BACKGROUND INFORMATION ON 5-YEAR CLIENT SURVEY OUTCOME MEASURES

IMPORTANT: Values of some measures have been changed to protect sample
members' confidentiality. (See notes in codebook) The data file
N5RS1621.TXT contains the original values and is available to
researchers (on a restricted basis) at the National Center for Health
Statistics.  See www.aspe.hhs.gov/hsp/newws/data-info.htm for more
information.

I. PARTICIPATION IN EMPLOYMENT-PREPARATION ACTIVITIES (SECTION D)


The Five-Year Client Survey Participation section was designed to supplement
the more detailed questions on participation in Two-Year Client Survey.  It
focuses on program and control group members' participation in employment
related activities during the year before the 5-interview.

Respondents were asked about participation in seven activities.  These
activities are distinguished by a 2-letter stem in the variable names:

..JC....   Job club/job search
..ES....   ESL: English as a Second Language
..AB....   Adult Basic Education, GED Preparation, Regular High School
..CO....   Post-Secondary Education, College, Community College
..VO....   Vocational Training
..UJ....   Work Experience, Unpaid Job
..OJ....   On-the-Job Training (OJT)

Respondents could also describe up to three activities as "Other":

..OTH...

Respondents who indicate that they participated in "Other" activites were then
asked to describe them, either with a code, corresponding to one of the seven
activity types listed above, or (if the activity did not seem to fit any
category) in a verbatim response:  D5OTH1-D5OTH3.

D5OTHn variables were set to 94 when the respondent provided a verbatim
response.


For each activity, the interview includes a sequence of 3 questions.  Answers
are designated by the same suffix:


1) Since [random assignment date], have you ever attended [activity type]?:
....[no suffix]

2) Did you participate in [activity type] in the last year?: ....LY

3) Since [date: 1 year before interview] for how many months did you attend
[activity type]?:   ....MO

   NOTE:  Attendance in job club was measured in weeks: ....WK


Skip Patterns

For each activity, the interviewer skips to the next activity if the respondent
answers NO to the 1st or 2nd question.   Skipped questions are set to MISSING.


Adjustments to Values

1) The number of months/weeks of participation measure included a code to
designate a fraction of one month (week).  These codes were reset to 0.5.

2) Recoding of "Other Activity" verbatim responses

MDRC reviewed verbatim responses and recoded D5OTH1-D5OTH3 to a specific
activity code when responses fit the description of one of the 7 activity types
listed above.  Otherwise the value was left at 94.


ADDITIONAL MEASURES


1) 0/1 "Ever participated" since random assignment date

a)  Based on responses to the 5-Year Survey only:

V5JC
V5ES
V5AB
V5CO
V5UJ
V5OJ
V5VO
V5OT

These measures are initialized to 0.  Respondents get a value of 1 if either
the original activity measure=1 (YES) or the code for ..OTH1-..OTH3 indicates
that activity type.  For instance:

V5JC=1 if D5JC=1  or [any of] D5OTH1-D5OTH3=1  [code for job club/job search]

The file also includes summary measures for related activities:

V5BE: Any basic education:=1, if V5ES=1 or V5AB=1

V5VC: Any vocational training, college, or post-secondary education:
      =1, if V5VO=1 or V5CO=1

V5WE: Any work experience or OJT: =1, if V5UJ=1 or V5OJ=1

V5ET: Any education or training: =1 if V5BE=1 or V5VC=1

V5PT: Any activity: =1 if V5JC=1 or V5ET=1 or V5WE=1


b) Based on responses to the 2-Year and 5-Year Client Survey


These measures are initialized to 0.  Respondents get a value of 1 if either
the V5.. measure=1 or the corresponding measure from the 2-Year Client
Survey=1:

VJC5Y
VES5Y
VAB5Y
VCO5Y
VVO5Y
VUJ5Y
VOJ5Y

The file also includes summary measures for related activities:

VBE5Y:  Any basic education: =1 if VES5Y=1 or  VAB5Y=1

VVC5Y:  Any vocational training, college, or post-secondary education
        =1 if VCO5Y=1 or  VVO5Y=1

VWE5Y:  Any work experience or OJT: =1 if VUJ5Y=1 or VOJ5Y=1

VET5Y:  Any education or training: =1 if VBE5Y=1 or VVC5Y=1

VPT5Y:  Any activity: =1 if VJC5Y=1 or VET5Y=1 or VWE5Y=1

2) 0/1 "ever participated in the year before interview"

VJCY5
VESY5
VABY5
VCOY5
VUJY5
VOJY5
VVOY5
VOTY5

These measures are initialized zero and equal 1 if:

a) The corresponding ....LY variable= 1 (YES)

   OR

b) Any of D5OTHLYn variables=1 and the corresponding D5OTHn has the code
indicating that activity type


3) Total months of participation in the year before interview

 KJCM5
 KESM5
 KABM5
 KCOM5
 KUJM5
 KOJM5
 KVOM5

These measures=
D5..MO
or
OR D5..MO + D5OTHMOn if any of D5OTHLYn variables=1 and the corresponding
D5OTHn has the code indicating that activity type.

ATTITUDES TOWARD WORK AND WELFARE

This section consists of 8 statements about work and welfare to which the
respondent is supposed to state the extent to which she agreed or disagreed.
Values range from 1 (agree a lot) to 5 (disagree a lot).

Respondents are then asked two questions related to their "reservation wage.":

M5LOWAGE: If respondent were unemployed, what is the lowest wage per hour she
          would accept for a full-time job with full medical benefits

M5FAIRWG: If respondent were unemployed, what would be a fair wage per hour
          for a full-time job with full benefits, given her skills and
          experience


NOTE:  These questions were asked of all respondents, irrespective of whether
they were working at the time of their 5-year interview. The responses may be
used together with responses from other sections and surveys, including

1) Questions on costs of child care and transportation for employment; why
respondents left their most recent job; and why did not accept a job offer
(Employment: Section F)

2) [For Child Outcomes Study Survey respondents:] Questions on domestic violence
and barriers to work (Working: Section GG)

3) Questions on attitudes toward work and welfare from the 2-Year Client Survey

4) Questions on attitudes toward work and welfare from the Private Opinion
Survey administered on the day of random assignment


II. DEGREE RECEIPT (SECTION E)

This section repeats the questions on degree receipt that were asked during the
2-Year Client Survey interview.  The questions were included to record degree
attainment that occurred during years 3 to 5 of follow-up, but respondents
could also report degree attainment during years 1 and 2, or during any
time before random assignment date.  Thus, this section may provide new
information on degree attainment, the same information as was recorded in
the 2-Year Survey, or information that differs from what was recorded
in the 2-Year Survey.


The interview recorded two types of measures:

 (1) Yes/no (1/2) variables for each degree which indicate whether a respondent
had ever received that degree


E5TR    Trade license
E5GE    GED certificate
E5HS    High school diploma
E5AS    Associate's degree
E5BA    Bachelor's degree
E5GR    Graduate degree

Also:

E5CRED  Since (random assignment date), earned credits towards college degree


 (2) Date variables (MM/YY) which indicate the date the degree was received. A
DD was added to facilitate calculation of SAS dates. The E5..SD variables are
stored as YYMMDD:

E5TRSD   Trade license
E5GESD   GED certificate
E5HSSD   High school diploma
E5ASSD   Associate's degree
E5BASD   Bachelor's degree
E5GRSD   Graduate degree


Note: interviewers recorded dates of degree attainment even if they
occurred many years before random assignment.

Interviewers implemented a skip pattern in this section:

If a respondent indicated that she did not have a high school diploma or GED,
then she was not asked whether they she had a higher degree, such as a
bachelor's or associate's degree. These measures are set to MISSING.


Changes to Degree Receipt Dates

If the respondent reported the year of degree attainment, but not the month,
the MM and DD of E5..SD were set to 0702 (the middle of the year).  However, the
date was left as missing if no YY nor MM were reported.


