CONTENTS



Introduction

Background

Data analysis

Telephone follow-up

Conclusion and recommendation

LIST OF TABLES

Table 1 Non-response to ancestry questions in the 1986 Census and August 1993 Census Test

Table 2 Persons with Australian ancestry in the 1986 Census and August 1993 Census Test

Table 3 'Yes' responses to the August 1993 Census Test questions on ancestry

INTRODUCTION

Before each Census, the Australian Bureau of Statistics (ABS) consults widely with the community and users of Census data about the topics which should be included.

In February 1993, a publication entitled 1996 Census of Population and Housing: ABS Views on Content and Procedures (2007.0) was released. It outlined ABS views on the content and procedures to be used in the 1996 Census and invited submissions about the views expressed in the publication.

In response to a number of submissions requesting the inclusion of an ethnicity topic in the Census, the ABS included the topic in a Census Test in Melbourne in August 1993. The key issue to be assessed was the extent to which the information from an additional question was compatible with, but added to the data collected in, existing questions. In particular, given the origins of this proposal, the level of compatibility with the results of the 1986 Census question was to be assessed.

This report explains why and how a Census question on ethnic origin was tested, and analyses the results obtained from the test.

BACKGROUND

The ABS assessment of the topic of ethnic origin in 1996 Census of Population and Housing: ABS Views on Content and Procedures was that the needs of most users were well met by the data available from other Census indicators of ethnicity such as birthplace, parents' birthplace, citizenship, religion and language use, and that an additional direct question on ethnic origin similar to that used in 1986 was not justified, particularly in view of the high cost of processing the question.

However, during the submission and consultation processes, it became obvious that some users wanted more comprehensive information on ethnic origin. There were representations from a range of academics, government departments and ethnic community groups supporting the inclusion of a question similar to that asked in the 1986 Census. It was also clear that the ancestry question could not replace any of the surrogate measures of ancestry which were used for service delivery, and one key issue under debate was whether the additional information from the ancestry question warranted the additional respondent load and resources needed to collect and process the question.

Two different questions on ancestry were tested in the August 1993 Census Test on two Census Test forms (referred to hereafter as Form 3 and Form 4). The forms were delivered to alternate households. Both questions attempted to collect 'additional' ancestry information to that collected in other questions but without requiring as many responses to be coded as the 1986 Census ancestry question (and therefore not being as expensive to process). At the request of the Australian Statistics Advisory Council (ASAC), the question on Form 4 (see below) was based on a 'self-identification' approach rather than the ancestry approach used in the 1986 Census.

The Form 3 question:

12 Is the person's ancestry different ( ) No
from their country of birth? ( ) Yes - please specify ancestry
∙ The person may identify with more than
one ancestry.


The Form 4 question:

11 Does the person identify with an ( ) No
ancestry different from their country ( ) Yes - please specify ancestry
of birth?
∙ The person may identify with more than
one ancestry.


The 1986 question:

15 What is each person's ancestry? Ancestry ................................
For example, Greek, English, Indian, Armenian, ................................................
Aboriginal, Chinese etc. ................................................


In order to gauge how well the ancestry questions would work, some test areas were chosen to include a relatively high proportion of overseas born persons. In addition, four Collection Districts with a high proportion of people reporting Judaism as their religion in the 1991 Census were included (since it was suggested that these persons may report 'Jewish' as an ancestry).

DATA ANALYSIS

Analysis of the data gathered with the two questions tested and the follow-up interviews enable the following assessment of their performance to be made. Non-response poses problems for the interpretation of data obtained with both test questions.

TABLE 1: Non-response to ancestry questions in the 1986 Census and August 1993 Census Test (per cent)
August 1993
Census Test
1986 Census Form 3 Form 4
Non-response rate
All persons 6.8 7.3 12.2
Australian-born 7.0 6.3 8.7

Non-respondents by birthplace
Australian born 79.7 56.5 45.5
Overseas born 16.3 41.5 47.3
Birthplace not stated 4.0 2.1 7.2
Sub-total 100.0 100.0 100.0

Non-respondents by birthplace of parents
Both Australian-born 59.7 23.3 17.2
Both overseas-born 8.3 56.6 64.6
One born overseas 9.5 11.4 9.4
Neither stated 22.5 8.8 8.8
Sub-total 100.0 100.0 100.0


Table 1 shows that actual non-response to the question on Form 3 was 7.3 per cent which compares favourably to the non-response rate for the 1986 Census question (6.8% for Australia and 8.0% for Victoria). On the other hand, non-response to the Form 4 question was much less acceptable at 12.2 per cent. It should be noted that question response rates in tests are commonly higher than those realised in Censuses due to their voluntary nature.

