The Canadian Experience witli Post
Censal Surveys

by Adele D. Furrie , Program Manger

Post-Censal Surveys Program

and Craig McKie ' Editor

Canadian Social Trends, Sttatistics Canada

Abstract

The Canadian experience with post-censal survey is, at

this point in time, limited to two surveys - one which
used the 1971 Census returns as the sampling frame, and
the 1986 survey, which used the 1986 returns as the

sampling frame, utilized the Census field organization to

collect the data, and used the 1986 Census data to

supplement the data collected 'n the post-censal survey.

Post-censal surveys provide efficiencies in terms of

overall costs because of the accessibility of the Census
data to identify relatively rare populations. The availa-

bility of a field organization to a) select the sample
immediately following the collection of the Census data

and b) collect the data reduces the overall cost to the

post-censal program since the hiring and some of the

training costs are absorbed by the Census. The availabil-

ity of the Census data, not only for the population of

interest, but for the data base. Respondent burden is

reduced because many of the demographic and socio-

economic variables are included in the Census. This is

true also for family and household-related data.

Some consideration is being given to the conduct of a

similar post-censal disabihty survey following the 1991

Census. In addition, there is the possibility for a survey

on the senior population and one on aboriginal persons.

The Canadian experience with post-censal surveys is

limited to two surveys - the Highly Qualified Manpower
Survey (HQMS) conducted following the 1986 Census.

While the methodology of the two surveys differed

significantly, each provides an interesting application of

post-censal methodology which could be considered by
those who are planning such activities. The methodology
for each survey will be described and an evaluation will

abe provided of the methodology.

The Highly Qualified Manpower Survey - HQMS - was
conducted in the fall of 1973, just over two years after

the 1971 Census was conducted. The objective of the

survey was to assess past expenditures for education in

relation to the utilized labour force status. The data were
needed to assist in the formulation of policy relating to

l(Mig-range planning in the fields of education and
manpower planning.

The sample was selected from the Census database based
on the individual's age-sex-labour force status and level

of post-secondary attainment. Because name and address

were not part of the Census database, the original Census

questionnaires had to be accessed, and information such

as telephone number and the name of the head of the

household as well as the name and address of the selected

person was transcribed. Quality control procedures for

the transcription and creation of the name and address

file were employed to control for errors.

The questionnaire covered a limited range of topics such

as field of study, current labour force status and current

earnings, and an employment profile over time. These

data were supplemented by the information collected in

the 1971 Census, so that the combined database covered

a wide range of topics, such as ethnicity, immigration

status, marital status, etc.

The survey questionnaire was mailed to approximately

138,000 selected persons (the number of persons meeting

the selection criteria were approximately 720,000); about

72,000 or ahnost 70% returned the completed question-

naire. A tracing operation was conducted to establish the

current address fw those selected persons who had

moved. Follow-up of non-respondents was conducted by

mail, and in some instances, by personal visit, during the

period September, 1973 to March, 1974.

Because the survey was conducted over two years after

the Census, there was some difficulty in locating some of

the selected persons and some of the data retrieved from

the Census were out-of-date. For example, the individual

may have married or additional children may have been

bom, and these differences would not have been ac-

counted for in the combined database.

The 1986 post-censal survey, the Health and Activity

Limitation Survey (referred to in this paper as HALS)
employed a different collection methodology, taking into

account the experience of the HQMS. HALS was
comprised of three distinct surveys - two household

surveys - both of which used the disability question on

the Census as the screen to identify the sample and an

institutions survey which used the Census to identify the

location, size and type of the institution.

The need for a comprehensive database on disabled

persons was articulated in 1981 in the report entitled

Obstacles, the report from the Special Parliamentary

Committee on the Disabled and the Handicapped. It

Spring 1990



noted that there were no national data available on
disabled persons. Not only were there no national

estimates on the number of disabled persons in Canada
and the nature and severity of their disability, but, as

important, there were no estimates on the barriers which
these disabled Canadians face in the conduct of their

everyday activities. Because many of the programs and
services offered to disabled persons are the responsibility

of provincial and local governments, it was important

that these data be available for relatively small geo-

graphic areas. As programs and services differ for

different age groups, d^e sample would also have to be
large enough to be able to generate estimates within the

geographic areas for each age group. The methodology
of a post-censal survey was considered as the only viable

alternative to obtain this level of detail. Other options

such as the use of existing survey vehicles - the monthly
Labour Force Survey or the annual General Social

Survey - were considered but it was determined that

neither would yield a sufficiently large sample of dis-

abled persons. There were also limitations in the cover-

age for both of these survey vehicles. The Labour Force
Survey excludes the more remote areas of Canada,
residents of Indian reserves, and residents of institutions.

The General Social Survey generaOy utilizes a random-
digit dialing methodology to create a sample, therefore,

households without telephones would be excluded from
the survey. The General Social Survey is also household
based so that residents of institutions would be excluded
from the survey.

The decision to use the Census as the method to identify

the samples for the household surveys necessitated the

inclusion of a disability question on the Census question-

naire. It was decided that this question would be added
to the "long" questionnaire - the one completed by one
out of every five households. The question asked if the

individual was Umited in the kind or amount of activity

he/she could undertake because of a health problem or
condition. The second part of the question asked if the

individual had any long-term disabilities or handicaps.

Households were advised through the Guide that was
included with their Census questionnaire that the disabil-

ity question was to be used to identify a population for a
more-intensive survey on the issues facing disabled
persons.