Additional Outcome Measures

1) Relative month of degree receipt (Random assignment month=1):


E5TRRLM  Trade license
E5GERLM  GED certificate
E5HSRLM  High school diploma
E5ASRLM  Associate's degree
E5BARLM  Bachelor's degree
E5GRRLM  Graduate degree

NOTE: Negative values indicate receipt prior to random assignment.


2) 0/1: Ever received degree after random assignment, based on 5-Year Survey
responses.  These measures are initialized to 0 and =1 if respondent reported a
degree receipt date on or after random assignment date.

VE5TR5Y  Trade license
VE5GE5Y  GED certificate
VE5HS5Y  High school diploma
VE5AS5Y  Associate's degree
VE5BA5Y  Bachelor's degree
VE5GR5Y  Graduate degree


3) 0/1: Ever received degree after random assignment, based on 2-Year and 5-Year
Survey responses.  These measures are initialized to 0 and =1 if respondent
reported (during either interview)  a degree receipt date on or after random
assignment date.

VTR5Y   Trade license
VGE5Y   GED certificate
VHS5Y   High school diploma
VAS5Y   Associate's degree
VBA5Y   Bachelor's degree
VGR5Y   Graduate degree

Additional 0/1 summary measures:

VDP5Y   GED or high school diploma
VTD5Y   Associate's degree or trade license
VDG5Y   Any degree


III. EMPLOYMENT AND EARNINGS (SECTION F)

The 5-Year Client Survey contains detailed information on respondents' current
or most recent job.  The survey also recorded information on instances after
random assignment in which respondents refused a job offer--including
respondents who reported that they had not worked since random assignment.

Respondents are first asked if they ever worked for pay since random assignment:
(F5JBANY)

  If YES: The remaining questions are asked

  If NO: If no, respondents are asked if they ever worked at irregular or odd
         jobs to make ends meet.

         If YES: The remaining questions are asked, including questions
         about instances in which they refused a job offer.

         If NO: Questions on employment are skipped.  Respondents are asked
         about instances in which they refused a job offer.


The following information was collected on respondents' current of most
recent job:

Job history:

    JB5SD: Job start date (YYMMDD)
    JB5ED: Job end date (YYMMDD)
    F5JBCURC: Still working at job at interview?
    F5JBNUM : Number of current jobs, if still working at interview
    F5JBNOJB: Number of jobs in year prior to interview

Job characteristics:

    F5JBDSCR: Type of work (U.S. Census occupational codes)
    F5JBHRS : Number of hours per week of work
    F5JBSHFT: Job hours regular or irregular?
    F5JBBEGH: Time job begins, hour
    F5JBBEGN: Time job begins, minute
    F5JBSBAP: Time job begins, AM/PM
    F5JBENDH: Time job ends, hour
    F5JBENDN: Time job ends, minute
    F5JBSEAP: Time job ends, AM/PM
    F5JBGT24: Type of 24 hr shift
    F5JBSVAR: Work hours fixed or rotate?
    F5JBSAL : Gross earnings from job
    F5JBBART: Paid in barter?
    F5JBTAX : Are reported earnings- pre or after-tax?
    F5JBFREQ: Pay period of wage payments
    F5JBDAYS: Number of days per week of work, if paid by day
    F5JBINKD: Paid in barter, goods, or reduced expenses?
    F5JBIKVL: Value of barter, goods, reduced expenses?
    F5JBIKPR: Pay period of barter payments

Fringe benefits:

    F5JBSICK: Job provides paid sick days?
    F5JBVAC : Job provides paid vacation days?
    F5JBHEAL: Employer offered a health plan/medical insurance?
    F5JBACCP: Respondent accepted employer's health plan/medical insurance?


Work-related expenses:

Transportation:

    F5JBTRAV: Amount spent on transportation to job
    F5JBDRIV: Usually drive to work?
    F5JBWALK: Usually walk to work?
    F5JBCRPL: Usually get ride to work?
    F5JBPBTR: Usually  take publc trnsp to work?
    F5JBOTH : Other transportation taken to work?
    F5JBWKHM: Work at home?
    F5JBTOWK: Number of minutes to commute to job


Child Care for employment:
NOTE: The following measures are stored with other child care measures on
Record #


    F5JBRESP: Working and respondent cares for a child under 13 years of age?
    F5JBCC  : Has child care while at work?
    F5JBCCPY: Cost per week of child care

If no longer employed at interview:

    F5JBLEFT: Reason leaving most recent job
    F5JBQUI1-F5JBQUI3: Reasons for leaving most recent job
    F5JBQUIM: Main reason for leaving most recent job

Refusing a job offer since random assignment date:
NOTE: The following questions were asked of all respondents:

    F5JBDECL: Ever refused job offer since random assignment?

If YES: For most recent occasion when a respondent did not accept a job offer:

    F5JBDEC1-F5JBDEC3: Reasons for refusing most recent job offer
    F5JBDECI: Main reason for refusing most recent job offer


CHANGES TO JOB START AND END DATES:

JB5SD: Job start date (YYMMDD)

Respondents were asked to provide the year and month when they started their
current or most recent job.  The DD was added by MDRC to facilitate creation of
date values in SAS or a different software package.  Usually, the DD was set to
15.


Start dates were changed in the following situations:

1) Set date to random assignment date, if start date before random assignment
and end date after random assignment.

2) Set DD to 01 of interview month if job started during interview month and
respondent was interviewed before the 15th of the month.  [Prevents end date
before start date]

3) Set start date to end date (and reset end date to missing) if
start date was recorded as occurring after end date.

4) Set start MM and DD to 07/02 (i.e., the middle of the year), if only the YY
of start date recorded

5) Impute the start date if missing and end date after random assignment date,
based on the average number of months from start to end -- calculated for
for all respondents in the same site and research group who had valid start and
end dates.

Start date= LATER of: End date - (average months) OR random assignment date


    JB5ED: Job end date (YYMMDD)


End dates were changed in the following situations:

1) Set to interview date, if respondent reported that she was currently working
for pay (F5JBCURC=1).

2) Set end MM and DD to 07/02 (i.e., the middle of the year), if only the
YY of end date recorded.

3) Set to interview date if start date before interview date and end date
recorded as later than interview date.

4) Impute the end date if missing and start date on or after random assignment
date, based on the average number of months from start to end -- calculated for
for all respondents in the same site and research group who had valid start and
end dates.

End date= LATER of: Start date + (average months) OR interview date




Additional outcome measures created from 5-Year Client Survey
employment start and end dates

JB5SRLM    Relative month of job start: random assignment month=1

JB5ERLM    Relative month of job end: random assignment month=1

F5SEMP     Ever employed after random assignment

SEMPY5     Ever employed in year 5

JB5STLIN   Employed at 5-year interview

JB5LOSS    Left most recent job before 5 year interview

F5JB       Employed during month before interview (used for research on income:
           See Section J below)


Also:

SEMP5Y     Ever employed after random assignment, based on responses to the
           2-Year and 5-Year Client Survey (=1 if respondent reported a
           start date of employment or or after random assignment on
           either survey)



NOTE: For about 50 respondents, the job reported on the 5-Year Client Survey may
not be their most recent job.  These jobs have end dates that occurred before
the end date of a job that the respondent reported on the 2-Year Survey.  MDRC
did not change the employment data on the 5-Year Survey for these respondents,
although some researchers may wish to.



CREATION OF HOURS OF WORK MEASURES


Hours of work per week were


1) Top-coded at 95

2) Imputed for respondents who reported an employment start or end date during
the follow-up period but did not report hours of work.  The following strategy
was used:

    a) Copy hours per week from respondent's most recent job from the Two-Year
       Survey, if available.  ELSE

    b) Substitute mean value for respondent's site and research group.


