I am beginner user for SAS. Before I go to my questions, I really hope the person who answer can able to give me a clear guide to the problems as some books does not explain so clear.
1. do anyone have the experience in running SAS PROC MIXED, SAS PROC CALIS, SAS PRODCLIN, proc factor for multilevel data?
2. do anyone have a copy of macro language to analyze a multiple mediators in multilevel modeling?
3. may I know what is the proper steps in re-coding reversed variables Note; I do have several items that need to re-code.
Thanks in advanced.
At a certain level, your questions look like you are trying to do difficult things before you have learned how to do the simpler things. You may need to first do the exercises in something like "The little SAS book" before starting on the main problem.
You should look at Littel, et al's book on MIXED and Hatcher's on SEM. Both are readable, but assume you already know the basics of manipulating SAS.
To your third question, if the scales are numeric valued and ordinal or rational scaling, a simple way to reverse the scoring is to use
newscore = maxval - oldscore + minval;
where
oldscore is the existing scoring,
maxval is the maximum value for the scale, and
minval is the minimum value for the scale.
For instance, if the scale goes from 1-5 and you want to reverse it (so 4 becomes 2, etc.), then
new = 5 - old + 1;
Doc Muhlbaier
Duke
Thanks Duke,
first I apologized for the very late reply as I was busy reading up those books that you recommended me as you noted in your last reply. Yes, I grape some important points out of it.
Together with this reply, I have attached my project framework for your better understanding of my research. as we know in the process of data collection, we couldn't run away from missing data. Below is some of my research data and some proc which I am going to use to analyze my data.
****************************************************************************************************************************************************************************************************************************************************************************************
* LABEL
L5, L9, L11, L13 ='IDEALIZED ATTRIBUTES'
L2, L7, L12, L19 ='IDEALIZED BEHAVIOURS'
L4, L6, L14, L20 ='INSPIRATIONAL MOTIVATION'
L1, L3, L16, L18 ='INTERLECTUAL SIMULATION'
L8, L10, L15, L17 ='INDIVIDUALIZED CONSIDERATIONS'
OP24, OP25, OP33, OP34 ='RATUINAL GOAL'
OP27, OP31, OP32, OP36 ='OPEN SYSTEM'
OP21, OP23, OP26, OP29 ='INTERNAL PROCESS'
OP22, OP28, OP30, OP35 ='HUMAN RESOURCE'
JS37-JS56 ='JS'
AC57-AC62 ='AC'
OCB65, OCB80, OCB83, OCB84, OCB86 ='CONSCIENTIOUSNUESS'
OCB64, OCB66, OCB69, OCB78, OCB81 ='SPORTMENSHIP'
OCB68, OCB71, OCB73, OCB74 ='CIVIC VIRTUE'
OCB67, OCB70 OCB76, OCB79, OCB82 ='CERTESY'
OCB63, OCB72, OCB75, OCB77, OCB85 ='ATTRUISM';
DATA MDATA1;
INPUT #1 @1 (COID) (3.)
@5 (LEADER) (1.)
@7 (NONLEADER) (1.)
@9 (SEX) (1.)
@11 (AGE) (2.)
@14 (RACE) (1.)
@16 (MARITAL) (1.)
@18 (TENURE) (3.)
@22 (TENUREJOB) (3.)
@26 (MONTHLYS) (5.)
@32 (EDULVL) (1.)
@34 (POSITION) ($20.)
@55 (L1-L20) (1.)
#2 @1 (OP21-OP36) (1.)
#3 @1 (JS37-JS56) (1.)
@22 (AC57-AC62) (1.)
