Summary of 1980 Census Edit and Allocation Procedures1

Go Back to Editing and Allocation Procedures Index

RELATE, SEX, AGE, and MARST

A. Purpose of Edit Procedures

The purpose of the edit and allocation procedures for the items on household relationship, sex, age, and marital status was to insure that the data for each person on the final files [whether from the "short" form questionnaire or from the "long" form (sample) questionnaires] include allowable values for each of the four characteristics. The procedures also insured that certain consistencies were maintained among these highly interrelated characteristics.

The 100-percent portion of each questionnaire appeared as the first two pages of questions on the questionnaires (pages 2 and 3 of both the short form and the long form). In the complete-count processing, pages 2 and 3 of all short-form and long-form questionnaires were microfilmed in one exposure; no other pages of the long forms were filmed at that time. There was no manual editing of these pages in the central processing office before the original microfilming. In the sample processing, the entire long-form questionnaire was microfilmed after clerical coding of selected write-in entries had been completed. Due to the availability of additional information for persons on the sample questionnaire, separate edit procedures were prepared for the 100-percent items on the complete-count and sample data files. However, the two edits of the 100 percent items differed only in minor respects. The following description will refer basically to the sample edit, with appropriate remarks inserted to indicate how the complete-count edit of the 100-percent items was different.

B. Major Assumptions

Among the six 100-percent population items on the short form, it was assumed that household relationship, sex, age, and marital status are highly interrelated and hence should be edited jointly. Race and Spanish origin were essentially edited separately, using the edited data for the other four population items.

Both the complete-count edit and the sample edit were subdivided into two separate procedures, one for persons in households and the other for persons in group quarters (both institutional and other group quarters). It was assumed that, on the whole, persons in households and in group quarters have different characteristics. These separate edits allowed different sets of requirements to govern the edit procedures for persons in the two general types of living arrangements.

C. Edit Procedures

Date of Birth and Age. In both the complete-count and the sample edits, there was a year-of-birth, quarter-of-birth, and age pre-edit that was done on all persons in households and group quarters who had a "response" to year of birth. In the general edit that followed the pre-edit, age information may have been blanked or changed to accommodate the household information regarding relationship of members to the householder. The purpose of the age pre-edit, therefore, was to check the FOSDIC information to insure the best age information possible from the date-of-birth item only, i.e., with no reference to other 100-percent items. For the household universe, this pre-edit was done prior to the regular edit of persons in households. For the group quarters universe, the pre-edit was actually done at an early stage of the regular edit of persons in and type of group quarters.

  1. In the pre-edit of year of birth, a check was made to see whether the century digit (8 or 9) was present when the decade and year digits were specified. If there was no century digit, a digit (either 8 or 9) was supplied consistent with the requirement that no person could be born before 1868 or after 1980. Year and quarter of birth were checked to insure that any person born in 1980 was coded as the son or daughter of the householder and was assigned the first quarter.
  2. Ever-married persons born in 1965 were assigned first quarter, and all other persons with quarter of birth not reported were assigned a quarter by rotation among the digits 1, 2, 3, 4.
  3. For persons with acceptable year and quarter of birth, an age recode was calculated at this point.

Persons in Households. For persons in households only, it was required that there be an age difference of at least 15 years between any person reported as a householder, spouse, mother, or father and any other person reported as a son or daughter in the same household. (In the complete-count edit, this was the last step in the age pre-edit.)

In the sample edit of households, there was one final preliminary age edit in which age was allocated when not stated for persons who were reported as "now attending" school (excluding householders, spouse, and parents, and any other ever-married person).

Relationship. In both the complete-count and sample edit procedures, the first phase of the main edit was to check the consistency of responses on household relationship, with reference to responses on age, sex, and marital status. This consistency check was followed by allocation of relationship, sex, marital status, and age (generally in that order) wherever a non-response occurred. In the sample edit, additional procedures were included to process information on detailed relationship (produced by clerical coding from write-in entries for "other relative") and on subfamily number (also produced by clerical coding).

  1. Establish householder: In the consistency check for relationship, the first step was to assign the value of householder to one and only one person in the household. In most cases this was accomplished by making the person in column 1 the householder.
  2. Householder consistency check: If there was at least one person in the household whose relationship to the designated householder was that of father or mother, and if the designated householder was under 19 years old and never-married, then a relationship switch was performed in which one of the stated parents was reassigned to be the householder and the originally designated householder was reassigned to be son or daughter. All other persons in the household were also appropriately reassigned so that they would have a correct relationship to the newly assigned householder. In the sample edit only, if the reported age of the householder was less than 15 years, the age of householder was made non-response, and all entries for detailed relationship and family number were also made non-response.
  3. Spouse processing: The next step was to check for the presence of a spouse of householder. If there was more than one spouse reported, the first spouse in the reporting sequence was retained and all others were made non-response on relationship to householder. In those situations where the householder was the first person reported in the household, and was married but there was no spouse reported, the following procedure applied. If the second person in the household had non-response on relationship, but was of the opposite sex and now married, then the second person was made spouse.
  4. Husband-wife household edit: When a household had been identified as containing a householder-spouse combination, reference was made to the sample item on children ever born for those cases where the spouse had no reported sex or had the same reported sex as the householder. If the spouse had an allowable entry for the children-ever-born item, the spouse was made female and the householder was made male. Otherwise, if the householder had an allowable entry for children ever born, the spouse was made male and the householder female. (This step was not performed in the complete-count edit.)

Failing the above check, if sex was not reported for both husband and wife, or if both reported the same sex, then the householder was assigned male and the spouse female. When the sex of only one was not reported, that person was assigned the sex opposite that for the reported husband or wife. If the householder and spouse were reported with opposite sex, no change was made; thus, the householder in a husband-wife household could be either male or female, depending on how it was reported.

In the next step of the husband-wife edit, given that a householder and spouse had been identified through the relationship item, the marital status of each of these two persons was made now married if not so reported. At that point, if the age of the spouse was less than 15 years old, the age of the spouse was changed to non-response. Thus the relationship of householder and spouse took priority over sex, age, and marital status in the household edit. (At this point in the complete-count edit only, the age of the householder was made non-response if the reported age of the householder was less than 15 years.)

Relationship (excluding householder and spouse). This step (not performed in the complete-count edit) was a procedure to check the consistency of detailed and basic relationship items (excluding-householders and spouses, processed above). In general, if the clerically coded information on detailed relationship and family number were present and the basic relationship was not reported, or if detailed and basic relationships were not consistent, basic relationship was changed to become consistent with detailed relationship. In addition, for any person with detailed relationship "02", "03", or "04" (son/daughter, brother/sister, or father/mother) and clerically-coded family number was greater than 0 (i.e., in a subfamily), a check was made to insure that the person's basic relationship to householder was brother/sister, unless the person had no age reported or was at least 15 years younger than the householder (requiring a basic relationship of son/daughter), or the person was at least 15 years older than householder (requiring a basic relationship of father/mother.)

Inverted Generation Edit. Done only once in any given household, this procedure comprised a child consistency edit and then a parent consistency edit. For each reported child (of householder) with an age greater than or equal to the age of the older of the householder and spouse, either the basic relationship of the child was made non-response (if the child was ever-married or had no response on marital status) or the age of the child was made non-response (if the child was never married). For each reported parent (of householder) the basic relationship was changed to non-response either if the reported age of the parent was less than 30 years or if the parent's age was less than or equal the lesser of the householder's and spouse's ages. Following this step, the marital status of every household person under 15 years old was made never-married. (This marital status check preceded the "inverted generation edit" in the complete-count edit.)

D. Allocation Procedures

Relationship. Up to this point in the edits, missing values were supplied in the consistency checks by specified values or were simply made non-response. In the remaining portion of the data household edits, missing data were supplied either by specified values or from hot-deck matrices.

In the complete-count edit, the next step involved allocating basic relationship from a hot-deck matrix according to

  1. Marital status of person (categories: never-married, ever-married, not reported);
  2. Type of household (husband-wife, other male householder, other female householder, sex of householder not reported); and
  3. Age differences between person and householder (age not reported, person 17 or more years older, 1 to 16 years older, 0 to 16 years younger, 17 to 44 years younger, and 45 or more years younger).