Additional measures include:

FT5EMP:   Employed for at least 30 hours per week at current or most recent job

FT535EMP: Employed for at least 35 hours per week at current or most recent job

PT5EMP:   Employed for less than 30 hours per week at current or most recent job

PT535EMP: Employed for less than 35 hours per week at current or most recent job

J5FTEMP:  Employed for at least 30 hours per week at current job (JB5STLIN=1)

J5FT35EP: Employed for at least 35 hours per week at current job (JB5STLIN=1)

J5PTEMP:  Employed for less than 30 hours per week at curren  job (JB5STLIN=1)

J5PT35EP: Employed for less than 35 hours per week at current job (JB5STLIN=1)



CREATION OF WEEKLY WAGE MEASURES

These measures were initialized to 0.


WKLYWAG5: Weekly pay, current or most recent job


Weekly wage=Sum of cash and barter wages (truncated)


Cash and barter wages were calculated based on the value of F5JBFREQ
(Pay period) and related questions:


F5JBFREQ       CALCULATION

1              F5JBSAL * F5JBHRS            /*HOURLY*/
2              F5JBSAL * F5JBDAYS           /*DAILY*/
3              F5JBSAL * 1                  /*WEEKLY*/
4              F5JBSAL/2                    /*EVERY 2 WEEKS*/
5              F5JBSAL*24)/52               /*TWICE A MONTH*/
6              (F5JBSAL*12)/52              /*MONTHLY*/
7              F5JBSAL/52                   /*YEARLY*/




Barter wages were calculated in the same way, using


F5JBIKVL: Value of barter, goods, reduced expenses

F5JBIKPR: Pay period of barter payments

and

F5JBHRS

F5JBDAYS


Changes to the value of weekly wages:

Weekly wages were top-coded at $1153  (equivalent of $15,000 per quarter)



IMPUTATION OF WEEKLY WAGES

Impute weekly pay according to the following hierarchy:


1) Substitute monthly earnings from Income Section (Section J)

2) Substitute the highest value of:

a. F5JBHRS  *
   (hourly wage of current or most recent job from 2-YEAR CLIENT SURVEY)

b. (Value of UI earnings for the quarter of interview) / 13

c. F5JBHRS  *  (Site and research group mean of hourly wage of current or most
   recent job from 5-YEAR CLIENT SURVEY)


Note: respondents with imputed values have a value of 1 on one or more Missing
Value indicators included on the file.  See below for details.


Additional Weekly Wage Measures

J5WKLYPY: Weekly pay, current job

J5ADJWKP: Adjusted weekly pay, current job:  equals J5WKLYPY / 0.85 if
respondent indicated that she was resporting after-tax earnings (F5JBTAX=2).



HOURLY PAY

These measures were

1) Initialized to 0.

2) Top-coded at $25 per hour.


HRLYWAG5:  Hourly wages, current or most recent job:  equals WKLYWAG5 / F5JBHRS,
if pay period other than "hourly".  Else=F5JBSAL

J5HRLYPY:  Hourly wage, current job

J5ADJHRP:  Adjusted hourly wage, current job: equals J5HRLYPY / 0.85 if
respondent indicated that she was resporting after-tax earnings (F5JBTAX=2).

NOTE: In most instances, hourly pay was imputed whenever weekly pay was imputed.
(The exception was when the respondent reported being paid by the hour and
reported the wage amount, but did not report her hours per week of work.)


ADDITIONAL MEASURES BASED ON REPORTED FRINGE BENEFITS:


All of the following measures were initialized to 0.  All are 0/1 measures:


For current or most recent job:

V5JBMDFR: Offered medical coverage
V5JBMDCC: Accepted medical coverage
V5JBSICK: Sick days with pay
V5JBVAC : Paid vaction
V5JBSAM : Morning start to job
V5JBSPM : Afternoon or evening start to job
V5JBFXED: Fixed hours
V5JBRTAT: Hours rotate
V5NJBMFR: Not offered medical coverage
V5NJBMCC: Didn't accept medical covrage, when offered
V5NJBSCK: No sick days with pay
V5NJBVAC: No paid vaction
V5FTOMC : Employed 30 hours or more per week and offered medical coverage
V5FTAMC : Employed 30 hours or more per week and accepted medical coverage


For current job (JB5STLIN=1):
J5JBMDFR: Offered medical coverage
J5JBMDCC: Accepted medical coverage
J5JBSICK: Sick days with pay
J5JBVAC : Paid vaction
J5JBSAM : Morning start to job
J5JBSPM : Afternoon or evening start to job
J5JBFXED: Fixed hours
J5JBRTAT: Hours rotate
J5NJBMFR: Not offered medical coverage
J5NJBMCC: Didn't accept medical covrage, when offered
J5NJBSCK: No sick days with pay
J5NJBVAC: No paid vaction



DATA PROBLEM FLAG VARIABLES

These variables identify respondents with imputed data.  Researchers who wish
to implement their own imputation strategy or who wish to work with unimputed
data may use these measures to select respondents with imputed responses and
reset values to missing:


JB5NMREC:  Most recent job on 5-year survey ended before most recent job on
           2-year survey.

JB5FIXMM Job start or end date missing MM,reset to 7

JB5FLG2  Job date flag:  value 3 (missing and imputed start date) or
         values 4,5,7 (missing and imputed end date)

HRS5FLG  Current or most recent job missing hours
MISSAL5  Current or most recent job missing salary amount
MISSBSL5 Current or most recent job missing barter amount
JBBRCAL5 Current or most recent job missing barter amount per week
JBHRCAL5 Current or most recent job missing hours per week
JBWKCAL5 Current or most recent job missing weekly pay
IMPTWAG5 Current or most recent job: weekly wage imputed


IV. CHILD CARE

Section F (Employment) of the survey asks questions about respondents' child
care use while employed; section G (Transitional Benefits) asks about receipt of
transitional child care benefits.

CHILD CARE WHILE EMPLOYED

1. Source Variables
F5JBRESP captures responses to the question "Do/Did you have any children you
are/were responsible for while you are/at the time you were working at your
current/most recent job? Only count children who are/were younger than age 13
and living with you now/at the time." For this variable, 1=yes and 2=no.

F5JBCC captures responses to the question "Do you have/Did you have any child
care arrangements for those children while you are/were working?" For this
variable, 1=yes and 2=no.

F5JBCCPY captures responses to the question "Altogether, how much do/did you
usually pay per week for child care out of your own pocket when you are/were
working, counting all of the arrangements you have/had for all the children
younger than age 13? Don't include any expenses that are/were paid or reimbursed
by someone else." For this variable, respondents could give an unrestricted
dollar amount.

2. Skip Patterns
There were three questions about child care while employed. The skip patterns
for these questions are as follows:

a) Only respondents who reported working (F5JBANY=1 or F5JBIRR=1) were expected
to answer the questions about child care while employed.

b) Respondents who answered that they did not have any children who were younger
than age 13 and living with them (F5JBRESP=2) skipped the remaining questions
about child care while employed.

c) Respondents who answered that they did not have any child care arrangements
during their most recent or current job (F5JCC=2) skipped question did not
answer the question about the cost of child care.

3. Created Variables
F5CHCARR is a dummy variable that equals 1 if respondents reported using child
care during their most recent or current job (F5JCC=1).

F5CCMCST is a continuous variable that equals the monthly cost of child care
used during current or most recent job. The weekly dollar amount reported in
F5JBCCPY was multiplied by 4.3 to get a monthly cost.

F5PYCCST is a dummy variable that equals 1 if respondents reported paying out of
pocket for child care during their current or most recent job. Respondents who
gave a dollar amount of at least 1 for F5JBCCPY are coded as paying out of
pocket for child care.

CCAREMIN is a dummy variable that equals 1 if respondent reported using child
care during current or most recent job and was currently employed at the time of
the 5-year interview.

F5EMCCST is a continuous variable that equals the cost of child care used during
current or most recent job, for those who were currently employed at the time of
the 5 year interview. Because this is an experimental measure, those who were
not currently employed have a value of 0 for this measure.

PCSTEMIN is a dummy variable that equals 1 if respondent reported paying out of
pocket for child care during their current or most recent job and was currently
employed at the time of the 5-year interview.