@29 (OCB63-OCB86) (1.);
CARDS;
614 1 0 2 37 1 1 36 003 4 Advertising Sale 44534454444433544444
5656565555454556
614 0 1 2 37 1 1 36 003 4 Advertising Sale 44535454345545444434
55545444543435354443 434435 434544544545454543543 43
614 0 1 1 2 1 4 Advertising 42442442344533443444
33433344434332443355 323234 415253142414443244241434
614 0 1 2 49 2 3 341 257 11000 2 Advertising 32343443443343232433
43443344535324443445 342225 315154155544444144154544
614 0 1 2 55 1 3 372 060 10000 4 Advertising 43443543423334333444
45434444444453444435 544124 324234132414414235232544
614 0 1 2 35 1 3 014 014 9000 4 Advertising 45444543444454444445
44544445544444444545 342214 425145143424554245154544
614 0 1 2 49 1 3 288 064 4 Advertising 43343443433444344444
44443444444334444334 442234 324224244444454345233444
614 0 1 1 54 3 3 432 012 3 Advertising 44443544444444343444
43334433343333333333 552225 425344344434443155223433
614 0 1 1 47 1 3 292 081 4 Advertising 23232232432422321233
44432343334333444334 332224 42524414 424444135245444
272 1 0 2 36 2 3 60 60 5 Audit 53541444433342443333
6534555556546564
272 0 1 1 31 2 1 25 25 4150 4 Accounts 44445543344444444443
44444444344444444344 431114 424244144344444444134444
272 0 1 2 48 2 3 272 53 7150 4 Estate 44445544444444444444
44444444444444444444 441114 415244144445354334144444
272 0 1 1 46 2 3 252 120 6180 4 Accounts Secretery 53543433534443444454
53444454444444444335 443224 43523414443443424534 133
793 1 0
793 0 1 2 27 2 1 7 7 2000 4 Corporate Planning 34443444434444444444
24444444444223444554 442224 434354345454313253245553
793 0 1 1 2 3 61 61 6000 4 Complience & Ledal 54444444444455544555
55555545555555555555 531115 414354141423444243343444
793 0 1 1 33 3 1 17 17 4 Finance 11123534114323222223
23222333332332333442 223333 134243453333313341233313
793 0 1 1 43 2 3 51 3500 3 33444444323333333333
33233333433332343333 342224 334143151344444333243433
793 0 1 1 34 2 3 108 72 4 Finance 53433444444544444444
34434443344443344444 442223 414144143434444243243333
(until it ended with (;) on the following line
DATA MDATA;
SET MDATA1 MDATA2;
PROC PRINT Data=mdata (obs=10);
RUN;
*Creating New Data Set for Analysis;
DATA MDATA_NEW;
SET MDATA;
*Recoding reversed Variables;
AC59 = 6 - AC59;
AC60 = 6 - AC60;
AC61 = 6 - AC61;
AC = (AC57 + AC58 + AC59 + AC60 + AC61 + AC62) / 6;
OCB64 = 6 - OCB64;
OCB66 = 6 - OCB66;
OCB69 = 6 - OCB69;
OCB78 = 6 - OCB78;
OCB81 = 6 - OCB81;
S = (OCB64 + OCB66 + OCB69 + OCB78 + OCB81) / 5;
*DATA MANIPULATION;
IA = (L5 + L9 + L11 + L13) / 4;
IB = (L2 + L7 + L12 + L19) / 4;
IM = (L4 + L6 + L14 + L20) / 4;
IS = (L1 + L3 + L16 + L18) / 4;
IC = (L8 + L10 + L15 + L17) / 4;
TFL = (IA + IB + IM + IS + IC) / 5;
RG = (OP24 + OP25 + OP33 + OP34) / 4;
OS = (OP27 + OP31 + OP32 + OP36) / 4;
IP = (OP21 + OP23 + OP26 + OP29) / 4;
HR = (OP22 + OP28 + OP30 + OP35) / 4;
OP = (RG + OS + IP + HR) / 4;
JS = (JS37 + JS38 + JS39 + JS40 + JS41 + JS42 + JS43 + JS44 + JS45 + JS46 +
JS47 + JS48 + JS49 + JS50 + JS51 + JS52 + JS53 + JS54 + JS55 + JS56) / 20;
CONS = (OCB65 + OCB80 + OCB83 + OCB84 + OCB86) / 5;
CV = (OCB68 + OCB71 + OCB73 + OCB74) / 4;
C = (OCB67 + OCB70 + OCB76 + OCB79 + OCB82) / 5;
ATT = (OCB63 + OCB72 + OCB75 + OCB77 + OCB85) / 5;
OCB = (CONS + S + CV + C + ATT) / 5;
*Create Value Label for the Demographic Section;
PROC FORMAT;
VALUE Sex 1 = "Female" 2 = "Male";
VALUE Race 1 = "Chinese" 2 = "Malay" 3 = "Indian" 4 = "Others";
VALUE Marital 1 = "Single" 2 = "Married" 3 = "Others";