Of more concern than the level of non-response is the ethnic background of those who did not respond in the test. With more than two thirds of the non-respondents to both form types having at least one overseas-born parent, it is possible that a significant number of latent 'Yes' responses have been lost. In this connection, the performance of Form 4 is worse than that of Form 3 due to a higher level of non-response (Form 3, 7.3%; Form 4, 12.2%) and a higher ratio of non-respondents with an overseas-born parent (Form 3, 68.0%; Form 4, 74.0%).

While 16.3 per cent of non-respondents to the 1986 Census question were overseas-born, 41.5 per cent of Form 3 non-respondents and 47.3 per cent of Form 4 non-respondents were overseas-born. Sixty-eight per cent of Form 3 non-respondents and 74.0 per cent of Form 4 non-respondents had at least one parent born overseas, compared to only 17.8 per cent in the 1986 Census (see Table 2).

TABLE 2: Persons with Australian ancestry in the 1986 Census and August 1993 Census Test (per cent)

August 1993
Census Test
1986 Census Form 3 Form 4
Australian born and Australian ancestry
All 25.8 72.9 74.7
Both parents born in Australia n.c. 92.6 89.0
Both parents born overseas n.c. 27.7 44.5
One parent born overseas n.c. 39.0 51.1

All persons 20.3 47.8 47.7

n.c. Not calculated

For both questions in the test, the proportion of Australian only ancestries derived (i.e. an answer of 'no' by people born in Australia) was almost 48.0 per cent compared with the 20.3 per cent of people who provided 'Australian' as their first or only response to the 1986 Census question on ancestry (Table 2). For Australian-born people with both parents born overseas the proportions of 'Australian' ancestries ranged between 27.7 per cent (on Form 3) and 44.5 per cent (on Form 4).

For Australian-born persons in the test, the proportions of 'Australian' responses were 72.9 per cent (Form 3) and 74.7 per cent (Form 4) compared with 25.8 per cent in the 1986 Census data for Australia (Table 2). For Australian born people with one parent born overseas, the proportion of Australian ancestries were 39.0 per cent and 51.1 per cent respectively.

In other words, many Australian born people who clearly have a non-Australian ancestry have not reported it (Form 3) or identified with it (Form 4).

TABLE 3: 'Yes' responses to the August 1993 Census Test questions on ancestry (per cent)

August 1993 Census Test
Form 3 Form 4

All persons 19.0 16.7
Australian born 20.8 16.7
Birthplace England 10.5 8.4
Birthplace Scotland nil (a) 12.5
Birthplace Italy 3.5 7.1
Birthplace Greece nil (b) 8.0
Birthplace New Zealand 29.2 27.6
Birthplace 'Other' 18.6 19.2

Australian born
Both parents born overseas 59.4 37.4
One parent born overseas 50.8 34.4

All persons by age
0 - 9 years 16.0 18.1
10 - 19 years 22.6 14.6
20 - 29 years 24.3 18.5
30 - 39 years 21.6 20.8
40 - 49 years 19.7 18.4
50 - 59 years 12.1 8.6
60 - 69 years 14.9 14.7
70 - 79 years 8.7 14.0
80 - 89 years 14.8 8.5
90+ years and write-in only nil (c) nil (d)

(a) Two Scottish-born persons answered 'No'.
(b) Twenty-nine persons born in Greece answered 'No' and five failed to respond.
(c) Four persons in this category failed to respond.
(d) Seven persons answered 'No' and three failed to respond.

As expected, the Form 4 question yielded a lower rate of 'yes' responses (people claiming to identify with an ancestry different from their birthplace) than the Form 3 question (people claiming to have an ancestry different from their birthplace). This was true for most categories (see Table 3).

It is unlikely that similar results obtained from the Census would be accepted by users of this data.

TELEPHONE FOLLOW-UP

Follow-up telephone interviews were included in the test to explore a number of issues, including ancestry. In all, the interviews obtained information about ancestry responses for 310 people on both form types, including instances of non-response and answers of 'yes' and 'no' to the test questions. The aim was to determine what respondents understood by each of the test questions and how answers were determined in each case.