TRANSITIONAL CHILD CARE BENEFITS

1. Source Variables
G5CHCARR captures responses to the question "After you stopped getting cash aid
from AFDC, were any of your children under 13 in any kind of child care
arrangements?" This question refers to any time during the follow up period. For
this variable, 1=yes and 2=no.

G5WELFPY captures responses to the question "Did local JOBS program or the
welfare department pay for any of the cost of that child care?" This question
refers to any time during the follow up period. For this variable, 1=yes and
2=no.

*Note: The same questions were asked of respondents in the two-year survey
interview.

2. Skip Patterns
There were two questions about transitional child care benefits. The skip
patterns for these questions are as follows:

a) Only respondents who reported leaving welfare because of employment or
increased earnings (G5NOAFDC=1) were expected to answer the questions about
transitional child care benefits.

b) Respondents who answered that they did not have any children who were younger
than age 13 and in any kind of child care arrangements (G5CHCARR=2) skipped the
question about receiving transitional child care benefits.

3. Created Variables
CHCARR5 is a dummy variable which equals 1 if respondent reports using child
care after being off welfare due to employment or increased earnings during the
follow up period. If responses to the 2 year survey were available, they were
used to update this measure: If respondent answered yes to either survey, they
were coded as a yes for this variable.

PYCCST5 is a dummy variable which equals 1 if respondent reported that local
JOBS program or the welfare department paid for any of the cost of child care
after leaving welfare during the follow up period. If responses to the 2 year
survey were available, they were used to update this measure: If respondent
answered yes to either survey, they were coded as a yes for this variable.


V. WELFARE RECEIPT FROM JANUARY 1996 THROUGH THE MONTH BEFORE INTERVIEW
(SECTION H)

NOTE: The measures in this section were not used by MDRC for analysis of program
impacts.   MDRC did not attempt to resolve the few data problems found in this
section or reconcile inconsistencies in responses to welfare receipt data from
the Income Section (Section J) or from administrative records.

The section consists primarily of a series of yes/no questions about welfare
receipt for each month, starting with January 1996:  H5AD0196--H5AD0499.

Respondents are asked first if they ever received welfare beginning in January
1996 (H5ADC). If they answered NO (2), then the rest of the questions are
skipped. Values are set to missing, not NO (2).

If respondents answered YES (1), they are asked if they received welfare
continuously from January 1996 to the month before interview (H5CNTADC).  If
they answered YES (1), then all monthly receipt variables are set to 1 through
month before interview.   [NOTE: this wasn't done in a few cases.]

If they answered, NO (2), then the respondent is asked if she received welfare
during each month through month before interview.  Responses are 1=YES and
(missing)=no.


VI. MEDICAL COVERAGE AND NON-CASH BENEFITS

Section K (Non-Cash Benefits) of the survey asks about health care coverage for
respondents; section L (Child Outcomes) asks about health care coverage for
respondents' children; section G (Transitional Benefits) asks about transitional
health care coverage for respondents; section F (Employment) includes
information about employer-provided medical coverage.

COVERAGE FOR RESPONDENTS

1. Source variables
K5MEDICD captures answers to the question "In the month prior to the 5 year
interview, were you covered by Medicaid/MediCal?" For this variable, 1=yes and
2=no.

K5PRVINS captures answers to the question "In the month prior to the 5 year
interview, were you covered by any kind of private insurance?" For this
variable, 1=yes and 2=no.

2. Created Variables
XSELFMD5 is a dummy variable that equals 1 if respondent had health care
coverage in the month prior to the 5-year interview. Respondents were considered
to have health care coverage if 1) they answered yes to K5MEDICD, 2) they
answered yes to K5PRVINS, 3) they responded in the employment section that they
were working and had accepted an employer's health insurance plan in the month
before the 5 year interview, 4) they reported receiving welfare in the month
before the 5 year interview, or 5) they reported receiving SSI in the month
before the 5 year interview.

SMDPRIV5 is a dummy variable that equals 1 if respondent had private health care
coverage in the month prior to the 5-year interview. Respondents were considered
to have private coverage if they answered yes to K5PRVINS or if they reported
working and accepting an employer's health insurance plan in the month before
the 5 year interview.

SMDPUB5 is a dummy variable which equals 1 if respondent had public health care
coverage in the month before the 5-year interview. Respondents were considered
to have public coverage if they answered yes to K5MEDICD or if they reported
receiving welfare or SSI in the month prior to the 5-year interview.

*Note: SMDPRIV5 and SMDPUB5 are not mutually exclusive.

SMDPRPB5 is a dummy variable which equals 1 if respondent had both private and
public health care coverage in the month before the 5-year interview.

A series of dummy variables were created which combined respondents' health care
coverage and work and welfare status in the month prior to the 5-year interview.

EYADNSL5 = employed, not on welfare, had health care coverage
EYADYSL5 = employed, on welfare, had health care coverage
ENADYSL5 = not employed, on welfare, had health care coverage
ENADNSL5 = not employed, not on welfare, had health care coverage
EYADNSN5 = employed, not on welfare, did not have health care coverage
EYADYSN5 = employed, on welfare, did not have health care coverage
ENADYSN5 = not employed, on welfare, did not health care coverage
ENADNSN5 = not employed, not on welfare, did not have health care coverage

J5EMPMD is a dummy variable that equals 1 if respondent was employed and had
health insurance at the time of the 5-year interview.

3. 2 Year/5 Year Variables
A series of variables were created which combine responses to the 2-year and 5
year surveys. Only respondents who answered both surveys have values for these
variables. These variables are dummy variables that equal 1 when there are
different combinations of yes/no for health care coverage variables created from
the 2 year and 5 year surveys. The prefix for all of the variables is SMD, to
denote coverage for the respondent. The variables' suffixes denote the
following:

Y2Y5 = had health care coverage in the month prior to the 2 year interview and
the month prior to the 5 year interview
Y2N5 = had health care coverage in the month prior to the 2 year interview; did
not have coverage in the month prior to the 5 year interview
N2Y5 = did not have health care coverage in the month prior to the 2 year
interview; had coverage in the month prior to the 5 year interview
N2N5 = did not have health care coverage either in the month prior to the 2 year
interview or the month prior to the 5 year interview

COVERAGE FOR RESPONDENT'S CHILDREN

1. Source Variables
The questions that measure children's health care coverage from section L
(Child Outcomes) are all structured the same way. First, respondents are asked
if any of their children had a certain type of coverage. For example, "In the
prior month were any of your children covered by Medicaid/MediCal?" For the
remainder of this memo, these questions will be referred to as "general child
questions." If the respondent answered yes to the general child question, she
was then asked to say whether or not each child had that type of coverage. For
example, "Was child 1 covered by Medicaid; Was child 2 covered by Medicaid; Was
child 3 covered by Medicaid etc." For the remainder of this memo, these
questions will be referred to as "specific child questions." If a respondent
answered no to a general child question, she did not answer any of the specific
child questions and skipped directly to the next general child question. Answers
to specific child questions were only collected for 10 children (in this memo,
the children will be referred to as child 1 - child 10) even if the respondent
reported having more children. The numbers 01 through 10 at the end of the child
outcomes variable names refer to child 1 through child 10 respectively.

L515 is the variable which captures answers to the general child question "Since
(focal date), were there any periods when your child/any of your children were
not covered by health insurance or Medicaid/MediCal?" For this variable, 1=yes
and 2=no. Focal date is the date of the two-year interview if the respondent was
interviewed at the two-year point. If the respondent was not interviewed at the
two-year point, focal date is the date of random assignment. L515YN01-
L515YN10 are the specific child variables which correspond to this question. For
these variables, 1=yes and 2=no.

L516 is the variable which captures answers to the general child question "In
the month prior to the 5 year interview, was your child/were any of your
children covered by Medicaid/MediCal?" For this variable, 1=yes and 2=no.
L516YN01-L516YN10 are the specific child variables that
correspond to this question. For these variables, 1=yes and 2=no.