In the sample edit, the allocation of basic and/or detailed relationships depended on which type, if any, was reported. If basic relationship was missing but a detailed relationship was stated (clerically coded as other than 0), then basic relationship was made "other relative." If both types of relationship were missing, both were allocated from a hot-deck matrix according to the same categories stated above for the complete-count edit of basic relationship. In the event that there was a stated basic relationship of son/daughter, brother/sister, or father/mother and the clerically-coded family number was greater than 0, detailed relationship was made 0, to indicate membership in a subfamily; where family number was not stated the detailed relationship was left blank. Finally, if basic relationship was other relative and detailed relationship was blank, the latter was allocated from a hot-deck matrix according to the same categories specified above for basic relationship allocation.

E. Post Allocation Consistency Check

Group Quarters Conversion: If any of the above procedures resulted in a household situation with 9 or more persons not related to the householder, the household was then converted to group quarters and subsequently processed by group quarters edit.

RACE

A. Purpose of Edit Procedures

The edits assigned a race to each person whose race was not indicated in the questionnaire and thus provided a complete racial distribution of the entire population and made the data more useful to data users. The edits were necessary since the data on race are used to implement governmental programs.

B. Major Assumptions

The assumptions underlying the edit and allocation procedures for the race item were:

  • Reported race did not have to be consistent with other items on the questionnaire.
  • The race of a person is likely to be the same as that of another person in the household (in cases of multi-person households).
  • A person's race is more likely to be the same as one's parents or siblings or children than of one's spouse.
  • All of the foregoing relatives are more likely to be of a person's race than are other relatives.
  • A person's race is more likely to be that of relatives than of non-relatives in the same household.
  • Others in the household, whether relatives or not, are more likely to be of the same race as the person in question than are others outside of the household.
  • A person's race is more likely to be identical to others residing nearby (i.e., in the same enumeration district) than others living farther away.
  • The racial composition of residents of group quarters who did not report their race is likely to be similar to that of residents of the same group quarters who did report their race.
  • The racial composition of a group quarters facility where all or most persons did not report their race would be similar to the racial composition of that type of group quarters in 1970.

C. Edit Procedures:

There were no edits or consistency checks on responses to the race item.

D. Allocation Procedures

No population controls were used in hot-deck allocations, which employed other persons in the same enumeration district as a basis of allocating race. There were separate flags for households and for group quarters.

Households:
1. From relative in household
2. From non-relative in household
3. Hot-deck
In group quarters:
1. Cold deck
2. Hot-deck

Sample Edit and Allocation -- Households

If the race entry was missing on the questionnaire for a member of the household, an answer was assigned in the computer according to the race of other household members using specific rules of precedence of relationship. Table A below, shows the sequence of relationship priority used to assign race from other household members. For example, if race was missing for the child of the householder, then race was assigned from the female householder or spouse; however, if there was not such a person in the household, then race was assigned from the householder. If race was not reported for anyone in the household (excluding paid employees), then race was assigned to the householder and everyone else in the household without a race from a hot-deck matrix. The hot-deck matrix contained the race of the householder of previously processing households.

Table A. Relationship Priority for Assignment of Race for Persons In Households: 1980
Relationship of Person with Race Not Reported Race Reported for Another Person in the Household -- Relationship Priority
Householder Mother, father, brother/ sister, son/ daughter, grandmother, grand- father, spouse, grand-children, nephew/niece, uncle/aunt, cousin, son-in-law, daughter-in-law, father-in-law, mother-in-law, other relative, partner/roommate, roomer/boarder, or other non-relative
Spouse Mother-in-law, father-in-law, brother-in-law/ sister-in-law, son/ daughter, or householder
Child of Householder Female householder, female spouse, or male householder
Father-in-law, mother-in-law, brother-in-law, and sister-in-law Spouse of householder
Father, mother, brother or sister Householder
Children in subfamilies Female subfamily primary person, female subfamily spouse, or male subfamily householder
Partner or roommate Partner/roommate or householder
Roomer or boarder Roomer/boarder, other non-relative, or householder
Paid employee Paid employee, other non-relative, or householder
Other non-relative Other non-relative, any other non-relative, or householder

Sample Edit and Allocation -- Group Quarters

In group quarters where 80 percent or more of the persons had a reported race, persons with a not reported race were assigned a race from other persons in the group quarters. Specifically, a hot-deck matrix was constructed within the group quarters and the reported races were assigned in rotation; that is, the reported race sequence was repeated throughout the non-responses. However, in group quarters where all or most of the people did not have a reported race; race was assigned from 1970 census distributions.

(NOTE: The sample edit and allocation procedures for race are similar to those used in the 100-percent processing. For example, the relationship sequence used in 100-percent processing differed because the full relationship detail for "Other relatives" (grandchildren, nephew/niece, uncle/aunt, and so on) and subfamily information was not available.

SPANISH ORIGIN

A. Purpose of Edit Procedures

The Spanish origin item used self- identification. As a result, a response was accepted as true and was not edited based upon responses to other items. However, allocations were made based on associated information when the item was left blank.

B. Major Assumptions

The major assumptions for the Spanish origin edit and allocation procedures were:

  • Reported origin does not have to be consistent with other related items on the questionnaire.
  • Origin can be allocated with confidence from other related questionnaire items.
  • Persons within the same household are of similar origin (Spanish or non-Spanish).
  • Persons of the same race in the same geographic area are of similar origin (Spanish or non-Spanish).

C. Edit and Consistency Check Procedures

As was mentioned in sections A and B above, there were no edits or consistency checks on responses to the Spanish origin question.

D. Allocation Procedures

In the sample record, allocations for non-response could have come from several sources:

  1. Cold-deck allocations which took advantage of the high correlation of Spanish origin with such identifiers as ancestry, country of birth and language;
  2. Response of other household members; and
  3. Hot-deck allocation using race as a population control.

In the complete count record, allocation could have come from responses of other household members or hot-deck allocation.

EDUCATION

A. Purpose of Edit Procedures

The computer edits establish two preliminary consistency checks. First, a consistency edit of education items P8 (School Enrollment), P9 (Highest Grade Attended) and P10 (Finished Highest Grade) compares the data, where appropriate, to age. Inconsistencies, such as between age and grade (a 6-year-old enrolled in college) or among responses to items 8, 9, and 10, were resolved and blanks were replaced with responses by using a hot-deck allocation controlled for age, sex, race, and Spanish origin.

A separate edit was designed to achieve consistency among items P33 (Income in 1979), P9 (Highest Grade Attended) and P10 (Finished Highest Grade) for a short list of selected professional specialty occupations (P30). When an inconsistency occurred (such as, income was moderate or high and education was low), more appropriate levels of education were assigned. For example, for a pharmacist reporting income of $10,000 or more and four years of high school or less, P9 and P10 were recoded to having completed college.

B. Major Assumptions

The computer edits for education make the following two assumptions:

  1. Current age data are more accurate than the responses on education.
  2. For persons 3 to 22 years old, if item P10 is "Finished," age is at least three years greater than the highest grade attended (P9). For those persons in the same age group, if P10 is "Now Attending," or "Did Not Finish," age is at least two years greater than the highest grade attended.

C. Edit Procedures

The consistency edit of education items checks responses to items P8 (School Enrollment), P9 (Highest Grade Attended) and P10 (Finished Highest Grade) for contradictions. For example, if the value for item P8 is "No," for P9 is "Elementary 6," and P10 is "Now Attending," items P8 and P10 are inconsistent. For persons 3 to 22 years old, both enrolled in school and not enrolled, the age/grade edit based on assumption 2 assigns item P9 from a hot-deck matrix if age/grade criteria are not met. In the age and grade check, age is recoded to age as of October 1, when modal grade is more clear for each age. The procedure is simple since quarter of birth is recorded. For persons 23 years old and over, any grade is accepted. For persons 7 years old and older, nursery school or kindergarten are not acceptable responses to question P9. The age/grade edit allows children 2 years old or less to be enrolled in or to have completed nursery school; thus, if item P9 is greater than nursery school, it is changed to "blank." There are no allocations for children 2 years old or under.