VALUE Edulvl 1 = "Primary" 2 = "Secondary" 3 = "Diploma" 4 = "Degree" 5 = "Master's" 6 = "Doctorate";
RUN;
OPTION NOCENTER NODATE NONUMBER NOLABEL;
DATA mdata_new2a;
SET mdata_new;
IF LEADER=1;
DATA mdata_new2b;
SET mdata_new;
IF NONLEADER=1;
/*****Descriptive Statistics for Level-2 Variables*****/
PROC MEAN DATA=mdata_new2a MEAN STD;
VAR TFL OP;
RUN;
/*****Descriptive Statistics for Level-1 Variables*****/
PROC MEAN DATA=mdata_new2b NEAB STD;
VAR TFL JS AC OCB;
RUN;
/*****MODEL 1*****/
TITLE "Unconditional Mean Model";
PROC MIXED DATA=mdata_new METHOD=ML COVTEST Noclprint;
CLASS coid;
MODEL OP = / Solution;
RANDOM intercept / TYPE=UN Subject=coid;
RUN;
/*****MODEL 2*****/
Title "MODEL 2A: Including Effects of Transformational Leadership (LEVEL-2) Predictors";
PROC MIXED DATA=mdata_new2a METHOD=ML COVTEST Noclprint;
CLASS coid;
MODEL op = tfl / Solution DDFM = bw;
RANDOM Intercept / TYPE=UN SUBJECT=coid;
REPEATED / TYPE=VC SUBJECT=coid;
RUN;
/*comparing the degrees of freedom with and w/o option ddfm = bw*/
PROC MIXED DATA=mdata_new2a METHOD=ML COVTEST Noclprint;
CLASS coid;
MODEL op = tfl / solution ;
RANDOM intercept / TYPE = UN subject = coid;
REPEATED / TYPE=VC SUBJECT=coid;
run;
Title "MODEL 2b: Including Effects of Transformational Leadership (LEVEL-1) Predictors";
PROC MIXED DATA=mdata_new2b METHOD=ML COVTEST Noclprint noitprint;
CLASS coid;
MODEL op = tfl / Solution DDFM=bw Notest;
RANDOM Intercept tfl / SUBJECT=coid TYPE=UN;
REPEATED / TYPE=UN SUBJECT=coid;
RUN;
/*****MODEL 3*****/
Title "Model 3: Including Both Level-1 and Level-2 Predictors";
Title2 "--Predicting op from leader, momleader, tfl and";
Title3 "cross level interaction of nonleader and tfl";
PROC MIXED DATA=mdata_new noclprint covtest noitprint;
CLASS coid;
MODEL op = leader nonleader tfl leader*tfl nonleader*tfl / Solution DDFM=bw Notest;
RANDOM Intercept tfl / Subject=coid TYPE=UN;
RUN;
**********************************************************************************************************************************************************************************************************
if you notice, I had made it bold in some of the parts in the proc step and one of the datalines. The one which I bold in the dataline is a missing case. Meanwhile the proc step which I bold produce an error while I submit for process. would you mind help me to check out what It had been wrong and how can I include the level-1 data to be in the equation when I make a relationship with the outcome variable which is in Level-2
Another problem which I am having right now is how to conduct a diagnostics test on multilevel data with SAS MACRO_DX? do you know how to fit in the data into the macro to run the diagnostics test for multilevel modeling data?
Thank you very much.
HI,
I haven't used MIXED in quite a while, so I can't help you on the specifics. If you post the section of your log with the error message, maybe someone else can help.
Doc Muhlbaier
Duke
Build your skills. Make connections. Enjoy creative freedom. Maybe change the world. Registration is now open through August 30th. Visit the SAS Hackathon homepage.
Register today!Learn the difference between classical and Bayesian statistical approaches and see a few PROC examples to perform Bayesian analysis in this video.
Find more tutorials on the SAS Users YouTube channel.
Select SAS Training centers are offering in-person courses. View upcoming courses for: