The Household-Level Identifiers

Restricted-Use Data

Each household received two household identifiers.  The first identifier was assigned by an interviewer within the household’s respective village on the compiled household list (Form H) for that village.  This identifier is only unique in combination with the household's village identifier.  The second identifier was recorded by an interviewer on the household’s questionnaire.  This identifier includes the village number and the house number (or unique sequential number in 1994) and is unique on its own.  Below are the variable names for each of the household identifiers:

Identifier Source	1984 Survey	1994 Survey	2000 Survey
Form H	HOUSE84	LEKTI94	HOUSE00
Questionnaire	HHID84	HHID94	HHID00

Each of the household identifiers that come from Form H are in alphanumeric format (with leading zeroes) with a length of 3 digits.   The 1994 variable is named LEKTI94, since the number is basically the ban lek ti, which is an administrative number assigned by the government.   The 1984 and 2000 variables are simply sequential numbers assigned by the interviewers.   The questionnaire identifiers are different a bit different across the three survey years.  The 1984 household questionnaire identifier is in alphanumeric format with a length of 9 digits.  HHID84 can be broken down into several different parts, but the main parts are the village number and the house number.  (Note:  Due to the fact that HHID84 did not appear in the data until 1994, this variable has been rarely used.  Most analysts will use VILL84 and HOUSE84 in combination to uniquely identify a household in 1984.)  The 2000 household questionnaire identifier is in alphanumeric format (with leading zeroes) with a length of 9 digits.  HHID00 is simply the village number concatenated to the house number.  The 1994 household questionnaire identifier is the exception to the rule as it is a numeric variable and is not in alphanumeric format.  It has a length of 8 digits, which can be broken down into three parts:  the village number, a sequential number that is not related to LEKTI94, and a final digit that indicates whether the household is old or new.
 

Public-Use Data

Due to reasons of confidentiality, a new set of household identifier variables was created for the public-use data archive.  The values for this new set of variables were randomly generated based on the original values from the original household identifiers.  Each variable name has a "PHHID" prefix followed by the last two digits of the survey year:

1984 Survey	1994 Survey	2000 Survey
PHHID84	PHHID94	PHHID00

The public-use household identifiers are in numeric format, start with 1 and increase to the total number of households in that particular survey year.

---

Household-Level Concepts & Issues...

Old & New Households in 1994 and 2000

The concepts of old and new households first appeared in the data during the 1994 survey and continued through the 2000 survey:  

Old households are those households in 1994 and 2000 that existed in the village in the prior round(s) of data collection, who had at least one member still living in the village in the current round of data collection.  For the 1994 survey, the household had to exist in the 1984 survey to be considered an old household.   For the 2000 survey, the household had to exist in either the 1994 survey or the 1984 survey. 

New households are those households in 1994 and 2000 that were formed in the years between rounds of data collection.  For the 1994 survey, new households were those households that formed between 1984 and 1994.  For the 2000 survey, new households were those households that formed between 1994 and 2000.

In both 1994 and 2000, the old and new households completed separate and appropriately named questionnaires - Old Household Questionnaire and New Household Questionnaire.  


The Old Household Designated Individual

For old households in the 1994 and 2000 surveys, there could be one and only one household that corresponded to each household from the previous survey year(s).  In order to identify this one and only household, or successor household, the interviewers looked for a designated individual on the household roster from the previous survey year.  The search for the designated individual started with the senior generation and went to the more junior generations, and from the females to the males.  For some households, the search was a bit complex given the various ways the households changed between the rounds of data collection.  Some households split into two or more households, but the rule again was that one and only one was designated as the successor household.  Once a designated individual was found, then the household that s/he lived in was designated as the successor household.  The successor household was then labeled an old household and was administered the Old Household Questionnaire.   

