cancel
Turn on suggestions
Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.
Showing results for 
Search instead for 
Did you mean: 
BlueNose
‎09-08-2018
  • 190 Posts
  • 36 Likes Given
  • 0 Solutions
  • 4 Likes Received

Re: GLM with data that doesn't follow the normality assumption

by Quartz | Level 8 in Statistical Procedures
‎09-08-2018 12:02 PM
‎09-08-2018 12:02 PM
If the errors follow the normal distribution, doesn't it mean that Y will also be normal ? It is easier to check Y.   Does PROC ROBUSTREG support something like LSMEANS ? ... View more

GLM with data that doesn't follow the normality assumption

by Quartz | Level 8 in Statistical Procedures
‎09-08-2018 04:59 AM
‎09-08-2018 04:59 AM
Dear all,   I wish to model a dependent variable Y, continuous, vs. a categorical independent variable X, along some covariates. I want to produce lsmeans and test for differences between all pairs of categories of X.   Y doesn't follow the normality assumption. Some values are 0 so I can't use the log transformation. I used the square root instead. It has improved things a bit, but it is still not normally distributed.   Is there a way in PROC GLM to get something like robust standard errors ? Is there an alternative, a procedure that is robust and produce clean lsmeans with tests ?   Thank you ! ... View more

Re: Converting Wide to long with two sets of variables

by Quartz | Level 8 in SAS Programming
‎09-07-2018 03:39 PM
‎09-07-2018 03:39 PM
Thank you !   Will it work if I have more IV's than DV's ? The loop runs until 4 only. ... View more

Converting Wide to long with two sets of variables

by Quartz | Level 8 in SAS Programming
‎09-07-2018 11:58 AM
‎09-07-2018 11:58 AM
Dear All,   I have a dataset with 4 dependent variables, 6 independent variables and 9 covariates.   For each combination of dependent and independent variable, I wish to run a model with the specific independent variables alongside the covariates (let's ignore the significance level issue for now). In order to avoid "copy+paste" of the same procedure 24 times, I was thinking to transpose my dataset to long format and use the by statement, I'll be more specific:   My data look like this:     And I want it like this:     So I can run a procedure such as GLM and do it "by DV, IV".   I know how to use PROC TRANSPOSE when I have one set of variables which I want to transpose from wide to long. But two...can't figure out how to do it. I wish to avoid the dirty work of doing set and where 24 times. Is there a way to do it ?   Any assistance will be most appreciated !   Thank you. ... View more

Aggregating and Plotting Quality Control Data

by Quartz | Level 8 in SAS Programming
‎08-24-2018 10:57 AM
‎08-24-2018 10:57 AM
Dear SAS Users,   I have a complex question, it involves mainly programming, but also a little bit of graphics and methodology, so I chose this board, I hope it's okay !   I have quality control data coming from a lab (photo attached).   The lab has two rooms. Each room has several areas. Each area has several locations within this area. For each location within area within room, two different bacteria tests are being made. Each has a maximum allowed limit. The test was performed 3 times a day for 16 days in a row. Each measure is the counts of bacteria found.   I have this data organized in a SAS dataset, a "long file", the way SAS "likes it" 🙂   I wish to do the following tasks, and I am not sure how to:   1) Aggregate - calculate the average count of bacteria per day, so I end up with a daily reading rather than 3 of them.   2) Graph the averages - 16 points. The x-axis is the date, the y-axis is the average. In addition, I wish to add a horizontal line with a value coming from the "Limit" column. This task should be done BY room / area / location / type (limit will vary by this combination). The data is already sorted for the by statement to work. Dots should be connected by a line.   3) In addition to the stated graph above, I thought of using a control chart, so I can see / test for trends in the data. However, non of the control charts I know seems to fit (Xbar, C, U, P, NP). The only one that might work is the Xbar, but I am not sure it's correct to use it. Which control chart shall I use, and is it possible to plot, for example, the Xbar chart (which aggregates the daily data I think), and to add another horizontal reference line to the plot, again, doing it by all the variables ?   I have a feeling that these tasks are fairly easy, if you know how to....   I am using SAS 9.4.   Thank you in advance for any tips and assistance you can provide. Without using SAS and the by statement I will have quite a lot of work doing this manually....     ... View more