After consistency edits among the three education items are completed, all remaining blanks in questions P8, P9, and P10 are allocated from hot-deck matrices using the control variables, single years of age, sex, race, and Spanish origin.

For persons 16 to 24 years old living in college dormitories, consistency edits are the same as for the remainder of the population; however, allocations are from hot-deck matrices containing only data from other persons living in college dormitories, since persons living in such dormitories are more likely than other persons in the population to be attending college.

PLACE OF BIRTH

A. Purpose of Edit Procedures

The edits of question 11 are designed to provide data that are consistent with each person's citizenship and to allocate missing data.

B. Major Assumptions

We assume that foreign-born family members (excluding those related by marriage) were most likely born in the same country. Likewise, we assume that a child of the reference person whose state of birth is not reported was most likely born in the same state as the next oldest child with state of birth reported. Finally, we assume children under 5 with no state of birth reported were most likely born in the state of current residence.

C. Edit Procedures

If question P11 is blank, the edit tries to assign the state of birth of another family member based on the assumptions in section B above. Assignments are only made if consistent with the person's responses to question P12 on citizenship.

D. Allocation Procedures

Person's whose place of birth cannot be assigned through the edits are provided with a state of birth through a hot-deck allocation. Specific country of birth is not allocated; instead, a code for "Abroad, not specified" is assigned. The population controls used in the hot-deck matrix are age, race, and Spanish origin. All cells in the matrix are initially loaded with state of current residence.

CITIZENSHIP and YEAR OF IMMIGRATION

A. Purpose of Edit Procedures

Editing of the citizenship item was designed to ensure that the item was answered only for foreign-born persons, as intended. Year of immigration was checked for consistency with age to be sure that the person was old enough to have immigrated during the reported interval of immigration.

B. Major Assumptions

The edit procedures assumed that the place of birth reported was more accurate than the report of citizenship status for persons who were born in the United States or in an outlying area and for whom, therefore, reporting citizenship status was improper. Also, the allocation procedures assumed that one could make reasonably accurate assignments of citizenship for persons from the citizenship status of parents in the household. Finally, it was assumed that date of birth is more accurately reported than year of immigration, since almost everyone knows his or her date of birth.

C. Edit Procedures

In the edits, entries were examined to ensure that only foreign-born persons had a response in the citizenship question and that there vas consistency between a person's age and the reported year of immigration.

D. Allocation Procedures

Allocation for blank items was determined for some persons based on the place of birth or citizenship of parents present in the household. Hot-deck allocations for blank items were determined using information on country of birth and year of immigration in the case of citizenship status, and using information on race and age in the case of year of immigration. Responses to citizenship and to year of immigration were each stored in matrices for the most recently reported case of foreign-born persons.

LANGUAGE SPOKEN AT HOME and ENGLISH ABILITY

A. Purpose of Edit Procedures

The computer edits of question P13 are designed to ensure that for persons 3 years old and over, the language items are filled in a logical and consistent manner; that is, for persons who reported speaking a language other than English (P13a), a specific non-English language (P13b) as well as English ability (P13c) were to be reported also.

B. Major Assumptions

The computer edits for language usage make the assumption that responses to question P13b are more accurate than those to question P13a or P13c. Thus, if a person reported a specific non-English language in P13b but reported that he/she only spoke English in P13a, then P13a was changed to "Yes."

C. Edit Procedures

The pre-edits to the language question check that the language code (question P13b) that was filled in by census coders based upon the respondent write-in entry is a valid language code. If the three digit code is found to be incomplete or one that is not used, then the entire code is blanked for allocation.

The consistency edits make sure that if there is a "Yes" response in question P13a there is a language code filled for P13b and an ability filled for P13c. If P13b was filled it was considered to be the most reliable part of the three questions; thus P13a was made "Yes" and P13c, if it had not been filled, was allocated.

Frequently, on the other hand, for persons who only spoke English (question P13a), English ability (P13c) was also filled. If persons reported only speaking English at home and there was no non-English language in P13b, then the computer edits blanked the response in item P13c; if item P13b was filled, P13c was accepted. If however, item P13b was filled and P13c was blank, item P13c was allocated from a hot-deck matrix.

D. Allocation Procedures

Allocation of the three part question was accomplished in the following two steps:

  1. When there was no response in P13b, other close family members' responses were used in a type of consistency edit to fill P13b with a specific language.
  2. If there was no close relative with a non-English language reported, then cold deck allocation matrices were used to assign the missing values. The population controls used in the allocation matrix for P13a/P13b are race, Spanish origin, ancestry and country of birth. The control variables used in the allocation matrix for P13c (ability), once a specific language had been assigned, are country of birth, year of immigration for the foreign born, language in P13b, age and years of school completed. When persons reported that they were born in a region of the world such as Europe or Western Europe, rather than in a specific country, the code for unspecified language was assigned.

ANCESTRY

A. Purpose of Edit Procedures

The ancestry item was one of self-identification, and the reported response was accepted. If the question was not answered, the non response was also accepted, with no attempt being made to allocate for non-reporting. The rationale for not allocating this item when blank was that the absence of a reported ancestry category in itself constituted important information, and there was interest in the extent of non-reporting. In addition, the Bureau, with advice from ethnic advisors, concluded that allocation for non-response could possibly bias the ancestry data.

B. Major Assumptions

The major assumptions for the ancestry edit procedures were:

  • Reported ancestry does not necessarily have to be consistent with other related items in the questionnaire.
  • Certain ancestries that are described in two or more words are truly single ancestries.
  • "No response" is a legitimate entry indicating lack of understanding or true uncertainty of one's ancestry.

C. Edit Procedures

The coding procedures provided for the identification of every single- and double-origin category reported, as well as selected triple-origin categories. The edits were primarily intended to correct for coder error. For example, ancestries that consisted of two or more words, but that nonetheless identified a single ancestry, were tabulated as single-ancestry groups. (Examples are "Spanish American" and "British West Indian.") Also, the pre-edit operation identified any nonexistent code that may have been incorrectly assigned to a reported ancestry or to a non-response.

D. Allocation Procedures

There was no allocation for non-response; therefore, there were no allocation flags.

RESIDENCE 5 YEARS AGO (Migration)

A. Purpose of Edit Procedures

The edits of question P15 are designed to provide data that are consistent with each person's age and to allocate missing data.

B. Major Assumptions

We assume that family members most often move together; that children most often move with their mother if both parents do not move together. Further, we assume that most moves are short-distance moves; that is, within the same county. In addition, we assume that EDs (enumeration districts) are processed in state order.

C. Edit Procedures

The pre-edit computes the check digit of the 6-digit code, blanks all "bad" codes (out of range or check digit fails to compute), and does no further processing of the questionnaires not in the migration sample.

D. Consistency Edit

The consistency edit of question P15a changes the person's response to be consistent with the age reported in question P5. Family members 5 years old or over with P15a blank are assigned "same house" or "different house" based on the responses of another family member using the assumptions in B above.

The consistency edit of question P15b has several steps:

Step 1:
Converts the 6-digit code to the full geographic codes, including state, county, MCD and place codes.
Step 2:
Makes the response to question P15b(4), "Inside the city limits," consistent with the presence of a place code the data reported by that other family member.
Steps 3, 4 and 5:
Assigns the codes from another family member for persons with blanks for state, county or MCD in the Northeast, providing any partial data given matches.

E. Allocation Procedures

  1. If P15a is still blank following assignment of codes by the consistency edit, a response is allocated using a hot-deck matrix. The population controls used in the hot-deck matrix are age, race, years of school completed, Armed Forces status, current residence (in a central city, balance of an SMSA, or outside an SMSA), and region of birth (including same state and born abroad). All cells of the matrix are initially loaded with "same house" for those currently living outside SMSA's OR ages over 30; all other cells are initially loaded "different house."
  2. If state, county, or MCD codes are still blank following the consistency edit of question P15b, missing data are allocated using a series of hot-deck matrixes.

