TECHNICAL NOTE 3 PEOPLE AND DWELLINGS MISSED


PEOPLE AND DWELLINGS MISSED IN THE CENSUS

1 Given the purpose of the PES is to check the coverage of the Census and provide estimates of how many people were missed (or counted more than once), the details of people and dwellings missed in the Census are necessary in order to obtain estimates of gross undercount, which then feed into net undercount estimates. This section presents a more detailed analysis of the data available on the people and dwellings missed in the Census.


People missed in the Census

2 In the PES, people are asked a number of questions about where they usually live, and addresses where they might have been counted in the Census, including:

• whether they were included on a Census form (and if so where);
• whether they could have been included on a census form at other addresses (and if so, where); and
• where they stayed on Census night.

3 For processing purposes, it was recorded whether the Census night address was the 'PES enumeration address' or a 'search address', that is, a different address supplied during the PES interview.

4 The Census form corresponding to the PES enumeration address was checked to determine if the person was counted in the Census at that address. Any search addresses were then checked to determine if the person was counted elsewhere in the Census. A key difference in the quality of this information is that exact address information was available for PES enumeration addresses, while search addresses relied on details provided by respondents. In 2011, Automated Data Linking (ADL) also identified, for the first time in PES processing, search addresses that had not been volunteered by respondents using probabilistic linking. For further information about ADL see Linking and Matching.

5 Based on results from PES linking and matching processing, there were a total of:

• 5,241 people (5.4% of the total number of people responding to the PES) who should have been counted in the Census but were missed (i.e. undercounted people); and
• 1,932 people who were PES respondents not counted in the Census and should not have been. This included diplomats, people who were overseas on Census night, and babies born after Census night.

6 The table below shows whether PES respondents were counted in the Census and whether they thought they were included on a Census form. Data have been cross-classified by characteristics of their Census night address. The data in this table are unweighted and hence do not correspond to weighted estimates included in other parts of this publication.

PES RESPONDING PERSONS, Whether counted in the Census and Whether thought they were included on a Census form(a)
		WHETHER THOUGHT THEY WERE INCLUDED ON A CENSUS FORM
		Yes - counted at PES address	Yes - counted at UR address	Yes - counted at other address	Don't know if counted in Census	No - not counted in Census(b)	Total(c)
Census night address		no.	no.	no.	no.	no.	no.
Counted in Census(d)
	PES address	81 495	32	84	653	518	82 782
	Usual residence address	14	1 266	2	176	15	1 473
	Other address	525	8	4 031	309	222	5 095
	Overseas	0	0	0	0	1	1
	Total(e)	82 035	1 306	4 117	1 140	756	89 356
Missed in the Census
	PES address	2 087	6	27	330	1 095	3 545
	Usual residence address	2	112	0	89	26	229
	Other address	190	1	474	255	531	1 451
	Overseas	2	1	0	2	4	9
	Total(e)	2 281	120	505	678	1 656	5 241
Out of PES scope coverage
	All addresses	535	0	62	130	1 205	1 932
All persons(f)
	PES address	83 651	38	111	990	1 624	86 414
	Usual residence address	16	1 378	2	265	41	1 702
	Other address	716	9	4 535	578	767	6 605
	Overseas	467	1	32	111	1 185	1 796
	Total(e)	84 851	1 426	4 684	1 948	3 617	97 047
(a) As reported in the PES.
(b) Includes people who did not know whether they were included on a Census form.
(c) Includes people who did not report (or were not asked) whether they thought they had been included on a Census form.
(d) Includes people who should not have been counted in the Census.
(e) Includes people who did not provide a Census night address.
(f) Includes people who were not counted and should not have been counted in the Census.

7 The majority of people were expected to have been counted in the PES at the dwelling where they stayed on Census night. The Census night address matched the PES enumeration address for 3,545 persons (around 68%) of missed people, compared to 82,782 persons (around 93%) of all people counted in the Census.

8 Almost a third (1,656 persons) of missed people were believed to have not been included on a Census form, with a further 678 persons (around 13%) not sure if they had been included on a form. The remaining missed persons were believed to have been included on a Census form, with most thought to have been counted at their PES enumeration address.

9 Of the people who were counted in the Census (89,356 persons), less than one percent (756 persons) were thought to have not been counted in the Census, with a further 1.3% (1,140 persons) not sure if they had been included.


Dwellings not counted in the Census

10 For a variety of reasons, not all dwellings received a Census form. This may have been because the dwellings were not habitable, the Census collector did not identify them, or in the case of a diplomatic dwelling, did not contain people within scope of the Census. Census field procedures required that all dwellings, apart from the types listed below, be recorded, irrespective of a received (or returned) Census form. Hence all dwellings in Australia should have been counted in the Census, with only the following exclusions:

• Dwellings under construction;
• Derelict dwellings;
• Vacant tents;
• Unoccupied converted garages;
• Unoccupied caravans; and
• Unoccupied dwellings within marinas/manufactured home estates.

11 In 2011, the ability of the PES to match PES non-responding dwellings and sample loss dwellings to their corresponding Census dwellings was reduced. This was due to the introduction of ADL which changed the linking and matching processing methodology to focus on persons initially, then dwellings. Instead of dwelling matching being a necessary initial step in linking and matching, most dwellings were matched through links established for persons in ADL. This therefore altered the systematic nature of dwelling matching that took place in the previous PES methodology. For more information on ADL see Linking and matching.

12 Estimates of dwelling undercount have not been provided in this publication due to some concerns over the representativeness of PES non-responding and sample loss dwellings (which show a higher than expected proportion of missed dwellings, and a lower than expected proportion of unoccupied dwellings). Dwelling matching outcomes for the dwellings responding in the PES are included below.

DWELLINGS, PES response by Census response type - 2011
	Occupied	Un-occupied	Late returns	System created records	Missing	ICF not missing	ICF missing	Total
	no.	no.	no.	no.	no.	no.	no.	no.
Responding	32 228	1 504	377	656	549	551	41	35 906