Re: Time Series Data with both Daily and Monthly Variables

by Quartz | Level 8 in SAS Forecasting and Econometrics
‎08-14-2018 03:29 PM
‎08-14-2018 03:29 PM
I see what you mean, but even if I group it, the same problem remains, which is that for each day within the month, the income will be the same, a constant within a month. So the temperature and number of people vary by day, while income by month.     ... View more

Time Series Data with both Daily and Monthly Variables

by Quartz | Level 8 in SAS Forecasting and Econometrics
‎08-14-2018 04:51 AM
‎08-14-2018 04:51 AM
Dear all, I have daily data on how many people entered a certain shopping center, and the weather on that day (temperature). I wish to find out if there is a relation between the weather and the number of people who entered the shopping center. In addition, I have covariates such as the average income in that region.   Number of people entering the shopping center (daily) Temperature (daily) Mean monthly income in the region (monthly)   The problem is, the covariates, such as mean income, are monthly, not daily. So for my main dependent and independent variables, I have a daily time series, while the covariates are monthly. How should I handle this situation ? I thought of several options, not sure which is best: 1. Aggregate the daily variables using means, to make them monthly - I will lose information 2. Make the monthly data daily, i.e., for each day in this month, the income will be the same. This will lead to a model with random effect, won't it ? How would you handle this problem and which model would you use ? (regression, time series, mixed model) Thank you in advance !   (Using SAS 9.4) ... View more

Setting up a general linear model with multiple comparisons or contras...

by Quartz | Level 8 in Statistical Procedures
‎08-03-2018 10:50 AM
‎08-03-2018 10:50 AM
Dear all,   I have data containing one dependent variable, I'll call it Y for simplicity, and let's assume it is normally distributed.   I have independent variables X1, X2, X3, X4. All of them are ordinal, getting values of 1-4, where 4 is highest and 1 lowest. In addition, I have some covariates, such as age, gender, etc...   I wish to model Y vs. the independent variables and the covariates, but I am mostly interested in finding if there are significant differences between groups 1 and 4 of each independent variable.   Using the following code:   PROC GLM DATA = Dat; CLASS X1; MODEL Y = X1 Age Gender ; LSMEANS X1 / cl pdiff diff; RUN; Gives me a linear model, with multiple comparisons of all pairs of the variable X1.   Is there a better way to model in this situation ? Is there a way to define something like a contrast, so only groups 1 and 4 will be compared with higher power ? Is my code correct in the first place ?   Thank you in advance ! ... View more

Setting up a mixed model

by Quartz | Level 8 in Statistical Procedures
‎07-18-2018 06:45 AM
‎07-18-2018 06:45 AM
Hello all,   I have a sample containing data on 400 school students.   For each student I have 3 dependent variables: Y1, Y2, Y3   In addition, for each student I have several independent variables: Age, Gender, ....   The students come from classes. Some students share the same class, so for each student I have the class as another independent variable.   In a different file, I have teacher data, such as teacher age, teacher gender,...   I need to try and find out which variables relate, or affect Y1, Y2 and Y3.   Since I have students from the same class, I thought that class should be a random effect and I will need PROC Mixed for the analysis (Y1 Y2 and Y3 are all continuous).   I am not sure what to do with the teacher data. Teacher's age might affect the dependent variables, same is gender. Can I simply add these variables to the model as fixed effects, while class is random effect ? Obviously, same class means same teacher...   Is this code reasonable ?   proc mixed data=school; class Gender_S Gender_T; model Y1 = Gender_S Gender_T Age_S Age_T; random class; run;   And one more thing if I may. Is there a way to analyze all dependent variables together, like in MANOVA or MANCOVA, but only with mixed models ?   Thank you in advance ! ... View more