It was determined through a pre-test that this question
would identify most of Lhe more-severely disabled
population, but that additional questions would be
required to identify all disabled persons. A copy of the

Census disability question is included in Appendix A of
this paper.

The content of the HALS questionnaire was determined
through extensive consultation with representatives from,

government departments that provided programs for and
services lo disabled persons. Consultation with organiza-
tions of and for disabled persons was also undertaken to

ensure that their needs were reflected in the content. The
questions used to identify the nature and severity of the
individual's disability were, for the most part, developed
by the O.E.C.D. These questions are known as the

Activities of Daily Living and were developed to identify

physical and sensory disabilities. Other questions were
added to identify emotional, psychological and learning

disabilities and persons who are developmental!y de-

layed.

With the inclusion of the disability question on the

Census, the sampling frame was in place for the post-

censal survey of disabled persons. To maximize the

efficiency of this sampling frame, it was decided that an

operation should be integrated with Census processing to

select a sample of individuals who had responded "Yes"
to the Census disability question. This would allow for

the conduct of the post-censal survey shortly after the

Census, thus minimizing the follow-up required because
of inter residence moves taking place between the lime of

the Census and the conduct of the post-censal survey. It

would also enable the utilization of the Census field staff

to conduct the face-to-face interviews.

To accommodate the selection of the sample and to

ensure that the field staff would be available for further

work, geographic areas were identified prior to the

conduct of the Census. These areas were defined as the

geographic area within which the workload for one
Census Field staff (Census Representative) was located.

Census staff received additional training on the concepts

and definitions used in HALS and the face-to-face

interviews were conducted immediately following the

Census collection. The reference day for the 1986
Census was June 1. In most instances, the interviews

were completed during August and September, 1986.

There were an estimated 120,000 individuals selected for

the follow-up interview; the overall response rate was in

excess of 95%. The second household sample involved a

sample of individuals who had responded "No" to the

Census disability question. This sample was necessary

because the pre-test had indicated that some disabled

persons may not respond affirmatively to the general

disability question. A sample of approximately 80,000
individuals was selected during a later stage in the

Census processing, but before the Census documents
were returned to Head Office in Ottawa. The same
questionnaire was used and most respondents were
contacted by telephone. Those who did not provide

telephone numbers on their Census questionnaire were
contacted in person. The survey was conducted from the

Regional Offices of Statistics Canada during October and

November, 1986 by interviewers who are part of the

regular Regional Office staff. Approximately 90% of the

sample was contacted and agreed to participate in the

survey.

The data from, both household surveys was integrated

with the Census at the micro-record level, so that the

linked database contains information from both the

Census and HALS. The Census data, for die most part,

is for the selected individual, but included in the base are

also some variables about the family and household
within which the selected person resides. Because the

Census and HALS were conducted within six months of

each other, the variables taken from the Census such as

marital status should not have changed significantly.

14 ASSIST Quarteriy



Another feature which adds to the richness of this

database results from the sample being selected to

represent both "Yes" and "No" respondents to the Census
disability question. The HALS sample can be divided

into strata - those who are disabled and those who are

not For the non-disabled population, the data available

on the linked database includes all of the Census vari-

ables. For the disabled peculation, the data includes both

Census and HALS variables. This affords the user the

opportunity to make comparisons of the characteristics of

the disabled and the non-disabled populations.

The Census methodology did not include the use of the

long questionnaire in institutions; therefore, the disability

question was not asked in institutions. To obtain infor-

mation for residents of institutions, the Census was used

to identify the location, size and type of institution.

Penal institutions and correctional facilities were ex-

cluded because of operational difficulties. From the

remaining list, a sample of institutions was selected,

approximately 1,100 out of a total of approximately

5,300. Each of the selected institutions provided a list of

residents from which a sample was selected. Of the

18,200 residents selected from the 1,100 institutions, less

than 3% refused to participate in the survey. A personal

interview was conducted with slightly over 50% of the

sample. For the remaining sample of respondents whom
the institution administrator deemed to be too ill or too

disabled, the interview was conducted with an individual

who provided the day-to-day care.

Data from HALS was released in May, 1988 and has

been used by both the public and private sectors. Much
has been learned concerning the conduct of post-censal

surveys and the integration required with the Census
operations.

Planning is now underway for post-censal survey activity

following the 1991 Census. Based on consultation with

representatives involved in social programs, three

potential topics have emerged and further consultation is

now underway. The three topic areas are a survey of

seniors with the focus on support networks, a repeat of
the survey of disabled persons so that data are available

over time, and a survey of aboriginal persons, both on-

and off-reserves. The possibility of one or more of these

topics going forward is contingent on obtaining funding
for them. The 1986 survey cost seven million dollars.

That survey proved that a post-censal survey, closely

linked to the Census operation in terms of identifying the

sample, utilizing the Census field staff, and the Census
data is a viable option for surveys of relatively rare

populations or which require significant geographic
detail. Q

Appendix A

20. (a) Are you limited in the kind or amount of

activity that you can do because of a long-term

physical condition, mental condition or health

problem:

At home?

No, I am not limited

Yes, I am limited

At school or at work?

No, I am not limited

Yes, I am limited

Not applicable

In other activities, e.g. transportation to or

from work, leisure time activities?

No, I am not limited

Yes, I am limited

(b) Do you have any long-term disabilities or

handicaps?

No

Yes

' Presented at the IFDO/IASSIST 89 Conference held in

Jerusalem, Israel, May 15-18, 1989.

Spring 1990