    1. If the state code is blank, the population controls for the allocation matrix are division of birth, race, age, years of school completed, Armed Forces status, and current residence (in a central city, in the balance of an SMSA, or not in an SMSA). All cells are initially loaded with current residence.
    2. If the state code is for a state in the United States or D.C., a second matrix is used that is identical to the first matrix except that responses of foreign countries (98 in state code) cannot be used to fill the matrix. Initial values are the same as for the first matrix.
    3. If the state code (within U.S.) is given but the county code is blank, the third matrix is used. The population controls for this matrix are state of previous residence, race, age, years of school completed, Armed Forces status, and current residence. Initial values are provided.
    4. If state and county are given but MCD is blank (one of the nine Northeastern States), the fourth matrix is used. The population controls for this matrix are state and county of previous residence, race, years of school completed, Armed Forces status, and current residence. Initial values are provided.
    5. Finally, persons reporting New York City without specifying the county or borough are assigned to the county of the last reported case of New York city with county specified.

ACTIVITY 5 YEARS AGO

A. Purpose of Edit Procedures

The edits of question P17 are designed to provide data that are consistent with each person's age and to allocate missing data.

B. Major Assumptions

We are assuming that the current age data is more accurate than responses on activity at a date 5 years in the past.

C. Edit Procedures

  1. The consistency check changes any "Yes" in question P17 to "No" if the person was less than 16 years in 1975.
  2. Question P17a is changed, if necessary, to be consistent with the response to question P18a on veterans status.
  3. Question P17b is changed, if necessary, to be consistent with the response to question P9 on highest grade of regular school ever attended.
  4. Question P17c is changed, if necessary, to be consistent with the response to question P27 on the year the person last worked.

E. Allocation Procedures

If the consistency check has not changed a "blank" to "No," the response is allocated using a hot-deck matrix. The population controls used in each hot-deck matrix are as follows:

17a: age, race, sex
17b: age, race, sex, P17a
17c: age, race, sex. P17a, P17b

All cells in the matrix are initially loaded with "No."

VETERAN STATUS

A. Purpose of Edit Procedures

To classify all persons 16 years and over according to whether they have ever been on active-duty military service (except for training in the National Guard or military reserves) in the Armed Forces of the United States. The edit attempts to corroborate the response in question P18a or, failing that, to assign a value to P18a based on other information from the respondent or on a hot-deck allocation procedure. The edit establishes a value for each person that is consistent with the official Veterans Administration definition of a "veteran" and with other the person's other characteristics.

B. Major Assumptions

  • The values for the Employment Status Recode, Group Quarters Type. Industry, Occupation, and Activity Five Years Ago items take precedence over the response to question P18a, which is edited after final values have been assigned to these items.

    The response to question Pl8b (Period of Service in the Armed Forces), is edited before, and without reference to, the response to l8a. An edited value in 18b that shows that the person served on active duty takes precedence over the person's response to question P18a.A person must be at least 16 years old to be considered a veteran.

C. Edit Procedures

The edit examines some closely-related items to see whether there is any evidence from them that the person is now serving or ever has served on active duty. If the person's employment status is "Armed Forces;" or if the person resides in military barracks; or if his or her current or most recent industry and occupation is "Armed Forces;" or if the person is over 21 years old and he/she indicated in question P17 that, in April 1975, he or she was on active duty; or finally, if the edited value in question Pl8b indicates that the person was on "active duty at any time," then responses other than "Yes" in P18a are assigned a value of "Yes", and "Yes" responses are accepted as valid. If none of the above conditions for the closely-related items is present, then the edit accepts the person's response in P18a as the final value, or if Pl8a is blank, the edit assigns a final value from a hot-deck matrix.

D. Allocation Procedures

Population Controls in the hot-deck matrix are age, sex, and race (White/Black/Other Races).

PERIOD OF SERVICE IN THE ARMED FORCES

A. Purpose of Edit Procedures

The purpose is to identify for each veteran of active duty military service in the Armed Forces of the United States, the period or periods in which he and or she served. The edit is designed to ensure that there is a logical and reasonable relationship between a person's age and the period (s) he or she served in the Armed Forces. This requirement takes two forms: (1) that the person is old enough to have been of military age during any claimed period; and (2) that the person is not so old that it is unreasonable that he or she began military service during the earliest period claimed. The edit also aims to ensure that persons reporting more than one period do not have improbable combinations of periods.

B. Major Assumptions

  1. Persons must be at least 16 years old to be a veteran.
  2. To have served during a period, except the category "Any other time," the person must have been at least 16 years old during the time covered by the period. To have a response of "Any other time" accepted as valid, a person must be at least 45 years old (at the time of the census).
  3. Persons who earliest claimed period is "May 1975 or later" must be 50 years old or younger; persons whose earliest period is "Vietnam-era (1964 to April 1975)" must be 60 years old or younger.
  4. It is unreasonable for a person to claim service in the Vietnam Era and the World War II period and not claim service in the Korean Conflict period; and unreasonable for a person to claim Vietnam and World War I, without claiming either World War II or Korean Conflict. It is also unreasonable for a person to claim World War I and Korean Conflict without claiming World War II.
  5. Unreasonable age/period-of-service relationships or unreasonable period combinations are incontrovertible evidence of respondent error.

C. Edit Procedures

Step 1.
Checks the person's responses (filled circle for each period served) to see whether they meet the test of validity implied by assumption 2 above; any response not meeting the test is blanked.
Step 2.
Checks responses for the assumption 3 tests; any failure leads the edit to blank all the person's responses to question Pl8b.
Step 3.
Checks responses of "Vietnam Era" and "World War I" for the tests of assumption 4; once more, any failure results in the blanking of all the person's responses to question Pl8b.
Step 4.
There are seven "Periods" categories in question P18b. The final step of the edit assigns one of three final values to each of these categories: served; did not serve; or not in universe. If the person is determined by the Veteran Status edit to be a veteran (i.e., final value of "Yes" in question P18a), then any period of service category with a valid response is given the value "served" and all others are given the value "did not serve". If the veteran has no valid responses, a value of "served" or "did not serve" is assigned to each category in a joint hot-deck allocation procedure. If the person is a non-veteran (P18a = "No"), all the response categories are given the value "Not in Universe."

D. Allocation Procedures

Values for each of the seven response categories of question P18b are assigned jointly from an individual donor to any person who emerges from the veteran status and period-of-service edits with a final value of "Yes" for question P18a and blanks in all the categories of P18b. The population control for the period-of-service allocation matrix is age.

DISABILITY

A. Purpose of Edit Procedures

The purpose and gains from these procedures are:

  1. To ensure that data are present for each person in the universe.
  2. To make the responses between "work limitation" and "work prevention" consistent (persons who are work prevented must be work limited).
  3. To make the responses between "work prevented" and labor force activity consistent (persons in the labor force cannot be prevented from working).

B. Major Assumptions

Disability status is associated with the variables used in the hot-deck matrix.

C. Edit Procedures

Change "work limited" if inconsistent with "work prevented". Change "work prevented" if inconsistent with labor force activity status.

D. Allocation Procedures

Population controls used in hot-deck matrix are household relationship, age, race, highest grade completed, and employment status recode.

CHILDREN EVER BORN

A. Purpose of Edit Procedures

The question on the number of children ever born was asked of all women 15 years old and over, regardless of marital status. The respondent was asked to include all children, even those born before her present marriage, children no longer living, and children living away from home, as well as children who were still living in the home. In the 1980 census, a terminal category of "12 or more" was used for reporting the number of children ever born. For purposes of computing the total number of children ever born, this terminal category was given a mean value of 13.

B. Major Assumptions

In instances where no answer was given to this question and where an answer was recorded but appeared to be inconsistent with other demographic characteristics, the number of children ever born was imputed. Since single (never married) women were among those asked about their childbearing to date, there is the likelihood of some deliberate misreporting of the facts, especially among women who perceive out-of-wedlock childbearing as a social stigma. It is also quite possible that the level of misreporting or not reporting at all to the question on children ever born may differ systematically according to other demographic and social characteristics such as age, race, and household relationship. When the number of children ever born was imputed, the imputed value was obtained by matching these women with women of similar characteristics who did report on children ever born.