Change from baseline transformation

by Quartz | Level 8 in Statistical Procedures
‎04-06-2018 02:48 AM
‎04-06-2018 02:48 AM
Dear all,   I have a dataset containing data of patients with cancer. There are 60 subjects in this study, from which 30 got the standard treatment of chemotherapy, and 30 got chemotherapy along with nutritional supplements such as Turmeric and others. For each subject, the white blood count level was measured at 2 time points. Chemotherapy reduces the level of the white blood count, and the aim of the study is to show that in the treatment group, the decrease is more moderate, that the level of white blood count did not reduce as much.   I have a few questions regarding the analysis required (I am using SAS 9.4):   1. I wish to calculate the change from baseline. What is the right way to do it, absolute change from baseline, i.e. WBC2 - WBC1, or the percentage change from baseline, i.e. WBC2/WBC1 ?   2. If I use the absolute change from baseline, does it mean that a t-test or an equivalent is not a good choice, as I need to account for  the baseline measurement, leading me to a regression model ?   3. The data in both groups is not normally distributed (in one group the violation is minor, in the other it is more major). The sample size is not sufficiently large for the use of the central limit theorem (right?). A log transformation fails, since the absolute change from baseline gives negative values. How do I get out of this situation ? (I read somewhere about Fisher's transformation but as far as I know that is for correlations, and I am not sure I understand how to do it, theoretically and technically in SAS).    The mean of absolute change from baseline in the control group is -1.5, while in the treatment group -0.5. SD are 1.5 and 2 respectively.   Any advice will be most appreciated ! Thank you in advance ! ... View more

Hypothesis Testing for Odds Ratio

by Quartz | Level 8 in Statistical Procedures
‎06-11-2017 03:22 AM
‎06-11-2017 03:22 AM
Hello all,   I am trying to plan a study, in which I wish to compare two "rare proportions".   It is assumed that with the current treatment (the standard of care) , the proportion of events (bad events) is 7%. It is also assumed that with the new treatment this proportion goes down to 4%. The expected difference is 3%. An effect of 3% might not sound like much statistically, but clinicians claim it is very significant clinically.   I've tried calculating the sample size for the two proportion test (Fisher's exact test) and got that for power of 80% and significance level of 5%, the sample size is 971 per group, which is quite a lot. This sample size will give me the desired power to test the hypothesis that P1-P2=0.   Alternatively, I've calculated and found that under these assumptions, the OR is 1.8. The sample size to test the hypothesis that OR=1, under the same power and significance level, is 545 per group - far less !   Now I am confused and looking for the "downfall". My questions to you are:   1. Using SAS - how do I run an hypothesis testing for the OR = 1 vs OR >1 or OR NE 1 ? I would like to write a simulation so I need to know how to do it. Can I simply use Logistic Regression ?   2. What is the disadvantage of using the OR instead of the proportion difference ? There has to be a catch, I mean, why would anyone test for proportion difference if you can bypass it and test for the OR with a much smaller sample size ? There has to be a disadvantage I currently can't think about. Do you think that the FDA will accept using the OR instead of the difference or the common Fisher's exact test ?   3. Is there another approach to handle this situation of rare events, in which clinicians say 3% is amazingly important (each event is hard medically), statistically it's a small effect, and samples of ~1000 per group are way to expensive for a company to pay ? What would you recommend ? Maybe Bayesian Statistics ? Any advice will be mostly appreciated !   Thank you !     ... View more

Re: Power Simulation for a Two Proportions Test

by Quartz | Level 8 in SAS/IML Software and Matrix Computations
‎06-03-2017 05:28 AM
‎06-03-2017 05:28 AM
Rick, thank you. I think you solved the mystery, well, one of them anyway. I specified alpha=0.025 in order to get a 95% CI, but I forgot to test the P-Value vs 0.025. I changed it now and the results of both the CI and the hypothesis testing are identical. The only thing that is still a mystery is why proc power doesn't want to work, but I guess that with a good simulation in hand, I don't really need it. ... View more