L517 is the variable that captures answers to the general child question "In the
month prior to the 5 year interview, was your child/were any of your children
covered by any kind of private insurance?" For this variable, 1=yes and 2=no.
L517YN01-L517YN10 are the specific child variables that correspond to this
question. For these variables, 1=yes and 2=no.

2. Created Variables
NCHLDMD5 is a dummy variable that equals 1 if all of respondent's dependent
children had health care coverage in the month before the 5-year interview.
Dependent children were children were who 18 or younger at the 5-year interview
point. All dependent children were considered to be covered if respondents
reported that each dependent child was covered either by Medicaid or by private
insurance (L516YNn=1, L517YNn=1) or if the respondent reported being covered by
welfare or SSI in the month before the 5 year interview or if the respondent
reported that none of her children had been without health insurance since the
last interview (L515=2).

NNCHMED5 is a dummy variable that equals 1 if none of respondent's dependent
children had health care coverage in the month before the 5-year interview.
Dependent children were children were who 18 or younger at the 5-year interview
point.

NSMCHMD5 is a dummy variable that equals 1 if some (but not all) of respondent's
dependent children had health care coverage in the month before the 5-year
interview. Dependent children were children were who 18 or younger at the 5-year
interview point.

NCMDPRV5 is a dummy variable that equals 1 if all of respondent's dependent
children had private health care coverage in the month before the 5-year
interview. Dependent children were children were who 18 or younger at the 5-year
interview point. All dependent children were considered to have public coverage
if respondent reported that each dependent child was covered by private sources
(L517YNn=1).

NCMDPUB5 is a dummy variable that equals 1 if all of respondent's dependent
children had public health care coverage in the month before the 5-year
interview. Dependent children were children were who 18 or younger at the 5-year
interview point. All dependent children were considered to have public coverage
if respondent reported that each dependent child was covered by Medicaid/MediCal
(L517YNn=1) or if respondent reported receiving welfare or SSI in the month
before the 5-year interview.

*Note: NCMDPRIV5 and NCMDPUB5 are not mutually exclusive.

NCMPRPB5 is a dummy variable which equals 1 if all of respondent's dependent
children had both private and public health care coverage in the month before
the 5-year interview. Dependent children were children were who 18 or younger at
the 5-year interview point.

COVERAGE FOR BOTH RESPONDENTS AND THEIR CHILDREN

1. Created Variables
A series of dummy variables were created using XSELFMD5 and NCHLDMD5 to capture
combinations of respondent and child health care coverage:

NOSLFMD5 equals 1 if only respondent had health care coverage (XSELFMD5=1) (all
children did not have coverage; NCHLDMD5=0)
NOCHDMD5 equals 1 if only all children had health care coverage (NCHLDMD5=1)
(respondent did not have coverage; XSELFMD5=0)
NSFCHMD5 equals 1 if both respondent and all dependent children had coverage
(XSELFMD5=1 and NCHLDMD5=1).
NNOMD5 equals 1 if neither respondent nor all dependent children had coverage
(XSELFMD5=0 and NCHLDMD5=0).

TRANSITIONAL MEDICAL BENEFITS

1. Source Variables
G5NOAFDC is the variable that captures the answers to the question, "At any time
since random assignment, did you stop getting AFDC because you got a new job or
your earnings increased at your old job?" For this variable, 1=yes and 2=no.

G5MEDCOV is the variable that captures the answers to the question, "While you
were employed, were you or your spouse or children covered by Medicaid/MediCal
even after you stopped getting cash aid from AFDC?" This question refers to any
time during the follow up period. For this variable, 1=yes and 2=no. Only
respondents who answered yes to G5NOAFDC were expected to answer G5MEDCOV.

*Note: The same questions were asked of respondents in the two-year survey
interview.

2. Created Variables
VADSTP25 is a dummy variable that equals 1 if respondent reports being off
welfare due to employment or increased earnings during the follow up period. If
responses to the 2-year survey were available, they were used to update this
measure: If respondent answered yes to either survey, they were coded as a yes
for this variable.

VMDCOV25 is a dummy variable that equals 1 if respondent reported receiving
Medicaid/MediCal after leaving welfare. If responses to the 2-year survey were
available, they were used to update this measure: If respondent answered yes to
either survey, they were coded as a yes for this variable.



VII. HOUSEHOLD COMPOSITION  (SECTION I)

This section contains variables about marital status, household composition and
moving and residential status.

The entire survey sample was expected to answer questions in Section I
(Household Composition); there is no initial filter to pass through to be
eligible to answer the questions.  It was expected that each question would be
asked of all respondents, except where skip instructions were involved.

MARITAL STATUS

1. Source Variables
I5MARR captures answers to the question "Have you ever been married?" For this
variable, 1=yes and 2=no.

I5MARST captures marital status in the month prior to the 5 year interview.
Values for this variable are as follows:
1=married and living with spouse
2=separated or living apart from spouse
3=divorced
4=widowed

I5COUPLE captures answers to the question "In prior month were you living as a
couple with a boyfriend or partner?" For this variable, 1=yes and 2=no.

2. Skip Patterns
There were three questions about marital status in the survey. The skip patterns
for these questions are as follows:

a) If the respondent indicated she has never been married (I5MARR=2), then she
was not supposed to be asked if she was currently married, separated, divorced
or widowed.

b) If the respondent reported that she was married and living with her spouse in
the month prior to the interview (I5MARST=1), then she was not supposed to be
asked whether she was living as a couple with a boyfriend or partner.

3. Created Variables
Using responses to the three questions about marital status, four mutually
exclusive marital status variables were created to represent marital status in
the month before the 5 year interview:
I5SPOME  = Married
I5SNVMME = Single, never married
I5SNLMME = Single, no longer married
I5CCHME  = Living as a couple with boyfriend or partner

The marital status variables were created as follows:

If respondent reported being married and living with spouse (I5MARST=1), then
she was coded as married (I5SPOME=1).

If respondent reported having never been married (I5MARR=2) and reported not
living as a couple (I5COUPLE=2), then she was coded as single, never married
(I5SNVMME=1).

If respondent reported ever being married (I5MARR=1) and reported being
separated, divorced or widowed (I5MARST=2,3,4) and reported not living as a
couple (I5COUPLE=2), then she was coded as single, no longer married
(I5SNLMME=1).

If respondent reported living as a couple (I5COUPLE=1), then she was coded as
living as a couple with boyfriend or partner (I5CCHME=1).

Some respondents had missing marital status information and some respondents
reported marital status information in the 5-year survey that was discrepant
with the household composition section of the 5-year survey. If this occurred,
coding was as follows:

1. If married (I5MARST=1) but do not list a spouse or partner as a household
member (I5HH01-I5HH11 not equal 1) then still coded as married (I5SPOME=1).

2. If not ever married (I5MARR=2) and missing response to the question about
cohabitation (I5COUPLE=.) then all marital status variables coded as missing.

3. If living as a couple with a boyfriend or partner (I5COUPLE=1) but do not
list a spouse or partner as a household member (I5HH01-I5HH11 not equal 1) then
coded as living as a couple with a boyfriend or partner (I5CCHME=1).

4. If missing marital status information (I5MARR=., I5MARST=., I5COUPLE=.) then
all marital status variables coded as missing.

HOUSEHOLD COMPOSITION
1. Source Variables
Respondents were asked to list all people who stayed in their household for at
least 2 nights a week during the month prior to the 5-year interview.
I5HH01-I5HH11 capture the relationships of these household members to the
respondent. For values, please see the codebook.