Follow-up telephone interviews were conducted as part of the evaluation of responses to the ancestry questions to determine whether or not respondents understood the question and to provide an insight into how they determined their ancestry. Interviews were conducted with a range of respondents, including persons who did not respond to the ancestry question; persons who answered 'yes' to the ancestry question; and persons who answered 'no' to the ancestry question and who were born in Australia but who had at least one parent born overseas.

In the interviews, respondents were asked what the word 'ancestry' meant to them, how the answer was determined for particular persons or why there was no answer provided, why more than one ancestry was included (if applicable) and how respondents decided which ancestries to include.

The majority of those interviewed (60.0%) considered ancestry to refer to their forebears, parents or grandparents and about 13.0 per cent thought it referred to ethnic or cultural background. About 15.0 per cent said they did not know what 'ancestry' meant and the rest (about 12.0%) gave answers that were too vague or too confused to be categorised.

Although the majority of people interviewed appeared to have an understanding of the word ancestry, not understanding the question was the most common reason given for non-response (41.0% of non-respondents interviewed). About 20.0 per cent of the non-respondents interviewed said they inadvertently missed the question. Another common reason given for non-response in Form 3 (40.0%) was not knowing how many generations to go back. Interestingly, a greater variety of reasons were given for Form 4 non-response (which, as we have already noted, was high). Some 42.0 per cent said they did not understand the question, 31.0 per cent said they 'missed' the question, 4.0 per cent were uncertain about the number of generations to go back and 23.0 per cent gave a variety of other reasons.

A common observation made in a large number of the follow-up interviews was that, while respondents felt they knew what the word ancestry meant, they were not sure how they should respond to the question in either test form. This is not surprising, considering the mixing of concepts that the questions involve. The uncertainty and confusion in respondent minds was clearly more pronounced among respondents to Form 4 where the mixture of concepts was more complex. It was also evident from some of the telephone follow-up interviews that, given another opportunity, some respondents could well have answered the question differently from the way they had responded on the test form. These changes in response might have been due to the passage of time, the result of talking to an ABS officer or simply second thoughts (or indeed a whim). This brings into question the validity of the first response, and has implications for future responses.

On the other hand, the ancestry questions appeared to have worked slightly better for persons reporting Judaism as their religion. People reporting Judaism as their religion were much more likely than persons of other religious denominations to report an ancestry different from their place of birth. This was the case for both form types, although overall, Form 3 performed better. For Form 3, 45.4 per cent of persons reporting Judaism as their religion answered that they had an ancestry different from their birthplace. This compares with 29.9 per cent of Form 4 respondents.

This additional analysis for people with Jewish religion was undertaken because it was thought that they would form a group with a reasonably strong affiliation with a non-Australian ancestry (probably Jewish). Test answers seem to bear this out in that this group provided a higher proportion of 'yes' answers to both questions. Analysis of the responses also showed a difference between the kind of answers given in Form 3 and those given in Form 4. The Form 3 question tended to elicit factual responses of ancestry (such as parent's birthplace), with a small number reporting a Jewish/other ancestry combination. In Form 4, however, respondents who answered the ancestry question appeared much more likely to 'identify' with a Jewish ancestry.

Although the question about identifying with an ancestry worked well for people of the Jewish religion, that clearly was not the case with other ethnic groups. Response rates and general respondent reaction to this question show that its intention is not easily understood by most respondents.

CONCLUSION AND RECOMMENDATION

The test results indicate that Census data on ethnic origin would be subject to serious problems of interpretation, validity and stability.

The telephone follow-up results confirmed previously accumulated evidence that ancestry is not a concept that is clearly understood and consistently interpreted by the community. While most people feel they know what the word means, many answers about the subject in the Census Test context were ambiguous. It was also clear in some follow-up discussions that people found the combination of 'identifying' and 'ancestry' in the one question (in Form 4) particularly confusing.

Discussions also indicated that some respondents would have been happy to change their response from that shown on the test form. In some cases they had chosen between answers that appeared equally valid to them, in others they arrived at a different interpretation at the time of the follow-up. Because of this, data collected with such questions would be unlikely to be stable, and would have serious doubts over its validity.

Based on this experience, the ABS has recommended that no direct question on ethnic origin be included in the 1996 Census.