean that the exposed group has 1.3 times the cases than the reference group and if the CI does not cross 1 then it is significant. But, for example, in one of the models the IRR is 1.0028 (CI 1.0015 - 1.0042) p-value <0.0001. So this would only be 0.28% greater than the reference scenario and it is showing it as significant. But am I interpreting this incorrectly? I get similar results trying it with Poisson as well. Thanks again!

wernie · ‎04-08-2015

I am trying to calculate incidence rate ratios and 95% CIs for the interaction between two variables: area and time. My variables are n, area, period, and age (using log of population as the offset). I have two questions that tie together for the code I am using, which is: proc genmod data=inf; (where age=1.....depends if I need to specify age group or not. If I want all ages, then I just leave this statement out) class area age; model n=area period area*period / dist=nb link=log offset=logpop type3; estimate "Area 1 v area 2*period" area*period 1 -1 0/exp e; estimate "Area 2 v area 3*period" area*period 0 -1 1/exp e; estimate "Area 1 v area 3*period" area*period 1 0 -1/exp e; run; Q1 - I have 12 different age groups. If I include 'age' in the class statement, I get slightly different IRRs and certainly different p-values than if I do not include 'age' in the class statement. Should I be including age in the class statement for the model? There are times where I want to look at everything together (all-ages) and other times where I need to specify age groups that I want to look at, so I don't know if 'age' is needed in the class statement for each model or not. Q2 - I'm not totally sure how to be interpreting the output for the IRs and CIs. I understand these (e.g., how to interpret IRR of something like 1.3), but it isn't seeming to make sense when I'm looking at the output. For example, with the above code (NO age listed in class statement), I get the following: Area 1 v 2: IRR (1.0007) 95% CI (0.9999-1.0015) p-value 0.1043 Area 2 v 3: IRR (0.9985) 95% CI (0.9977-0.9994) p-value 0.0007 Area 1 v 3: IRR (1.0022) 95% CI (1.0015-1.0028) p-value <0.0001 This doesn't quite make sense looking at the p-value that goes with the IRRs and CIs. However, when I DO include age in the class statement for the same dataset, I get: Area 1 v 2: IRR (1.0006) 95% CI (0.9990-1.0022) p-value 0.4867 Area 2 v 3: IRR (0.9996) 95% CI (0.9979-1.0013) p-value 0.6697 Area 1 v 3: IRR (1.0009) 95% CI (0.9994-1.0025) p-value 0.2411 This seems to make more sense to me, as looking at the IRR and CI I wouldn't expect significant results. So does this mean that age should be included in the class statement? Thanks!

wernie · ‎02-04-2015

Thanks, Steve. I've been playing around with the LSMESTIMATE statement and have been reading up online, but am having difficulty translating this to my example of looking at an interaction between one variable (area) with 3 levels and another variable (period) with 17 levels. I tried to use LSMEANS to print out the matrix so I can see where the 0's and 1's fall (as it seems to me, but I could be completely wrong, that I need to input what I am trying to examine in the LSMESTIMATE, but I am very confused on how to do so with the numbers following the variables I specify), but then when I did a test run of code, it said that 'The level 0 is not valid for CLASS variable period. The level specifications for this variable must range from 1 to 17'. The code I tried is as follows: proc genmod data=want; class area period; model nsum=area period area*period / dist=NB link=log offset=logpop type3; lsmestimate area*period **I had tried something like [1 0 0] [0 1 0] [0 0 1] and that gave the error - do not understand how to put in the numbers to examine the area*period interactions** run; Hoping someone can help with a start on what I need to follow that 'lsmestimate area*period' statement. Thanks!

wernie · ‎02-02-2015

Hi Steve, Thanks for your help. I do understand what the significant interaction means - I had previously plotted everything prior to running through in SAS and was planning on using the stats to confirm what looked like it would be significant from my plots. So I guess my question then is if I should be looking at the type3 p-value for the interaction or if I should be looking through at all of the estimates and p-values for each year/area in comparison to the reference area? I would think it's the type3 p-value for the interaction that I should be looking at for the overall picture; however, someone had told me not to focus on the p-values. So I'm not entirely sure what I should be looking at/reporting. Thanks again!

wernie · ‎01-30-2015

I have hospitalization data (nsum) over a 17 year period for 3 different study areas (variables period and area). I have the log of the population as the offset. So, for my code I have entered: proc genmod data=want; class period area; model nsum=area period area*period / dist=NB link=log offset=log pop type1; run; I am not really interested in estimates just for area or just for period, rather, I am interested in the difference between the areas over the time period. In the output for the interaction of area*period, there are 3 rows for each year in the 17 year period so I have area 1, area 2, and area 3 (reference group). My question is - how do I interpret this output? Is there a way to broadly look at if the 3 areas differ significantly over this time period or do I have to look at the p-values for each year for areas 1 and 2 compared to the reference group? I also have the LR statistics for type 1 analysis, but I'm not really sure what this is telling me when I look at the p-value for the area*period term. Thanks for your help!

Online Status	Offline
Date Last Visited	‎04-28-2020 03:20 PM

Extracting word(s) before comma delimiter

Re: Transforming a dataset and calculating variables with dates

Transforming a dataset and calculating variables with dates

How to create new dataset with daily dates for each state?

Re: How to add value from variable of one observation to another obser...

Re: How to add value from variable of one observation to another obser...

How to add value from variable of one observation to another observati...

Add leading 0 to census tract value (character var)

Re: Need help merging two files to fill in values

Need help merging to expand dataset

Re: Extracting word(s) before comma delimiter

Re: How to create new dataset with daily dates for each state?

Re: How to create new dataset with daily dates for each state?

Re: How to add value from variable of one observation to another obser...

Re: How to add value from variable of one observation to another obser...

Re: How to add value from variable of one observation to another obser...

How to replace part of variable values?

Re: Help: Collapsing multiple observations with different variable val...

Re: Help: Collapsing multiple observations with different variable val...

Re: Help: Collapsing multiple observations with different variable val...

Help: Collapsing multiple observations with different variable values

Re: Adding and interpreting covariates in proc genmod (NB) model?

Re: Adding and interpreting covariates in proc genmod (NB) model?

Adding and interpreting covariates in proc genmod (NB) model?

Re: Vuong macro - scale= ?

Vuong macro - scale= ?

Re: IRRs in proc genmod

IRRs in proc genmod

Re: How to interpret proc genmod (NB) output?

Re: How to interpret proc genmod (NB) output?

How to interpret proc genmod (NB) output?