2. Created Variables
A series of dummy variables were created to capture household membership. Each
variable equals 1 if at least one person of that relationship is respondent's
household member:

I5HHSPO   =  spouse or partner
I5HHKID   = son(s) and/or daughter(s)
I5HHPAR   = with parent(s)
I5HHSIB   = with sibling(s)
I5HHGPAR  = grandparent(s)
I5HHUA    = uncle(s) and/or aunt(s)
I5HHCOS   = cousin(s)
I5HHNN    = nephew(s) and/or niece(s)
I5HHREAD  = other adult relatitve(s), including in-law(s)
I5HHUNAD  = unrelated adult(s)
I5HHUNCH  = unrelated child(ren)
I5HHGCHD  = grand child(ren)
I5HHALON  = alone. [Did not list any other people who stayed in household at
least 2 night a week in month prior to 5 year interview]

Using the dummy variables listed above, another series of dummy variables were
created to capture combinations of people in respondent's household:
I5SDONLY  = live with son(s) and/or daughter(s) only
I5CHONLY  = live with child(ren) only
I5RCHONL  = live with related child(ren) only
I5UCHONL  = live with unrelated child(ren) only
I5SPKID   = live with spouse or partner and son(s) and/or daughter(s)
I5SPACHD  = live with spouse or partner and any child(ren).
I5RELTV   = live with any relative(s).
I5ARLTV   = live with adult extended family/relatives
I5ARCHD   = live with adult extended family/relatives and child(ren).
I5UADONL  = live with unrelated adult(s) only. [*note: This measure does not
necessarily capture cohabiting partners.]
I5UADCHD  = live with unrelated adult(s) and child(ren).

For the above variables, the following definitions were used:
-Children are defined as anyone whose generation is lower than respondent's and
includes: son/daughter, nephew/neice, grandson/granddaughter, unrelated children
- Related  children are defined as any relatives whose generation is lower than
respondent's and includes: son/daughter, nephew/neice, grandson/granddaughter
-Adult extended family/relatives include parent(s), sibling(s), grandparent(s),
cousin(s), aunt(s), uncle(s), other adult relatives including in-laws

HOUSING STATUS

1. Source Variables
 I5LIVARR captures housing status in the month prior to the 5 year interview.
Values for this variable are as follows:
1=own your home
2=rent your home/room
3=live with family or friends and not pay rent
4=live with family or friends and contribute to part of rent
5=live in a group shelter
6=live in some other housing arrangement
7=jail
8=homeless
9=alone and rent free

I5MOVED captures answers to the question "Have you moved since random
assignment?" For this variable, 1=yes and 2=no.

I5NOMOVE captures answers to the question "Have you moved more than once since
random assignment?" For this variable, 1=yes and 2=no.

If respondents indicated that they had moved during random assignment, they were
asked why they had moved. They were allowed to give up to three reasons. These
reasons are captured in I5MOVE1, I5MOVE2 and I5MOVE3. For values, please see the
codebook. Respondents were also asked which of the listed reasons was most
important. Responses to this question are captured in I5MOVEI. For this
variable, 1=the 1st reason was the most important, 2=the 2nd reason, 3=the 3rd
reason. If respondents only gave 1 reason, then they were automatically given a
value of 1 for I5MOVEI.

K5PUBHSE captures answers to the question "do you live in public housing, that
is, housing owned or operated by a local housing authority or other government
agency?" For this variable, 1=yes and 2=no.

K5SECT8 captures answers to the question "Does your household pay less rent
because the government pays for part of it, such as in Section 8 housing?" For
this variable, 1=yes and 2=no.

2. Skip Patterns
There were 6 questions about moving and housing status in the survey, 4 from
section I (Household Composition) and 2 from section K (Non-Cash Benefits).
[NOTE: The variables from section K (Non-Cash Benefits) capture public and
subsidized housing and are grouped here with variables from section I (Household
Composition) about housing status. Researchers could also use measures from
section K for research on non-cash benefits.] This skip pattern for these
questions are as follows:

a) If the respondent reported that she had not moved since random assignment
(I5MOVED=2) then she was not supposed to be asked whether she moved more than
once and she was not supposed to be asked her reasons for moving.

3. Created Variables
I5DMOVE is a dummy variable that equals 1 if respondent reported moving in since
random assignment.

I5DNMOVE is a dummy variable that equals 1 if respondent reported moving more
than once since random assignment

A series of dummy variables were created to capture housing status at the 5 year
follow up point:
I5OWNHM  = own home.
IK5PUBHS = live in public housing
IK5RSUB  = live in subsidized housing
I5RENT   = rent home or room
I5OTHER  = live in other housing. [This includes group shelter, jail, alone and
rent free, homeless and other housing arrangement.]
IK5PHSB  = live in either public or subsidized housing

*Note:  If a respondent reports living in public housing, living in jail, or
having their rent subsidized (K5PUBHSE=1, I5LIVARR=7, or K5SECT8=1), AND reports
owning a home (I5LIVARR=1), then home ownership (I5OWNHM) is recoded to 0.

When creating the dummy variables above, any response about public or subsidized
housing from section K (K5PUBHSE, K5SECT8) overrides responses to housing from
section I (I5LIVARR).  (For example, the 10 respondents who report living with
family and friends and not paying rent in section I (LIVARR=3) and report having
their rent subsidized in section K (K5SECT8=1) will be coded as "subsidized
housing" (IK5SUB=1) only.)

EVFUPCH is a dummy variable that equals 1 if a new baby was present since random
assignment. This variable was created using the child birthdates from section L
and the random assignment date. The variable does not measure respondents'
fertility during the follow up. Because it is only based on birthdates, and not
on a question about having a child, EVFUPCH only measures presence of a baby.
This baby could have been adopted or could be a stepchild. So, the variable can
only be used a proxy for fertility but not as a direct measure of fertility.

UNRELADR is a dummy variable that equals 1 if respondent was living with an
unrelated adult in the month prior to the 5 year interview. To be consistent
with the U.S. census, cohabiting partners are included in this definition of
unrelated adults.

2-YEAR/5-YEAR VARIABLES

A series of variables were created which combine responses to the 2-year and
5-year surveys. Only respondents who answered both surveys have values for these
variables. These variables are dummy variables that equal 1 when there are
different combinations of yes/no answers to the created variables from the 2
year and 5 year survey. If coding rules were different at the 2-year follow up
point, 2-year measures were recoded to be consistent with 5 year measures before
this series of measures was coded. All of the variables were constructed the
same way. The variables' prefixes denote the following:
EV  = yes to either the 2 year or the 5 year variable
BC  = no to the 2 year variable, yes to the 5 year variable
STL = yes to both the 2 year and 5 year variables
NL  = yes to the 2 year variable, no to the 5 year variable
NEV = no to both the 2 year and 5 year variables

The suffixes are as follows:
MARR = married
LVAL = live alone
CHD  = live with any children
PHSB = live in public or subsidized housing
UADR = live with unrelated adults [this includes cohabiting partners]


VIII. SOURCES OF INCOME (SECTION J)

The Five-Year Client Survey collected information on seven types of income
during the month before interview.  Respondents' answers all begin with the
prefix J5.  The income types can be distinguished by the letters used in the
stem of the variable names:



INCOME TYPE                 STEM

Employment                 ...JB
Welfare                    ...ADC
Food Stamps                ...FS
Child support payments     ...CS
Social Security or pension ...SSA
SSI or Disability          ...SSI
"Other"                    ...OTHER


Respondents are asked about income for


RECIPIENT TYPE                            STEM

Themselves                               ..R...
All other household members combined     ..H...
Anyone in the household                  [none]


The questions follow a set sequence for each income type:

Did anyone in household receive income?

   If NO: skip to the next income type

   If YES:  1) Did respondent receive income?

                  If YES:  Ask for dollar amount

                           If respondent answers DON'T KNOW: Ask respondent if
                           income fell within a range of values
                           [see codebook for ranges]

                           Ask about other household members


                   If NO:  Ask about other household members


            2) Did anyone else in the household receive income?

                  If YES:  Ask for total dollar amount for all household members


                           If respondent answers DON'T KNOW: Set to MISSING.
                           Ask respondent to estimate income within a range of
                           values [see codebook for ranges]

                  If NO:   Skip to next income type


For all income measures:

If the respondent answers NO: the remaining questions in the sequence are
skipped and their values are set to MISSING.