Combined Old Households in 1994 and 2000

In general, a combined old household is when two or more households, in 1994 or 2000, have combined into one household since the last round of data collection.  This concept is mainly a product of the way the successor household was defined using the designated individual (see description above).  Basically, if the designated individuals from two or more households lived in the same household, then that household became a combined old household.  During the fieldwork, a separate Form 1 was completed for each household and only one set of Forms 2-6 were completed for the entire combined household.  Below is an example of a combined old household in the 1994 data:

COMHH	CEP94	Q1	Q2	Q3	VILL84	HOUSE84	CEP84	FORM	HHTYPE94
1	101	0	98	8	04	042	101	1	3
1	102	1	67	2	04	042	102	1	3
1	103	2	32	1	04	042	103	1	3
1	104	2	23	1	04	042	104	1	3
1	201	1	37	2	04	042	201	1	3
1	202	1	28	2	04	042	202	1	3
1	203	3	26	1	04	042	203	1	3
8	401	4	5	2				2	3
8	402	4	14	2	04	035	106	2	3
2	403	1	11	2	04	060	103	1	3
2	491	3	36	1	04	060	101	1	3
2	492	3	38	2	04	060	102	1	3

There are several things to note from the roster: 

1. The COMHH column indicates from which 1984 household roster the individual originated.   A code of '1' means the individual came from what has been termed the primary household.   This is the household that completed the first Form 1 during the fieldwork.  A code of '2' means the individual came from the secondary household or the household that completed the second Form 1.  There were no combined households with a tertiary household in 1994, but there was one in the 2000 survey.    The '8's are not applicable codes as these two individuals are new to the household (not on the primary's or secondary's household roster in 1984) and were listed on Form 2.
   
   
2. In the CEP94 column, there are some unusual numbers to note at the bottom, '491' and '492.'  These numbers, and all numbers in the 490's, were assigned after the fieldwork to individuals from the secondary households who did not currently live in the combined old household.  These unusual CEP numbers only occur in the 1994 data.   In 2000, the CEP numbers are sequential for all members on the household roster. 
   
   
3. A quick look at the Q1-Q3 columns shows that person '102' was the designated individual for the primary household (i.e. oldest female living in the village), while person '403' was the designated individual for the secondary household.  She was young, but her parents, '491' and '492',  had migrated (Q1=3) outside the village.
   
   
4. The 1984 identifiers - VILL84, HOUSE84, CEP84 - verify that the primary and secondary households link back to two different households in 1984.   Hence, this one combined old household was the successor household for these two 1984 households.  In the 1994 household-level data, there are two sets of linking identifiers for the combined households - VILL84 and HOUSE84 for the primary household and SVILL84 and SHOUSE84 for the secondary household.  Please note that although person '402' links back to yet another 1984 household, she was not the designated individual for her original 1984 household. 
   
   
5. Finally, combined old households and individuals in the combined old households are always coded as '3' for the HHTYPE94 variable.

The 2000 combined old households are very similar to the above 1994 example.  For both surveys, the number of combined old households is a very small percentage of the total number of households, although there was an increase from 1994 to 2000.  There were 41 combined households in 1994 and 124 combined households in 2000. 

Moved Households in 1994 and 2000

Moved households are those households in 1994 and 2000 which existed in the village in the prior round(s) of data collection that did not have at least one member still living in the village in the current round of data collection.  All of their members had either moved out of the village or died since the last round of data collection.  In 2000, this included households that moved since the 1994 survey as well as households that moved since the 1984 survey.  In 1994, this included only households that moved since the 1984 survey. 

In the 2000 survey, an informant(s) from the village helped the interviewers complete the Moved Household Questionnaire, called Form HM, which noted the location and living status of each of the former members of the household.  In the 1994 survey, however, there was no formal questionnaire for the moved households.  Rather, the interviewers obtained information from an informant(s) in the village and simply recorded notes on the photocopy of the household's 1984 questionnaire.  These notes mainly pointed to the new location of the moved household. 

Given the different methods of obtaining information, the identifiers for the moved households are not consistent from 1994 to 2000.   Since there was no formal questionnaire in 1994, the moved households in that survey year were not assigned 1994 identifiers.  The only identifiers attached to these households are their original 1984 identifiers.   The 2000 moved households, on the other hand, did have a formal questionnaire and subsequently were assigned 2000 identiifiers.