Power Simulation for a Two Proportions Test

by Quartz | Level 8 in SAS/IML Software and Matrix Computations
‎06-02-2017 05:18 PM
‎06-02-2017 05:18 PM
Hello all,   I am trying to make a power analysis for a future clinical trial in which two proportions are to be compared. The analysis to be done is a non-inferiority analysis, i.e., it will be a one sided test for proportions difference, and the difference of proportions will be compared to a non-inferiority margin. The null hypothesis is P1-P2=-0.02, while the alternative hypothesis will be P1-P2>-0.02. The proportion in the first group (control) is assumed to be 8% while in the treatment group 6% (a smaller proportion is better).   I wrote this simulation, very much based on Rick Wicklin's example which I read in his book (that's why I publish this on this board, hoping that Rick will see this). The simulation ran and I got results, however, three problems occurred:   1. The results differ significantly from the results obtained from a sample size software, which used a formula rather than a simulation. The sample size in my simulation turned out to be much higher (about 300 subjects more overall).   2. I saved not only the P-Values of the non-inferiority test, but also the lower limit of the corresponding CI, which is a common practice which I saw in several articles. I expected the two methods to give the exact same results. But they don't. Not exactly.   3. I tried using PROC POWER to comapre the simulation results, but I keep getting an error saying: "NTOTAL is not available as a result option for TEST=." despite copying my code from the SAS documentation examples and changing the numbers only. Can you kindly assist me in figuring out what is wrong with my simulation ?   Thank you in advance   %macro ODSOff(); ods graphics off; ods exclude all; ods noresults; %mend; %macro ODSOn(); ods graphics on; ods exclude none; ods results; %mend; /* Macro Variables */ %let NumSamples = 10000; %let nimargin = 0.02; /* Non-inferiority margin */ %let negnimargin = -0.02; /* Negative non-inferiority margin */ %let P_SOC = 0.08; %let P_Treatment = 0.06; /* 1. Simulating N samples of EACH group (Total = 2N) for each sample */ data PowerSizeSim(drop = i); call streaminit(321); Pc = &P_SOC; Pt = &P_Treatment; do N = 200 to 500 by 25 ; /* N - sample size per group */ do SampleID = 1 to &NumSamples; do i = 1 to N; c = 1; x1 = rand("BERNOULLI", Pc); output; c = 2; x1 = rand("BERNOULLI", Pt); output; end; end; end; run; data SimulatedData; set PowerSizeSim; if x1 = 0 then x2 = 1; else if x1=1 then x2 = 0; drop Pc Pt; run; /* 2. Compute Statistics */ %ODSOff proc freq data = SimulatedData; by N SampleID; table c*x2 / riskdiff(noninf margin=&nimargin method=fm) alpha=0.025; ods output PdiffNoninf = NITESTS; run; %ODSOn /* 3. Construct indicator var for obs that reject H0 */ data ResultsSize; set NITESTS; RejectH0 = (PValue <= 0.05); LowerCI = (LowerCL >= &negnimargin); run; proc freq data=ResultsSize noprint; by N; tables RejectH0 / out = SimPower(where = (RejectH0 = 1)); tables LowerCI / out = SimPowerCI(where = (LowerCI = 1)); run; /* 4. Output */ title2 'Simulated Power by Sample Size'; proc report data = SimPower; column N PERCENT; define N / display 'Sample Size Per Group'; define PERCENT / display 'Simulated Power'; run; title2; /* PROC Power Validation */ proc power; twosamplefreq test = FM groupproportions = (0.08 0.06) nullproportiondiff = -0.02 alpha = 0.025 sides = U power = 0.8 /*ntotal = .*/ NPERGROUP=.; run;   ... View more