C. Edit and Allocation Procedures

All responses were first checked against the respondent's age. Any woman whose response on children ever born resulted in an average of more than one birth a year since age 14, e.g., a woman 16 years old who reports 4 children ever born, was imputed a new value. Two allocation matrices were used in the computer edit specification, one for women living in group quarters and one for women living in households.

For women living in group quarters who needed a value imputed on children ever born, the number imputed from the allocation matrix was the number of children ever born. The characteristics used in the matrix were the race of the woman, her current age, and her marital status (never or ever married). An additional characteristic was used for ever-married women: the period of her first marriage if she was under 40 years of age, and her age at first marriage if she was 40 years of age or older.

For women living in households who gave an acceptable response on the number of children ever born, a difference between the reported number of children ever born and the computed number of own children present in the household was obtained. If this difference (representing the number of children absent from the household) was either positive or zero, the difference was stored in the appropriate cell of the allocation matrix, replacing the previously held value. If the difference was negative, indicating the woman had more own children in the household than she reported to have borne, her reported value on children ever born was accepted but no difference entered the allocation matrix (the previously held cell value was kept). This negative difference could occur in instances where there are adopted children in the household.

For women living in households who needed an imputed value of children ever born, the previously mentioned difference from the allocation matrix was added to the number of own-children the woman had in the household to obtain an assigned total for children ever born. If the resulting total exceeded "12" children, the woman was assigned to the "12 or more" category. If the resulting total exceeded the number she would be allowed to have based on her current age, she would be given the maximum allowable for her age. This latter procedural check was also used for women living in group quarters. No "differencing" scheme was used for women living in group quarters since, by definition, they would represent a household group.

The population controls used were age, race, marital status, and if ever-married, the woman's age or year of first marriage.

EMPLOYMENT STATUS RECODE

A. Purpose of Edit Procedures

The purpose of this procedure is to classify all persons 16 years old and over into one of the following six labor force categories (known as "employment status recode" or ESR categories), as officially defined by the Departments of Labor and Commerce:

ESR = 1 Employed, At Work (AW)
ESR = 2 Employed, With a Job But Not at Work (W/J)
ESR = 3 Unemployed (UN)
ESR = 4 Armed Forces, At Work (AF/AW)
ESR = 5 Armed Forces, With a Job But Not at Work (AF/WJ)
ESR = 6 Not in Labor Force (NLF)

These categories all relate to the person's activity during the census reference week. They enable all persons 16 years old and over to be placed into a unique official U.S. Government employment-status classification as follows:

Labor Force: Consists of ESR categories (1) (2) (3) (4) (5)

  • Armed Forces: (4) (5)
  • Civilian Labor Force:(1) (2) (3)

    • Employed: (1) (2)

      • At Work: (1)
      • With a Job But Not at Work: (2)
    • Unemployed: (3)

Not In Labor Force: (6)

B. Major Assumptions

In general, each person is placed into one of the ESR categories according to his or her combination of responses to the following questions: P22a, P22b, P23 (or P24b), P25, P26a, P26b, P27 and P28.

  1. Persons who answer "Yes" to P22a or who leave P22a blank but report hours worked in P22b are generally assumed to be "at work" and given ESR 1 regardless of their other responses.

    Persons answering that they did not work last week are generally assumed to be:

    • "with a job but at work" (ESR 2) if they indicate that they were on vacation or ill or otherwise temporarily absent from a job;
    • "unemployed" (ESR 3) if they indicate that they were "on layoff" or that they looked for work in the last four weeks and were available to take a job last week;
    • "Not in labor force" (ESR 6) if they were not temporarily absent from a job and not looking or not available for work.
  2. Persons who reside in military barracks or indicate in P28 that they were on active duty in the Armed Forces are generally assumed to be:

    • "Armed Forces at work" (ESR 4) if P22a is marked "Yes"; or
    • "Armed Forces, not at work" (EM - 5) if P22a is "No" or not reported.
  3. Persons with all other combinations of responses are generally given ESR 6, or are allocated an ESR.

C. Edit Procedures

Initially, the edit examines a person's inmate status and group quarter status. Inmates of institutions or persons who report that they resided in military barracks are immediately classified as "Not in the Labor Force" or an "Armed-Forces at Work," respectively.

Then the edit examines the industry item (P28). Persons identified in the industry item as being on active duty in the Armed Forces are given ESR categories of "AF/AW, " or "AF,W/Job;" those identified as homemakers or volunteer workers are classified as "Not in the Labor Force" or "Unemployed." The specific ESR category assigned these persons depends on their responses to the employment-status items (P22a, P22b, P25, P26a, P26b, and P27).

Next the edit examines the class of worker item (P30, whose possible categories are: private wage and salary workers, government workers, self-employed, and unpaid family workers). Class of worker is examined because the official U.S. employment status concepts treat unpaid family workers differently from other kinds of workers. Generally, the employment status assigned to unpaid family workers depends on the number of hours they worked during the census reference week. Therefore, those who worked less than 15 hours a week are generally assigned an ESR of "Not in the Labor Force," whereas those who worked 15 hours or more are usually assigned "At Work."

Persons not yet assigned an ESR are subjected to a series of tests that concentrate on their responses to the employment status items (P22a, P22b, P25, P26a, P26b, and P27) and the place of work item (P23) or, if P23 is blank, and P24b as follows:

  1. To be classified as "at work (AQ)," a person has to meet one of the following three conditions:

    1. A "Yes" response to question P22a;
    2. An entry of "Yes, on vacation" in item P25; or
    3. An entry in item P22b (hours worked) if item P22a is blank.

    For cases with inconsistent entries, the items on date last worked (question P27) and place of work (questions P23 or P24b) are considered in the classification.

  2. To be classified as "with a job, not at work (W/J)," a person could not have indicated that he was at work and had to have an entry of "Yes, on vacation .. " in item P25.
  3. To be classified as "unemployed (UN)" a person could not have indicated that he was at work or was with a job but not at work, and he has to meet one of the following conditions:

    1. An entry of "Yes, on layoff" in item P25; or
    2. A "Yes" response to looking for work in item P26a and an indication in item P26b that the person was available to accept a job. Availability to accept a job is assumed if there was an entry of "No, already has a job," "No, temporarily ill," or "Yes, could have taken a job" in P26b.
  4. To be classified as "Not in the labor force," a person could not have indicated that he was at work, that he was with a job but not at work, or that he was looking for work. In addition, two of the three items P22a, P25, and P26 must have an entry of "No" or an implied negative. For cases with inconsistent entries, such as implied negatives, the items on date last worked (question P27) and place of work (P23 or P24b) are considered in the final classification.
  5. Finally, when the employment information is so incomplete or so contradictory that the edit cannot assign a valid ESR, the edit assigns a temporary recode of "Needs Allocation." A person with this code is assigned from a hot-dock allocation matrix the same employment status as a person having similar demographic characteristics.

D. Allocation Procedures

The population controls used in the hot-deck matrix are household relationship sex, age, school enrollment, race, and farm/non-farm residence. These were chosen because of their high correlation with employment status.

HOURS WORKED

A. Purpose of Edit Procedures

Data on actual hours worked last week apply to persons 16 years old and over who were at work during the census reference week. This edit establishes this universe of persons for these data. The edit allocates a value for persons within this universe who do not answer P22b.

B. Major Assumptions - None

C. Edit Procedures

Persons whose employment status recodes (ESR) are "With a job, but not at work (ESR = 2)," or "Unemployed (ESR = 3)," or "Armed Forces, not at work (ESR = 5)," or "Not in Labor Force (ESR = 6)" are assigned the final value "Not in Universe," which means that actual hours data cannot appear in any tabulations for these persons. For the remaining persons 16 years old and over [that is, persons with ESR of "at work (ESR = 1)" or "Armed Forces, at work (ESR = 4),"] the edit accepts a response in 22b as the final actual hours worked value or, in the absence of a response, allocated from a hot-deck matrix a final value from a person with similar demographic characteristics.

D. Allocation Procedures The population controls of the hot-deck allocation matrix are the same as those used for the ESR edit.