The actual dollar amount and estimated dollar amount are stored as
separate measures, which can be distinguished by the suffix in the variable
name:

Actual:    ...AMT

Estimated: ...EST



ADDITIONAL OUTCOME MEASURES

A series of 0/1 variables were created from the original series of yes/no
questions.  These measures were initialized to 0 (rather than MISSING, as was
done for the original series.



These measures are named with:

Prefix:  VJ5.....

Income Type:  Same as above, except that OTHER is truncated to OTHR or OTH

Recipient type:

Respondent:               ...R....
Other household members:  ...H....
Any household members:    ...HH...

The file also includes an alternative measure of employment in the month before
interview,  VF5RJB, which is based on the start and end dates of respondent's
most recent job, recorded in the Employment section (Section F).  For
consistency with other employment information, this measure was used in the
tables in [add report name] and to create other measures listed below that
include information on respondent's employment. This measure was also used to
create an alternative indicator of employment for any household member: VF5HHJB.


Additional outcome measures based on the 0/1 variables include:

Respondent's reported employment and welfare status in the month before
interview:

             Employed?         On Welfare?

ERYADRN5     yes                  no

ERYADRY5     yes                  yes

ERNADRN5     no                   no

ERNADRY5     no                   yes


Employment status for the respondent and for other household members for the
month before interview:
                              Other
             Respondent       Household Members
             Employed?        Employed?

ERYEHN5      yes                  no

ERYEHY5      yes                  yes

ERNEHN5      no                   no

ERNEHY5      no                   yes





Total sources of income:  Prefix: KJ5.....

For respondent:                   KJ5RINC
For other household members:      KJ5HINC
For any household member:         KJ5HHINC


Have any income any source (0/1): Prefix: VJ5.....

For respondent:                   VJ5RINC
For other household members:      VJ5HINC
For any household member:         VJ5HHINC


Received income other than earnings, welfare, and Food Stamps (0/1):

VJ5HHSTH: Other household member received: SSI, SSA, or other income

R5ADNSIY: Respondent received no welfare, other household member receivd SSI

VJ5RSTHR: Respondent received SSI,SSA, or other income

VJ5HSTHR: Other household member received: SSI,SSA, or other income




MEASURES BASED ON INCOME AMOUNT (J5...AMT  and J5...EST)

MDRC only created a series of 0/1 variables from these measures:


Earned $1,000 or more: for the respondent and (total earnings) for other
household members for the month before interview:


VF5R1000: Respondent

VJ5H1000: Other household members

V5HH1000: Any household member

[NOTE: V5HH1000=1 IF VF5R1000=1 OR VJ5H1000=1.  This measure does not indicate
that the SUM of all earnings among all household members exceeded $1,000.]


VERNH1K5: Other household members, when respondent not employed


MDRC did not create measures that show dollar amounts of total income.
Researchers who choose to do so will need to sum the actual amount (for those
who reported it) with the estimated amount (for those who estimated).
Researchers will need to assign a value to each range -- most likely the
midpoint. However, researchers will need to devise special rules for
assigning dollar values for respondents who place the dollar value within the
highest range, which is open-ended. For example: measures of estimated income
from employment for respondent and for "all other household members" have the
top range "More than $1,000."

Note: for respondent's income from employment: researchers may use an
alternative measure based on calculated pre-tax weekly earnings (multiplied by
4.33) from the Employment Section: F5RJBARN.



Also:
MDRC did not attempt to reconcile contradictions in values for similar
variables, either:

1) Across sections

   a) Employment/Earnings:  Employment Section (F) /Income Section (J)

   b) Welfare Payments: Welfare Payment History Section (H)/ Income Section (J)

2) With administrative data stored on the Full Sample Impact File.



IX. OUTCOMES FOR ALL OF RESPONDENTS' CHILDREN (SECTION L)

L501 asked respondents how many children they had. Respondents were supposed to
include birth children, legally adopted children and step children whether or
not they lived with the respondent. If respondents answered 0 to this question,
they skipped all subsequent questions in this section. If respondents refused
to answer this question, they skipped all subsequent questions in this section.

The questions which measure child outcomes in this section are all structured
the same way. First, respondents are asked if any of their children had done a
certain thing. For example, "Have any of your children repeated a grade?" For
the remainder of this memo, these questions will be referred to as "general
child questions." If the respondent answered yes to the general child question,
she was then asked to say whether or not each child had done a certain thing.
For example, "Did child 1 repeat a grade; Did child 2 repeat a grade;  Did
child 3 repeat a grade etc."  For the remainder of this memo, these questions
will be referred to as "specific child questions."

If a respondent answered no to a general child question, she did not answer any
of the specific child questions and skipped directly to the next general child
question. Although respondents could list any number of children in L501,
answers to specific child questions were only collected for 10 children (in
this memo, the children will be referred to as child 1 -  child 10).
The numbers 01 through 10 at the end of the child outcomes variable names refer
to child 1 through child 10 respectively.

L5FC01-L5FC10 are indicator variables which equal 1 if that child is the focal
child who is part of the Child Outcomes Study. If the child is not the focal
child, these variables equal missing.

L506 is the variable which captures answers to the general child question
"Since (focal date), has your child/have any of your children had an accident
injury or poisoning requiring a visit to a hospital emergency room or clinic?"
For this variable, 1=yes and 2=no. Focal date is the date of the two year
interview if the respondent was interviewed at the two year point. If the
respondent was not interviewed at the two year point, focal date is the date of
random assignment. L506YN01-L506YN10 are the specific child variables which
correspond to this question. For these variables, 1=yes and 0 or 2=no.

L507 is the variable which captures answers to the general child question
"Since (focal date), has your child/have any of your children not lived with
you because you couldn't care for him/her them?" For this variable, 1=yes and
2=no. Focal date is the date of the two year interview if the respondent was
interviewed at the two year point. If the respondent was not interviewed at the
two year point, focal date is the date of random assignment. L507YN01-L507YN10
are the specific child variables which correspond to this question. For these
variables, 1=yes and 0=no.

L508 is the variable which captures answers to the general child question "Has
your child/have any of your children attended kindergarten, first grade or a
higher grade in school?" For this variable, 1=yes and 2=no. L508YN01-L508YN10
are the specific child variables which correspond to this question. For these
variables, 1=yes and 2=no. L508 was the filter question for three questions
about school outcomes (repeat a grade, suspended/expelled, drop out). If
respondent answered no to L508, she skipped the three questions that followed
and skipped to question L-12. If respondent answered no to any of the specific
child questions, she skipped the three specific child questions that followed
for that child.

L509 is the variable which captures answers to the general child question
"Since (focal date), has your child/have any of your children repeated a grade
for any reason?" For this variable, 1=yes and 2=no. Focal date is the date of
the two year interview if the respondent was interviewed at the two year point.
If the respondent was not interviewed at the two year point, focal date is the
date of random assignment. Only respondents who answered yes to L508 were
expected to answer this question. L509YN01-L509YN10 are the specific child
variables which correspond to this question. For these variables, 1=yes and
0=no.

L510 is the variable which captures answers to the general child question
"Since (focal date), has your child/have any of your children ever been
suspended, excluded or expelled from school?"  For this variable, 1=yes and
2=no. Focal date is the date of the two year interview if the respondent was
interviewed at the two year point. If the respondent was not interviewed at the
two year point, focal date is the date of random assignment. Only respondents
who answered yes to L508 were expected to answer this question.
L510YN01-L510YN10 are the specific child variables which correspond to this
question. For these variables, 1=yes and 0=no.

L511 is the variable which captures answers to the general child question "Has
your child/have any of your children ever dropped out of school?" For this
variable, 1=yes and 2=no. A time frame is not specified for this question. Only
respondents who answered yes to L508 were expected to answer this question.
L511YN01-L511YN10 are the specific child variables which correspond to this
question. For these variables, 1=yes and 0=no. L511AY01-L511AY10 are specific
child variables which capture the answers to the question "Did child return to
school after dropping out?" Respondents only answered these questions for the
children who they reported had dropped out. Thus, for each child that a
respondent answered yes to L511YN01-L511YN10 then she subsequently answered
L511AY01-L511AY10 for the same child. For L511AY01-L511AY10, 1=yes and 2=no.

