************************************************************
* PROGRAM:      SCM3.SAS                                   *
* PROGRAMMER:   Cindy Rains                                *
* DATE:         3/05/03                                    *
*                                                          *
* PURPOSE:      CxE datasets contain sociometric form data.*
*               This program creates a subscale for the    * 
*		    self-report items as well as one dataset   *
*		    with all of the data for public use.       *                                         *
* INPUT:        CxE, s=site, y=year, c=cohort              *
*                                                          *
* OUTPUT:       SCM3SC1                                    *
************************************************************;

** Fill out information below for data processed in this run **;

** Name library holding SAS input data files **;

  LIBNAME SASIN  "l:/datasets/child/sociometrics/u" ;


** Specify cohort number **;

        %let coh = 1 ;

** Enter year of data collection **;

        %let yr = 3 ;  /*  Grade 2  */

*************** DO NOT MODIFY BELOW THIS LINE ******************************;
** Specify library for output datasets **;

  LIBNAME SASOUT "d:\work\sociometrics\year 3" ;

** Allkey datasets **;

  LIBNAME ALL    "d:\allkey"       ;

options ps=80 ls=132 nocenter nostimer ;

** The macro, "linsum" creates derived variables which are means of input  **
** scale items.  In the case of missing input items, the output variable   **
** is defined to be missing if half or more of the input items are         **
** missing.  Otherwise, missing input items are replaced with the mean of  **
** the nonmissing items.  The macro call to linsum specifies the input     **
** items as a string "var_list" of variable names, and specifies the       **
** label and name of the output variable as "var_name" and "var_labl"      **
** respectively.                                                          **;

%macro linsum ;
data first ;
  set &INDATA ;
   count = 0                     ; 
   label &var_name = "&var_labl" ;
   meanz = 0                     ;
  array sumvars &var_list ;
  array working &var_list count meanz &var_name ;
    do i = 1 to dim(sumvars) ;
      if working(i) = . then count = count + 1 ;
    end ;
    do i = 1 to 1 ;
      meanz = mean(of &var_list) ;
    end ;
    do i = 1 to dim(sumvars) ;
      if working(i) = . then working(i) = meanz ;
    end ;
    do i = 1 to 1 ;
      if count le .5*dim(sumvars) then
        &var_name = sum(of &var_list) ;
        else &var_name = . ;
    end ;

  keep site tcid &var_name ;
  proc sort ; by site tcid ;

data &INDATA ;
  merge first &INDATA ;
  by site tcid ;
%mend linsum;
****************************************************************************;

DATA SCM1 ;
  SET SASIN.C&yr.E ;  
  **Need to convert the responses for SR1-8 from yes/no to numbers (1,0)**;
if C&yr.ESR1 in ('Y', 'N') then C&yr.ESR1x = (C&yr.ESR1 = 'Y');
if C&yr.ESR2 in ('Y', 'N') then C&yr.ESR2x = (C&yr.ESR2 = 'Y');
if C&yr.ESR3 in ('Y', 'N') then C&yr.ESR3x = (C&yr.ESR3 = 'Y');
if C&yr.ESR4 in ('Y', 'N') then C&yr.ESR4x = (C&yr.ESR4 = 'Y');
if C&yr.ESR5 in ('Y', 'N') then C&yr.ESR5x = (C&yr.ESR5 = 'Y');
if C&yr.ESR6 in ('Y', 'N') then C&yr.ESR6x = (C&yr.ESR6 = 'Y');
if C&yr.ESR7 in ('Y', 'N') then C&yr.ESR7x = (C&yr.ESR7 = 'Y');
if C&yr.ESR8 in ('Y', 'N') then C&yr.ESR8x = (C&yr.ESR8 = 'Y');

drop C&yr.ESR1-C&yr.ESR8;
RUN;
DATA NEW;
SET SCM1;
RENAME C&yr.ESR1x = C&yr.ESR1
	   C&yr.ESR2x = C&yr.ESR2
	   C&yr.ESR3x = C&yr.ESR3
	   C&yr.ESR4x = C&yr.ESR4
	   C&yr.ESR5x = C&yr.ESR5
	   C&yr.ESR6x = C&yr.ESR6
	   C&yr.ESR7x = C&yr.ESR7
	   C&yr.ESR8x = C&yr.ESR8;
PROC SORT ; BY SITE TCID ;
**Reverse Score items before creating subscale**;
DATA NEW2;
SET NEW;
C&yr.ESR3r=1-C&yr.ESR3;
C&yr.ESR6r=1-C&yr.ESR6;
C&yr.ESR7r=1-C&yr.ESR7;
PROC SORT; BY SITE TCID;
**Merge with allkey file**;