PLACE OF WORK

A. Purpose of Edit Procedures

The edits of question P23 are designed to provide data that are consistent with each person's employment status and internally consistent among the parts of the question. Place of work was coded with a 6-digit serial number called a GRIN Code, which is later converted to the actual census geographic codes that it represents. The sixth digit of the code is a check digit. If a person's place of work is coded to the block level, a 3-digit block code is added at the end of the 6-digit code, resulting in a total of 9 digits. The GRIN Code is converted to intermediate geographic codes of further editing, then the intermediate codes are changed to output codes for the Sample Detail File.

B. Major Assumptions

We assume that the employment status recode is correct.

C. Edit Procedures

The pre-edit computes the check digit of the GRIN Code, blanks "bad" codes (check digit fails to compute), and blanks block numbers if FOSDIC did not read all three digits. Finally, the check digit is dropped leaving a 5-digit code plus a 3-digit block code if applicable.

The consistency edit of question P23 has several steps:

Step 1.
If the person is not in the place-of-work/migration sample, or the person is in the place-of- work/migration sample but their employment status recode is not "at work" or "Armed Forces-at work," all place-of-work geographic codes are made "Not in Universe."
Step 2:

(a) If the person worked at home, the geographic codes for the place of residence are obtained and matched against the GRIN Code conversion file. The codes that match then supersede the GRIN Code and become the person's intermediate place-of-work codes for further editing.

(b) If the person did not work at home:

  1. If the person's place-of-work GRIN Code is the code for Puerto Rico; a foreign country; an outlying area; abroad; at sea; U.S., state not specified; or not reported, the code is changed to the appropriate intermediate codes without using the conversion file.
  2. If the person's place-of-work GRIN Code is the code for a state, county, minor civil division (Northeast only), place, or census tract in the United States, the code is matched against the conversion file and all the geographic codes which it represents are defined as intermediate codes.
Step 3.
If the GRIN Code converts to a census tract and a block code is also given, the block is checked to be sure it is a legitimate block number in that tract. If not, the block number is blanked.
Step 4.
If a GRIN Code converts to a census tract, the various locations indicator (filled by the coder if the respondent did not have a fixed place of work) is blanked.
Step 5.
The incorporated limits indicator is changed, if necessary, to make it consistent with the place code and the place description.
Step 6.
The intermediate codes, as edited, are made the final output codes to be carried on the Sample Detail File.
Step 7.
Extensive edit tallies are made for internal use.

TRAVEL TIME TO WORK

A. Purpose of Edit Procedures

The edits of question P24a are designed to provide data that are consistent with each persons employment status and to allocate missing data.

B. Major Assumptions

We assume that the employment status recode is correct and that persons of the same sex, living in the same area, and using the same means of transportation to work, spend about the same amount of time commuting to work.

C. Edit Procedures

The pre-edit blanks question P24a if only one of the two digits is read by FOSDIC. In addition, the consistency edit makes question P24a "Not in Universe" if the means of transportation reported in question P24b is "Worked at home."

D. Allocation Procedures

If question P24a is blank, a number of minutes is provided through a hot-deck allocation. The population controls used in the hot-deck matrix are sex, SMSA/non SMSA residence, and means of transportation. All cells in the matrix are initially loaded with 15 minutes.

MEANS OF TRANSPORTATION TO WORK

A. Purpose of Edit Procedures

The edits of question P24b are designed to provide data that are consistent with each person's employment status and to allocate missing data.

B. Major Assumptions

We assume that the employment status recode is correct and that persons living in the same area most likely use the same means of transportation to get to work.

C. Allocation Procedures

If question P24b is blank, the means of transportation is provided through a hot-deck allocation. The population controls used in the hot-deck matrices are SMSA/non SMSA residence, carpooling status (if reported), number of persons in carpool (if reported), and sex and race (if carpooling status and number of persons in carpool are not reported). All cells in each matrix are initially loaded with "Car" as means of transportation.

CARPOOLING

A. Purpose of Edit Procedures

The edits of question P24c are designed to provide data that are consistent with each person's employment status and means of transportation and to allocate missing data.

B. Major Assumptions

We assume that the employment status recode is correct and that persons living in the same area and using the same means of transportation are equally likely to carpool.

C. Allocation Procedures

If question P24c is blank, carpooling status is provided through a hot-deck allocation. The population controls used in the hot-deck matrices are SMSA/ non SMSA residence, means of transportation (if car, truck or van is reported in question P24b) and number of persons in the carpool (if reported in P24d). Additional population controls of race and sex are used if both questions P24b and P24d are also blank. The cells in the matrices are initially loaded with "Drives alone" if question P24d is blank and "Shares driving" if the number of persons reported in question P24d is 2 through 74. If means of transportation is not reported, the matrices are initially loaded with "Car" as means of transportation, "Drives alone" if question P4d is blank and "Shares driving" if question P24d is not blank.

CARPOOL OCCUPANCY

A. Purpose of Edit Procedures

The edits of question P24d are designed to provide data that are consistent with each person's employment status, means of transportation, and carpooling status, and to allocate missing data.

B. Major Assumptions

We assume that the employment status recode is correct, that persons living in the same area are equally likely to carpool, and that the number of persons in each carpool for those not reporting is similar to the number of those reporting.

C. Allocation Procedures

If question P24d is blank and question P24c indicates the person carpools, the number of persons in the carpool is provided through a hot-deck allocation. The population controls used in the hot-deck matrices are SMSA/ non-SMSA residence, means of transportation (car, truck, or van), and carpooling status. The cells in the matrices are initially loaded with 2 persons in the carpool. If carpooling status is blank, the matrix is initially loaded with "Drives alone" and carpool occupancy is "Not in Universe."

YEAR LAST WORKED

A. Purpose of Edit Procedures

This edit establishes consistency between employment status recode (ESR) and year last worked (question P27) and between worked in 1979 (P31a) and year last worked (P27). It provides as many persons as possible with final values for question P27; and indicates for the reminder that the final value is to be allocated from a hot-deck matrix.

B. Major Assumptions

The edit assumes that when there are inconsistencies between the person's response to question P27 and his or her ESR or final value for question P31a, the ESR or question P31a values are more accurate indicators of year last worked than the person's question P27 entry.

C. Edit Procedures

Persons whose employment status recode (ESR) is 1, 4, or 5 or who enter "1980" in question P27 are assigned a final value of "1980" for question P27.

For persons not assigned a final value by the above instruction, the edit accepts entries in question P27 as final values, except:

  1. If the entry in question P27 is 1979 and the final value in question P31 states that the person did not work in 1979 (P31a = "No"), the edit assigns "1980" as the final question P27 value.
  2. If the entry in question P27 is not 1980 or 1979 and the final value in question P31a states that the person did work in 1979 (P31a = "Yes"), the edit assigns "1979" as the final question P27 value.
  3. If the person's response in question P27 indicates that he last worked when he was 14 years old or younger (for example, a person 16 years old who reports that he last worked in "1969 or earlier"), the edit assigns the final value of "never worked" to question P27.

The remaining persons who did not respond to question P27 are assigned the intermediate value "needs allocation." These persons will be assigned a final value from a hot-deck matrix.

D. Allocation Procedures

Allocated values are assigned a value from one of the matrices in the joint-economic item allocation system.

INDUSTRY, OCCUPATION and CLASS OF WORKER

A. Purpose of Edit Procedures

The industry, occupation, and class of worker items are edited as a package because they all refer to one job. The class of worker item was self-coded at the time of enumeration, but the industry and occupation (I & 0) codes were assigned by coders in the clerical processing operation. The I & 0 coders assigned either a three-digit numeric code or a one-letter code for the industry and occupation entries. Because of all these possible sources of response and clerical errors, the edit should make all three items consistent wherever possible to each other and certain other items. The edit also identifies persons who must be assigned values for questions P28 through P30 from a hot-deck matrix.

B. Major Assumptions

There are several major assumptions:

  1. All the I & 0 entries must ultimately have valid 3-digit codes.
  2. The industry, occupation and class of worker entries must be consistent with each other since they refer to the same job.
  3. Sometimes occupation should also be consistent with education and earnings.

C. Edit Procedures

The first edit step is to convert the letter codes to three-digit codes and to make sure all the codes have three digits. The codes were also checked to make sure they were valid (only 732 of the 19,000 possible code combinations were legal). Most invalid codes were blanked but some were changed to valid codes. This was done only when one digit was off by one number. For example, 131 was changed to 130.