L512 is the variable which captures answers to the general child question "Does
your child/do your children go to a special class or special school, or get
special help in school for any physical, emotional, or mental condition?" For
this variable, 1=yes and 2=no. The time period is at the time of the five year
survey. L512YN01-L512YN10 are the specific child variables which correspond to
this question. For these variables, 1=yes and 0=no.

L513 is the variable which captures answers to the general child question "Does
your child/do your children have a physical, emotional, or mental condition
that demands a lot of your attention and makes it hard for you to go to school
or work?" For this variable, 1=yes and 2=no. The time period is at the time of
the five year survey. L513YN01-L513YN10 are the specific child variables which
correspond to this question. For these variables, 1=yes and 0=no.

L514 is the variable which captures answers to the general child question "Does
your child/do  your children have a physical, emotional, or mental condition
that requires frequent medical  attention, frequent use of medication, or the
use of any special equipment such as a wheelchair or  a breathing mask? [does
not count eyeglasses]" For this variable, 1=yes and 2=no. The time period is at
the time of the five year survey. L514YN01-L514YN10 are the specific child
variables which correspond to this question. For these variables, 1=yes
and 0=no.

L518 is the variable which captures answers to the general child question
"Has your child/have any of your children had a baby?" For this variable, 1=yes
and 2=no. A time frame is not specified for this question. Only respondents who
answered yes to L508 were expected to answer this question. L518YN01-L518YN10
are the specific child variables which correspond to this question. For these
variables, 1=yes and 2=no. L518AG01-L518AG10 are specific child variables which
capture the answers to the question "How old was child when he or she had his
or her first baby?" Respondents only answered these questions for the children
who they reported had had a baby. Thus, for each child that a respondent
answered yes to L518YN01-L518YN10, she subsequently answered L518AG01-L518AY10
for the same child. For L518AG01-L518AG10, respondents were free to answer any
age in years.

DOB1-DOB10 are SAS dates for the birthdates of each child. If respondent did
not give a day, month and year of birth, then DOBn will be missing for that
child.

AGERAD1-AGERAD10 are the ages of each child at random assignment. These
variables were created from the DOB1-DOB10 and the random assignment date. If
DOBn is missing, AGERADn will be missing for that child.

GT18CH1-GT18CH10 are dummy variables which equal 1 if the child was greater
than age 18 at random assignment. Respondents answered the questions in this
section for all of their children, regardless of age. All analyses conducted
for the final NEWWS report, however, were limited to children who were 18 or
younger at random assignment. These flags were used to identify children to be
excluded from the analyses. These variables were primarily created with
AGERAD1-AGERAD10. Those with missing age at random assignment could still have
a 1 for these variables if the respondent indicated that the child had had a
baby at age 25 or greater.

INTAGE1-INTAGE10 are the ages of each child at the time of the 5-year
interview. These variables were created from DOB1-DOB10 and the 5-year
interview date. If DOBn is missing, INTAGEn will be missing for that child.

For each set of general and specific child questions, one dummy variable was
created which equals 1 if any children in the family had experienced/done that
thing and 0 if no children in the family had experienced/done that thing. These
variables limited any children to children who were 18 or younger at random
assignment.

VHOSP5 corresponds to L506YN01-L506YN10
VNOTLIV5 corresponds to L507YN01-L507YN10
VREPGR5 corresponds to L509YN01-L509YN10
VSUSP5 corresponds to L510YN01-L510YN10
VDROP5 corresponds to L511YN01-L511YN10
VSPCLS5 corresponds to L512YN01-L512YN10
VEMPROB5 corresponds to L513YN01-L513YN10
VMEDPROB5 corresponds to L514YN01-L514YN10

All time periods for these dummy variables are the same as for the sources
variables listed above. For the remainder of this memo, these dummy variables
will be referred to as any child outcomes.

A respondent can get a 1 (yes) to an any child outcome in two ways:

1) At least one of her APPROPRIATELY AGED children (aged 18 or younger at RAD)
has a yes to the specific child question.

2) None of her children have a yes; at least one appropriately aged child has
as missing; and she responded yes to the general question.

A respondent can get a 0 (no) to an any child outcome in two ways:

1) Respondent answered no to the general question.

2) None of her appropriately aged children has a yes to the specific child
questions.

***Additionally, for outcome measures based on questions L9-L11, respondents
will get a 0 if none of their children had attended school. This applies to any
child repeated a grade, any child suspended, any child dropped out. For these
outcomes, if L508 is NO (no children attended school), then respondent gets a 0
(no).

A respondent can get a . (missing) to an any child outcome in three ways:

1) Respondent has 0 appropriately aged children. This could be because she
answered L501 (How many children do you have?) 0 or because all of the children
listed were 19 or older at random assignment.

2) Respondent refused to answer the answer L501

3) Missing for the general question.

An additional any child measure was created to capture any children ever had a
baby as a teen: VBABY5.

This measure was created from L518AG01-L518AG10.

A respondent gets a 1 (yes) for this measures if:

1) At least one appropriately aged child had a baby at age 18 or younger.
2) Respondent answers yes to L518 (Have any of your children had a baby?) and
answers the specific child follow up questions for each child but cannot answer
(missing) the age when each child had a baby. In this situation, if at least
one of the children who have a yes to the specific child question was 18 or
younger at the 5-year interview, the respondent gets a 1 (yes) to the any child
question. We decided this because we are capturing teen pregnancy. If we limit
it to those who were teens during the entire follow up period, we can code them
as teen parents even though we do not know the exact age that they gave birth.

Zeros and missings are calculated in the same manner as described for the other
any child outcomes above.

A series of variables were created which combine responses to the 2 year and 5
year surveys. Only respondents who answered both surveys have values for these
variables. These variables are dummy variables which equal 1 when there are
different combinations of yes/no answers to the any child questions from 2 year
survey and the created any child variables from the 5-Year Survey. All of the
variables were constructed the same way. The variables' suffixes denote the
following:

YFUP = yes to either the 2 year or the 5 year variable

N2Y5 = no to the 2 year variable, yes to the 5 year variable

Y2Y5 = yes to both the 2 year and 5 year variables

NFUP = no to both the 2 year and 5 year variables

The prefixes are as follows:

HOSP = any child had an accident, injury or poisoning requiring a visit to a
       hospital emergency room

SUSP = any child suspended, expelled or excluded from school

RPGR = any child repeated a grade

NLIV = any child did not live with respondent because respondent couldn't are
       for child

PROB = any child had physical, emotional or mental condition that demanded a
       lot of attention and made it hard for respondent to go to work or school


X. ATTITUDES TOWARD WORK AND WELFARE (SECTION M)

This section consists of 8 statements about work and welfare to which the
respondent is supposed to state the extent to which she agreed or disagreed.
Values range from 1 (agree a lot) to 5 (disagree a lot).

Respondents are then asked two questions related to their "reservation wage.":

M5LOWAGE: If respondent were unemployed, what is the lowest wage per hour she
          would accept for a full-time job with full medical benefits

M5FAIRWG: If respondent were unemployed, what would be a fair wage per hour
          for a full-time job with full benefits, given her skills and
          experience


NOTE:  These questions were asked of all respondents, irrespective of whether
they were working at the time of their 5-year interview. The responses may be
used together with responses from other sections and surveys, including

1) Questions on costs of child care and transportation for employment; why
respondents left their most recent job; and why did not accept a job offer
(Employment: Section F)

2) [For Child Outcomes Study Survey respondents:] Questions on domestic violence
and barriers to work (Working: Section GG)

3) Questions on attitudes toward work and welfare from the 2-Year Client Survey

4) Questions on attitudes toward work and welfare from the Private Opinion
Survey administered on the day of random assignment
