
06-15-2021
plf515
Lapis Lazuli | Level 10
Member since
06-23-2011
- 191 Posts
- 83 Likes Given
- 6 Solutions
- 63 Likes Received
About
I am a statistical consultant for researchers in the social, behavioral and medical sciences.
-
Latest posts by plf515
Subject Views Posted 928 06-11-2021 05:16 PM 3275 01-25-2020 07:49 AM 864 01-22-2020 07:59 AM 1846 01-22-2020 07:42 AM 3595 01-21-2020 02:37 PM 1362 01-16-2020 09:40 AM 3829 01-16-2020 09:39 AM 1517 01-07-2020 08:55 AM 677 01-02-2020 03:35 PM 622 12-31-2019 11:49 AM -
Activity Feed for plf515
- Got a Like for Meta analysis. 11-25-2021 05:37 AM
- Liked Re: How to put SAS on Demand for Academics in an icon on my screen? for Reeza. 06-15-2021 07:25 AM
- Liked Re: How to put SAS on Demand for Academics in an icon on my screen? for Reeza. 06-15-2021 07:25 AM
- Posted How to put SAS on Demand for Academics in an icon on my screen? on SAS Studio. 06-11-2021 05:16 PM
- Got a Like for Re: How to use correlations in PROC CALIS for confirmatory factor analysis?. 11-06-2020 12:11 PM
- Got a Like for How to use correlations in PROC CALIS for confirmatory factor analysis?. 11-06-2020 12:03 PM
- Got a Like for Meta analysis. 08-11-2020 03:13 PM
- Got a Like for Meta analysis. 07-27-2020 05:41 AM
- Got a Like for Re: Nominal outcome variable and Categorical independent variable. 04-08-2020 05:51 PM
- Got a Like for Re: What part of The Quad are you looking forward to visiting?. 02-20-2020 12:01 PM
- Got a Like for Re: What part of The Quad are you looking forward to visiting?. 01-30-2020 02:49 AM
- Got a Like for Re: SAS GF 2020 Sooo...What sessions do you have in your agenda?. 01-29-2020 11:20 AM
- Got a Like for Re: What part of The Quad are you looking forward to visiting?. 01-28-2020 02:07 PM
- Got a Like for Re: What part of The Quad are you looking forward to visiting?. 01-25-2020 12:50 PM
- Got a Like for Re: What part of The Quad are you looking forward to visiting?. 01-25-2020 10:16 AM
- Posted Re: What part of The Quad are you looking forward to visiting? on SAS Global Forum 2020. 01-25-2020 07:49 AM
- Liked What part of The Quad are you looking forward to visiting? for AaroninMN. 01-25-2020 07:44 AM
- Got a Like for Re: SAS GF 2020 Sooo...What sessions do you have in your agenda?. 01-23-2020 10:36 AM
- Liked Re: In a CATX function, how can I use a paragraph break as a separator? for Amir. 01-22-2020 08:24 AM
- Posted Re: Structural Equation Modelling on Statistical Procedures. 01-22-2020 07:59 AM
-
Posts I Liked
Subject Likes Author Latest Post 1 1 10 1 4 -
My Liked Posts
Subject Likes Posted 1 09-05-2019 05:24 PM 1 09-05-2019 04:17 PM 1 05-01-2015 10:03 AM 5 01-25-2020 07:49 AM 4 01-21-2020 02:37 PM
02-18-2018
12:17 PM
If you examine the output data set, you will see that you have the pieces that you need. It is just a matter of extracting the right pieces and re-assembling them properly. Here is a starting point:
proc transpose data=ken.days prefix=mean_group out=want;
id group;
var mean;
by time;
where _type_=3;
run;
Take a look first at KEN.DAYS to understand that the WHERE statement in PROC TRANSPOSE is selecting the proper observations.
Then take a look at the output data set WANT which will need a few more calculations, but has the data assembled in an easy-to-calculate form.
... View more
01-05-2018
09:16 AM
2 Likes
You can find the formula for the t-statistic here . The formula for C-hat can be found near the bottom of this page.
When looking at s.e.'s of parameters that have very small estimates, you can get into a situation where the calculation of that s.e. just becomes too treacherous. That's when you will see that 0 estimate for the s.e.
in the bigger picture, optimization algorithms perform better if the final parameter estimates for all parameters on the same scale. If your parameter estimates are more than a few orders of magnitude apart, then you may want to rescale one of the offending variables to bring it's final parameter estimate more in line with the others in your model.
... View more
01-17-2017
08:45 AM
1 Like
Yes, conditional and marginal interpretations are different. The drawback to the GEE approach seems to lie primarily in the area of covariance structure selection, using tools such as change in log likelihood or information criteria. Because the marginal approach uses pseudo-likelihoods, the "data" are not the same under different covariance structures. In contrast, the conditional approach uses a quasi-likelihood approach, so that information criteria can be used to select a best fitting covariance structure.