As stated above, the industry, occupation, and class of worker items referred to one job. Therefore, the codes for each person must be consistent. For example, a "mail carrier, postal service" must have an industry of "postal service" and a class of worker of "Federal government." Some of the larger blocks of occupations that required an edit to insure consistency with occupations were teachers, sales occupations, private household occupations, protective service occupations and farming, forestry and fishing occupations. There are other cases where most people in an occupation are in one industry. If the reported occupation is one of these occupations, and the industry is not reported, the code for the predominant industry is added. Some examples are dental hygienists who are mostly in offices of dentists; actors and directors who are mostly in theaters and motion pictures; and barbers, who are mostly in barber shops. Other occupations are concentrated in a few industries. One of the several industries containing the occupation was chosen to be assigned when the occupation was reported but industry was not.

Since most industries have many occupations, far fewer occupations can be assumed from the industry when occupation is blank. However, some assumptions can be made with the help of the class of worker entry. For example, most self-employed people working for law firms are lawyers. These obvious assignments were made in the edit.

Most of the edits dealt with the relationship between industry and occupation themselves. However, for a few occupations we looked at education. The educational requirements were not stringent in order to allow for people who may have advanced to occupations through experience. For example, in the edits persons in engineering and natural science occupations are required to have completed at least 12 years of school. If they do not meet this criterion, their earnings are checked. If earnings are $15,000 or more, the occupation code in kept and the education level is raised. If the earnings are below $15,000, the occupation is changed to one of the technician codes. The educational requirement is higher for some of the health practitioners and lawyers and lower for accountants and registered nurses.

D. Allocation Procedures

The blanks or non-responses remaining after the edit were allocated to one of the categories, in the classification system through a hot-deck allocation system. Where both industry and occupation were missing, the main characteristics used to allocate are sex, race, age, years of school completed and, if reported, work experience such as usual hours worked and weeks worked.

WORKED IN 1979

A. Purpose of Edit Procedures

This edit tests the validity of a person's responses to question P31a by examining related responses on the questionnaire. Responses corroborated by related information are accepted as final values. When a response cannot be corroborated or when question P31a is not reported, information from related items is used to infer a final value. When such inferences cannot be made, the edit declares that the final value is to be allocated in a hot-deck matrix from another person with similar demographic characteristics.

B. Major Assumptions

In general, we assume the following:

  1. A "Yes" response in question P31a is sufficient evidence that the person worked in 1979; we will accept it as the final value for P31a regardless of any inconsistencies between it and the responses in the earnings items (P31a, P32b, P32c) or the year last worked item (P27).
  2. If a person reports that he had earnings in 1979 (either by listing a dollar amount in questions P32a, P32b, P32c, or by answering "Yes" in P32a, P32b, P32c without listing a dollar amount), then we will, in most instances, assume that the person worked in 1979.
  3. If a person reports that he had no earnings in 1979, we will assume that the person did not work in 1979.
  4. Each of the following is considered evidence that the person worked in 1979:

    1. responses in P31b or P31c;
    2. an ESR of 1, 4, or 5; or
    3. a response of "1980" or "1979" in year last worked (P27).

    Conversely, each of the following is considered evidence that the person did not work in 1979:

    1. non-responses in both P31b and P31c;
    2. an ESR of 3 or 6; or
    3. an entry in the year last worked item (P27) other than 1980 or 1979.

C. Edit Procedures

For persons 18 years and over, the edit converts responses of "Yes" or "No" or non-responses in question P31a to final values of "Yes," "No," or to an intermediary value indicating that the final value is to be allocated from a hot-deck matrix, according to the following general rules:

  1. If there is a "Yes" response in question P31a, the edit makes the final value for P31a "Yes";
  2. If there is any indication that the person had earnings in 1979 (any dollar amount reported in question P32a, P32b, or P32c, or any of the "Yes" circles in P32a, P32b, or P32c are marked), the edit accepts a "Yes" in P31a as final, or changes "No" entries or non-responses in P31a to final values of "Yes."
  3. If the respondent indicates that he or she had no earnings in 1979 (by answering "No" in question P32a, P32b, and P32c, or by not responding to P32a, P32b, and P32c but answering "None" in question P33), then the edit accepts a "No" response in P31a as final or changes a "Yes" response or a non-response to "No."
  4. If the respondent provides no information about his or her earnings in 1979, then the edit examines the person's employment status recode, responses in questions P31b and P31c, and year last worked (P27):

    1. If the person's ESR is 1, 4, or 5, or if his entry in year last worked is 1980, then

      • If there are responses in questions P31b and P31c, the edit makes the final value in P31a "Yes";
      • If both questions P31b and P31c are left blank, the edit assigns an intermediate value (symbolized as "NA") to P31a, indicating that the final value is to be assigned from a hot-deck matrix;
      • If one but not both of questions P31b or P31c is blank, the edit changes a "No" response in P31a to the intermediate value "NA" and assigns to non-responses in P31a the final value of "Yes."
    2. If the person's ESR is 2, 3, or 6, then:

      • If the person's year last worked entry is 1979, the edit makes the final P31a value a "Yes."
      • If the year last worked item has an entry other than "1980" or "1979," the edit makes the final P31a value a "No."If the year last worked item is blank, the edit assigns P31a the intermediate value "NA."
  5. The major exception to the above rules applies when the person responds "No" in question P31a and does not respond to P31b and P31c: the edit makes the final value "Yes" only if the person reports that he received earnings in 1979 and if he last worked in 1979 or his ESR is 1, 2, 4, or 5; otherwise, the edit accepts "No" as the final P31a value.

D. Allocation Procedures

Allocated values for question P31a are assigned from hot-deck allocation matrixes that jointly allocate P31a and earnings and other income. Population controls are age, sex, ESR, actual hours worked last week, school enrollment, presence and age of children, and value of property and monthly rent. Before beginning the allocations, the computer sorts the file according to the following characteristics: sample size, sex, race and ethnicity, household relationship, educational attainment, type of area, district office, enumeration district, and serial number.

WEEKS WORKED IN 1979 and USUAL HOURS WORKED PER WEEK WORKED IN 1979

A. Purpose of Edit Procedures

This edit identifies the universe of persons 16 years old and over for whom data on weeks worked and usual hours worked are applicable. The edit also identifies persons for whom final values for question P31b and/or P31c are to be allocated from a hot-deck matrix.

B. Major Assumptions - None

C. Edit Procedures

For persons to whom the work experience in 1979 edit assigned a final value of "Yes" in question P31a, the P31b/P31c edit either accepts the response for P31b or P31c as the item's final value, or assigns the intermediate value "Needs Allocation (NA)" when the item is blank.

For persons assigned a final value of "No" in question P31a, the P31b/P31c edit assigns the final value "Not in Universe" to both P31b and P31c. This value means that these data cannot be tabulated for persons who did not work in 1979.

Finally, for persons assigned the intermediate value "Needs Allocation" in question P31a, the P3lb/P31c edit assigns the intermediate value of "Needs Allocation" to both P31b and P31c.

Persons assigned intermediate values of "Needs Allocation" will have their final values assigned from a hot-deck allocation matrix.

D. Allocation Procedures

Allocated values are assigned from one of 11 matrices as part of the joint-economic item hot-deck allocation system. Major population controls are age, sex, employment status, actual hours worked, earnings, class of worker, school enrollment, and value of property or monthly rent. The file is sorted as described for the work experience (P31a) edit.

WEEKS LOOKING FOR WORK

A. Purpose of Edit Procedures

This edit identifies the universe of persons 16 years old and over for whom data on weeks looking for work are applicable. It also establishes consistency between data on weeks looking for work (P31d) and weeks working (P31b). Finally, it identifies persons who must be assigned a final value for P31d from a hot-deck matrix.

B. Major Assumptions

By definition any week in which a person worked, even for one hour, cannot be included among weeks looking for work even if the person actively sought work during that week. In the question P31d edit, we sum the number of weeks the person worked (P31b) plus the number of weeks he or she reports that they looked for work (P31b). If the sum is less than or equal to 52 weeks, we assume that the person correctly interpreted the weeks looking question and entered only those weeks in which he did not work but looked for work.