DATA SCM2 ;
  MERGE NEW2 (IN=SCM) ALL.ALLKEY&coh. (IN=ALL) ; BY SITE TCID ;
  IF SCM AND ALL ; 

/* Specify definitions for SCALE VARIABLES below:         ** */
        %let INDATA = SCM2 ;
************************************************************;
        %let var_name = SCM&yr.SCS ;
        %let var_labl = %str(School Satisfaction - Sociometrics ) ;
        %let var_list = %str(C&yr.ESR1 C&yr.ESR2 C&yr.ESR3 C&yr.ESR4 C&yr.ESR5 C&yr.ESR6 
C&yr.ESR7 C&yr.ESR8) ;
        %linsum ;
************************************************************;
        

DATA SASOUT.SCM&yr.SC&coh. (label="Scored Sociometrics Y&yr. C&coh.") ;
  SET SCM2;
**This keeps all of the data from the original dataset, CxE, and creates a scored dataset which includes the one subscale**;

  Keep site tcid cohort sitecode year C&yr.ESCHID C&yr.ETCHID C&yr.EROSTR SCM&yr.SCS  
C&yr.ESR1-C&yr.ESR2 C&yr.ESR3r C&yr.ESR4 C&yr.ESR5 C&yr.ESR6r C&yr.ESR7r C&yr.ESR8 C&yr.EZLMS3 C&yr.EZLMSu C&yr.EZLLS3 C&yr.EZLLSU C&yr.EZAGGS C&yr.EZWDRS C&yr.EZPROS C&yr.EZHYPS C&yr.EZVICS C&yr.EZPRF3 C&yr.EZSIM3 C&yr.EZPRFU C&yr.EZSIMU C&yr.EGLLSU C&yr.EGLLS3 C&yr.EGLMSU C&yr.EGLMS3 C&yr.EGAGGS C&yr.EGHYPS C&yr.EGPROS C&yr.EGVICS C&yr.EGWDRS 
C&yr.EGPRFU C&yr.EGSIMU C&yr.EGPRF3 C&yr.EGSIM3 C&yr.ESTAT3 C&yr.ESTATU C&yr.EGSTA3 C&yr.EGSTAT C&yr.ECLASS C&yr.EVOTEE C&yr.ELMS3 C&yr.ELMSU C&yr.ELLS3 C&yr.ELLSU C&yr.EAGGS C&yr.EWDRS C&yr.EWDRS C&yr.EPROS C&yr.EHYPS C&yr.EVICS C&yr.ERR1-C&yr.ERR32 C&yr.EDK1-C&yr.EDK32 C&yr.ESEX
C&yr.EPERM C&yr.ERMAX C&yr.EPRATE C&yr.ERRGIV C&yr.ERRREC C&yr.ERRSTD C&yr.EG TREATMNT NORM SEX RACE C&yr.EG C&yr.ESR9-C&yr.ESR21 TCID5;

LABEL C&yr.ESR1="Self-Report 1: School fun?"
	  C&yr.ESR2="Self-Report 2: Nice things happen?"
	  C&yr.ESR3r="Self-Report 3: Feel unhappy?"
	  C&yr.ESR4="Self-Report 4: Like your school?"
	  C&yr.ESR5="Self-Report 5: Kids to play with?"
	  C&yr.ESR6r="Self-Report 6: Lonely at school?"
	  C&yr.ESR7r="Self-Report 7: Hard to make friends?"
	  C&yr.ESR8="Self-Report 8: Kids like you?";
	  
run;

proc contents;
run ;

proc freq;
table SCM&yr.SCS C&yr.ESR1-C&yr.ESR2 C&yr.ESR3r C&yr.ESR4 C&yr.ESR5 C&yr.ESR6r C&yr.ESR7r 
C&yr.ESR8;
run;