Steve Denham
... View more
11-01-2016
05:26 PM
Thanks Steve. I will look for it Peter ##- Please type your reply above this line. Simple formatting, no attachments. -##
... View more
06-09-2016
12:11 AM
1 Like
I guess you can't . But you could try proc sgplot .
... View more
05-11-2016
07:20 AM
Logistic regression is the way to answer these questions. The DV would be "answered" and the IV would be the reward for the first question. For the second you would add variables about times called and days since calling.
... View more
05-09-2016
01:16 PM
1 Like
If V2 is a class variable, you can use a SERIES statement with GROUP= option:
SERIES x=v1 y=pred / group=v2;
However, if V2 is continuous, this won't work so well. Instead, use the EFFECTPLOT statement to get what you want. It sounds like you might want to use the SLICEFIT option with the SLICEBY= suboption, as shown in the second example that I linked to.
If your regression procedure does not support the EFFECTPLOT statement directly, you can STORE the model to an item store and use PROC PLM, which does support the EFFECTPLOT statement.
... View more
04-07-2016
03:53 PM
1 Like
If the queues for each subject are long enough, those PROC FREQs should also give you some pretty good estimates of Stuart's tau-c and it's confidence bounds. I would like those intervals as a measure of agreement between request and response (I still don't know hwat the right vocabulary for these would be).
Steve Denham
... View more
03-30-2016
08:11 AM
1 Like
Hi Peter,
With the TOST option, the H0 option specifies the upper and lower equivalence bounds, so I would include it.
The one thing I would consider here is that the tests are not independent. It seems to me that there is likely a pretty strong correlation over time within subjects. I think that is what Xia is getting at in his comments.
The problem is that there is no real straightforward analog of Schuirman's approach when constructing simultaneous confidence intervals. The best I can think of is to use something like PROC MIXED, use the LSMESTIMATE statement for the three comparisons with appropriate ADJUST=, CL, and ALPHA= options. As far as the adjustment method, ADJUST=SCHEFFE strikes me as the best for confidence bounds.
Steve Denham
... View more
03-03-2016
01:06 PM
Hi,
Thanks a lot for your useful advice. Sorry. Yes, I meant the residuals were not normally distributed not the response. I did check the residual plots generated from GLM and they were skewed.
For the recoding of negative scores, actually, there was more reason for that but was not described fully. I agree that it is not a good idea to change the data in any way to fit our need.
For the non-parametric multivariate analysis models, thanks for sharing that paper. I am going to have a look at it.
Thanks,
... View more
03-01-2016
08:30 AM
@plf515 I agree partly ...and also disagree partly... Stepwise method have tendency to include too many variables. But, if it is clear when the results is reported that the associations was found by model section and not by testing well defined hypthoses, then there is no problem. Or, if the variable selection is done on a training dataset to generate hypothis, which then is tested on an other dataset it is also a valid approach.
... View more
02-24-2016
07:11 AM
Note that the standard errors in those cases is 0, so the calculation will involve division by 0. Without your data, it's impossible to say what is causing this, but it could be perfect collinearity, I think.
... View more
02-23-2016
06:57 PM
I guess that makes sense. I will try the graphing and see what comes out of it. Thanks!
... View more
02-15-2016
03:03 PM
2 Likes
I really like Jacob's approach.
Here is something a little simpler, along the lines of a Helmert contrast. You do have to expand this out, as I have only put in the time coefficients, so this would be for the first treatment arm:
LSMESTIMATE time*treatment 'pre vs all follow' -5 1 1 1 1 1 divisor 5,
'time 1 vs follow' 0 -4 1 1 1 1 divisor 4,
'time 2 vs follow' 0 0 -3 1 1 1 divisor 3,
'time 3 vs follow' 0 0 0 -2 1 1 divisor 2,
'time 4 vs follow' 0 0 0 0 -1 1/joint;
Additional arms, and comparisons between them follow from this.
Steve Denham
... View more
02-02-2016
11:25 AM
3 Likes
Luke,
Try adding an lsmeans statement to your model, to see if you get proportions out that you can understand. Also, using by subject processing can greatly improve your chances of convergence. I have rewritten your random statement to take advantage of this.
proc glimmix data=rcb ;
class site species mulch herb seedling ;
model Alive (event='1') = mulch herb species mulch*herb mulch*species herb*species mulch*herb*species/s
dist=binary link=logit;
random intercept mulch herb species mulch*herb mulch*species herb*species /subject=site s; lsmeans mulch|herb|species/diff ilink;
run;
Now, i would also suggest that perhaps your random statement probably overspecifies the number of random effects. You could probably get by with:
random intercept/subject=site;
Also, you should consider some points brought out by Walt Stroup in his book Generalized Linear Mixed Models (get a copy)--that the estimates and tests from this method tend to be biased, and that you may want to consider adding METHOD=LAPLACE to the PROC GLIMMIX statement.
Steve Denham
... View more