Multiple Regression - Comparing Adjusted Means

by Quartz | Level 8 in Statistical Procedures
‎05-05-2017 07:12 AM
‎05-05-2017 07:12 AM
Hello all, I have a dataset from the medical intudstry. An example of the data you can see in the attached table (image).     I have 3 dependent variables, all continous, some transformed due to non-normality, some not. In addition, I have two independent variables, which are the variables to be testes (the main research variables). Both came from a factor analysis (the first two factors).   The two first factor scores where categorized to quartiles (Q1, Q2, Q3 or Q4) using PROC RANK. In addition, I have several covariates, which I need for adjustment. Such covariates are age, BMI, smoking habbits, etc. The main target of the analysis is to look for differences in the means of all DV's by groups of both IV's, meaning, for each DV to see if there are differences between the different quartiles for each IV. I have several questions, which I hope you can assist me with, some of them are more technical, some methodological. I am using SAS 9.4. 1) I need a regression model that will yield the adjusted means (LSMEANS). I am not sure which SAS procedure to use here: REG, GLM or GLIMMIX. 2) In comparing the quartiles (for each IV), I am interested in checking if there is a trend (for example, if the adjusted mean of Q4> mean of Q3> mean of Q2> mean of Q1). Is there a test that does that in SAS?  In addition, I will be interested in simply comparing a couple of quartiles (mainly Q4 vs Q1). How should I get SAS to do that, to makes pair comparisons, while correcting the significance level? Should I use "contrast" or "estimate" ? 3) Methodologically, or theoretically, I need to adjust for some covariates, such as age and BMI. Since I have quite a few of them, putting too many variables in the model doesn't risk the effect of the IV's to be gone completely? Or on the other hand, doesn't it brings risk for overfitting?  Is there a risk of "over-adjusting", which will make effects to vanish? What is the best way to handle it ? Does SAS knows how to choose the best subset (including two way interactions) ? Thank you in advance for any tips you can give me. ... View more

Longitudinal Analysis with a Small Sample

by Quartz | Level 8 in Statistical Procedures
‎02-07-2017 04:04 AM
‎02-07-2017 04:04 AM
Hello all,   I have a samll data (N=7) with a longitudinal nature, and I am not sure how to analyze it. I am interested in a practical solution as well as an understanding of the correct one in case I had a larger sample.   My DV of interest is blood pressure. It is measured for each subject at baseline, and at several other time points post-treatment. There is only one group (one arm study).   My variables are: ID, Pressure (continuous), Time (categorical), Pressure at Baseline (I copied the values of the first time point into a column). The data is in long format, like SAS "likes it".   The differences between baseline and post treatment time points are relatively high (at the first time point the mean of difference is around 17, with a SD of 7). A simle Wilcoxon signed rank test is significant, despite the small sample.   Assuming that I had more data, what would be the correct way of analyzing this data ? My dilemma is which model is the best one. I wish to eventually compare the time points, each one to each one, but with a priority on comparing each time point to baseline.   The questions are:   1. If I put the pressure as a DV, and the Time as IV, it will include the baseline in it. Should I, or can I, enter the pressure at baseline as another covariate, or is it problematic? If I can't, how do I control for different baseline levels ?   2. Is it legal to enter the pressure change from baseline as a DV (after I calculate this variable) ? Now obviousaly I wish to test if the mean at any time point is not 0, how do I do that ?   3. What is the SAS code for models 1 and 2 ? I assume it will be either PROC MIXED or PROC GENMOD ( I am using SAS 9.4 but do not have the new PROC GEE yet) ?   4. In the case of a sample sample, how bad is it to simple to a paired t-test of Wilcoxson signed rank test for each time point vs. Baseline ?   Thank you in advance ! ... View more
Contact Me
Online Status
Offline
Date Last Visited
‎09-08-2018 12:32 PM
Likes given to
Likes from
Quick links