C. Edit Procedures

If the final value for question P31b (weeks worked in 1979) is 52 weeks, the P31d edit assigns the final value of "0" weeks to P31d.

If the sum of the final value in P31b plus the entry in P31d exceeds 52 weeks, then edit assigns the final value of [52 minus the final value of P31b] to P31d.

Otherwise, the edit accepts an entry in P31d as the final value, or assigns the intermediate value of "Needs Allocation" the respondent leaves P31d blank. Persons assigned "Needs Allocation" are assigned a final value from a hot-deck allocation matrix.

D. Allocation Procedures

Allocated values are assigned from one of 11 hot-deck matrices as part of the joint economic item hot-deck allocation system. Major population controls are age, sex, employment status, actual hours worked, earnings, class of worker, school enrollment, and value of rent of property. The file is sorted as described for the work experience edit (P31d).

INCOME

A. Purpose of Edit Procedures

The total income inquiry (question P33) was introduced in the 1980 census pretest program to aid in the resolution of questionable income entry configurations in question P32 reported by self-employed persons. However, test results indicated that a sizable number of all income recipients furnished responses to P33 but failed to provide answers to some or all parts of P32. Thus, the subsequent computer edits were developed primarily to achieve consistency between the sum of the entries in P32 and the entry in P33. Also, edits were created to align misreported earnings amounts in P32a, b, and c with the corresponding class-of-worker entries in P30 and to assign specific dollar amounts to respondents with unspecified "loss" entries in P32b, c, d, and P33. These edits were expected to improve the quality of responses to the detailed income questions and minimize the need for imputation of income entries for partial responses to the income questions.

B. Major Assumptions

The major assumption in the income edit was that the sum of completely reported entries in questions P32a-g takes precedence over the entry in P33, with some exceptions for self-employed workers and individuals reporting weekly or monthly amounts. Also, if an amount is reported in P33 and there are one or more missing entries in P32a-g, all or some part of the amount in P33 is assigned to one of the income fields in P32a-g in accordance with the person's work experience entries in P28 and P31.

C. Edit Procedures

The following general rules were implemented in the machine editing of income entries.

  1. A dollar amount in any income item was accepted as such, and any entry in conflict with this response was disregarded. The only exception was an age entry of less than 15 years, the effect of which was to cancel a dollar code for income.
  2. When the "No" code was marked and no dollar code appeared in a given item in P32a-g, it was accepted that there was no income from that source.
  3. If both a positive or negative dollar code and the "No" (or "None") circle were marked for the same question, the dollar code was accepted and the "No" circle was cancelled.
  4. If, in the standard 4-digit code for an amount, any digit was missing, the code was invalidated and the question was considered to be unanswered.
  5. For a person who did not work in 1979 (according to the entry in P28 or P31a), entries in earnings questions P32a, b, and c were treated as "No" entries.
  6. If a dollar amount was reported for only one of the three earnings items, and the type of earnings was inconsistent with the class-of-worker status (P30) of the respondent, then the dollar amount was moved to the earnings entry which was consistent. For example, if a person identified in question P30 as an employee of a private company reported the sole source of earnings under "non-farm self-employment income" (P32b), the amount was moved to wage or salary income (P32a).
  7. If a dollar amount or "No" is reported in all parts of questions P32a-g, the entry in P33 is usually disregarded unless the income recipient is classified as "self-employed in own business not incorporated". If these self-employed persons reported an amount in P33 that was significantly less than the corresponding amount in P32b or c, the amount in P32b or c was changed to agree with the difference between the P33 amount and the sum of any amounts in P32a, d, e, f, and g. In these cases, the original amount in P32b or c was considered to be gross self-employment income and the entry in P33 net self-employment income.
  8. Major inconsistencies between the sum of the entries in question P32a-g and the amount reported in P33 were resolved as follows:

    1. If a dollar amount in P32a, e, or f was less than 10 percent of the amount reported in P33, the entry in P32 was modified to agree with the difference between total income in P33 and the sum of all dollar amounts in the remaining income fields in P32.
    2. If there was a dollar amount in P33 and no dollar amounts in P32a-g, the amount in P33 was assigned to one of the parts of P32, depending on selected characteristics of the income recipient.
    3. If there was a combination of dollar amount(s) and blank entries in P32a through g and the dollar amount in P33 was greater than the sum of dollar amounts in P32, the difference between the amount in P33 and the sum of the entries in P32 was assigned to a specific blank field in P32.
  9. All remaining cases of non-response, impossible or unreadable codes for income questions were handled by assigning the income of a person with similar demographic and economic characteristics.

D. Allocation Procedures

Prior to the allocation of all economic variables, the sample computer records were sorted according to such characteristics as sex, race and ethnicity, household relationship, years of school completed, and geographic area. All persons requiring the allocation of earnings and/or other income entries were then assigned values from a series of hot-deck matrices, wherein information was supplied from the data of the latest fully reported person with the same socioeconomic characteristics. This assignment of income entries was based on such matrix variables as age, disability status, presence of children, veteran status, work experience in 1979, occupation, class-of-worker status, level of earnings in 1979, and value of property or monthly rent.

FARM RESIDENCE

A. Purpose of Edit Procedures

The edits of question H15a are designed to provide data that are consistent with other housing data on acreage of property and with geographic information on the urban/rural location of the housing unit, and to allocate missing data. The edits of question H15b are designed to provide agricultural sales data that are consistent with responses to H15a, and to allocate missing data.

B. Major Assumptions

Group quarters are not in the universe for the farm residence questions. Any structure of 5 or more units is considered on a city or suburban lot or on a place of less than one acre. No attempt was made to force consistency between H15a and H10a (one-family house on 10 or more acres). If H15a was blank, however, and H10a indicated the unit was on 10 or more acres, then H15a was made 10 or more acres. An urban or vacant unit or one on a city or suburban lot or place of less than one acre was not in the universe for H15b.

C. Edit Procedures

If the unit is group quarters, the edit assigns "Not in Universe" (NIU) codes to H15a and b, and the edit ends. If the structure has 5 or more units, the edit assigns the response "CSL or < 1 acre" to H15a and the NIU code to H15b, and the edit ends. If the response to H15a is "CSL or 4.1 acre," or if the unit is urban or vacant, the edit assigns the NIU code to H15b.

If H15a is still blank after pre-edit, the edit looks at the responses to H13 and H10a and assigns a value of "10 or more acres" if H13 and H10a indicate the unit is a one-family house on 10 or more acres. Otherwise a response to H15a is provided through a hot-deck allocation. If H10a is blank, a response is provided at the same time as H15a through a hot-deck allocation. The choice of hot-deck matrix depends on whether H10a and H15b also have to be allocated.

If H15b is still blank after pre-edit, the edit provides a value through a hot-deck allocation. The choice of hot-deck matrix depends on whether responses to H15a and H10a also have to be allocated.

D. Allocation Procedures

When responses to H10a, H15a, and H15b all have to be allocated for the same unit, the controls used in the hot-deck matrix consist of the amount of farm self-employment income of the householder (P33c) and the urban/ rural location of the unit. When just H10a and H15a have to be allocated, the controls consist of the responses to H15b and the urban/rural location of the unit. In any other situation where H10a must be allocated, the procedure is handled by other housing edits.

When both H15a and b must be allocated, the controls used in the hot-deck matrix are the amount of farm self-employment income of the householder (P33c) and the urban/rural location of the unit. When H15a only needs to be allocated, the controls consist of the responses to H15b and the urban/rural location of the unit. When H15b only needs to be allocated, the controls include the responses to H15a, the amount of farm self-employment income of the householder (P33c), and the urban/ rural location of the unit.

ENDNOTES:

  1. The edit and allocation procedures for the 1980 census were never published by the Census Bureau. The material in this section was included in a March 20, 1997 memo from the Bureau to the Historical Census Projects. In the original memorandum, each variable description included a display of the original census question(s) involved. Users can view the entire 1980 Census form in Section 3.2 of this volume.

Go Back to Editing and Allocation Procedures Index