|
|
|
Data Quality Issues
Summary of Data Quality Issues
There are several technical and discussion papers that discuss the data quality issues that we are aware of in the datasets. A summary of these data quality issues is provided in Table 38. As further research is carried out on a variety of data quality issues, this table will be added to.
Table 38: Summary of the data quality issues in the HILDA data
| Topic / variable |
Problem |
Where to get more information |
| Sample Representativeness |
| Wave 1 non–response |
The wave 1 response rate was 66% and non–respondents were more likely to be living in Sydney, male or unmarried, aged 20 to 24 or 65+, or born in non–English speaking country. |
Watson and Wooden (2002a) pp. 3-8 |
| Attrition |
The attrition rates from wave 2 are provided in Table 59. Attritors are more likely to be living in Sydney and Melbourne; aged 15 to 24 years; single or living in a de facto marriage; born in a non–English-speaking country; Aboriginal or Torres Strait Islander; living in a flat, unit or apartment; relatively low levels of education; unemployed; or working in blue–collar or low–skilled occupations. |
Watson and Wooden (2006);
Watson and Wooden (2004a) pp.2–14 |
| Missing data |
| Item non–response |
| General level of item non–response |
Overall, the level of non–response in the HF, HQ and PQs is generally relatively low – less than 2 per cent. The item non–response rates in the SCQ are higher – averaging around 2.5 to 2.8 per cent |
Watson and Wooden (2002a) p9;
Watson and Wooden (2004a) p15 |
| Missing income data |
10–16 per cent of respondents did not provide details for all financial year income components, resulting in 22 to 29 per cent of households with missing financial year income.
Analysis of Wave 1 data shows that individuals with missing financial year information were more likely to be female; living in Sydney and rural WA; or attach a high importance to their financial situation.
The income data is imputed. |
Section above ‘Income Variables and Income Imputation’; Section below ‘Missing income data and the extent of income imputation’; Watson and Wooden (2002a) pp. 9–12 |
| Missing wealth data |
14 per cent of respondents did not provide allperson–level wealth details and 20 per cent of households did not provide all household–level wealth details, resulting in 39 per cent of households with missing wealth data (in wave 2) and 29 per cent in wave 6.
The wealth data is imputed. |
Section above ‘Wealth Variables and Wealth Imputation’; Section below ‘Missing wealth data and the extent of wealth imputation’; Watson and Wooden (2004a) pp. 21–24 |
| Family background |
People living with both parents in wave 1 were not asked the family background questions on the assumption that this could be derived from the parent's interview. However, not all parents responded or it was impossible to determine what the parent was doing when the respondent was aged 14. |
Watson and Wooden (2002a) pp. 12-13 |
| Permanently unable to work |
452 respondents were incorrectly coded as ‘permanently unable to work’ at D21 in the PQ (interviewers were meant to check back to D6, but many used the response at D20 to code D21). As a result, the questions for those not in paid employment were not asked (such as whether looking for work, main activity, whether they would like work, and whether they have retired). Note that the retirement questions were asked in later waves. |
Watson and Wooden (2002a) pp. 13-14 |
| Incomplete households |
| Part–responding households |
8 to 10 per cent of households are partially responding (that is, some but not all adults in the household provide an interview). When using derived variables that sum information across individuals in the household (for example, income or wealth variables), there will be more missing data. |
Watson and Wooden (2002a) p14; Table 51, Table 53 to Table 57. |
| Accuracy of the data |
| Questionnaire design issues |
| Childcare costs |
The child care grids in the HQ are very complex and require the parent to split the costs by the type of children (those of school aged and those not yet at school). There is some (small amount of) evidence that some respondents struggled to do this, with the same amount being reported for the two groups of children when the number of children in each group is not the same. |
Watson and Wooden (2002a) p15 |
| Current wages and salaries |
There are some respondents who reported having current wages and salaries but who:- did not report having a job (13 respondents in wave 1).
- were recorded as an employer (414 respondents in wave 1).
There were also some respondents who did not report having current wages and salaries but who:- were recorded as an employee of their own business (126 respondents in wave 1).
- were recorded as an employee (16 respondents in wave 1).
There may be some circumstances that can explain these apparent discrepancies (for example, a spouses who have income from the family business but who do not actually work in the business). |
Watson and Wooden (2002a) pp.15-16 |
| Calendar |
In wave 1, we tried to separate jobs out based on whether they were full or part–time and asked the interviewers to record job numbers so we can identify jobs changing between part–time and full–time. However, this was not completed by the interviewers very often or was (mistakenly) not entered by the processing team.
Interviewers also did not have sufficient instruction on how to treat breaks in employment (such as long–term leave or infrequent hours).
The design of the calendar was modified between wave 1 and 2 due to some problems identified with the calendar in wave 1. |
Watson and Wooden (2002a) p16 |
| Marital status |
The HF and PQ in wave 1 asked whether respondents were ‘legally’ married with the intent of asking about a a ‘registered’ marriage. We suspect some defacto couples reported they were ‘legally’ married because they have certail legal rights under the Australian legal system.
From wave 1, we have revised the questions to talk about ‘registered’ marriages. As a result, there may be inconsistencies between wave 1 and later waves. |
Watson and Wooden (2002a) p16 |
| Time use |
While we undertake a large amount of checking and editing on the time use questions in the SCQ, it is likely that problems remain. The problem areas are:- Excessive hours reported suggest respondents find it difficult to think in terms of hours in a week.
- The same hours may be recorded against multiple tasks if respondents are doing more than one thing at a time (eg looking after children while doing the housework).
- Some confusion was caused by the layout of the boxes as some respondents tried to record both hours and minutes.
The design of the time use question has undergone some revision since wave 1 to try to address these problems, but it is expected that errors still occur. |
Watson and Wooden (2002a) p17 |
| Leave entitlements |
In the wave 1 SCQ, respondents were asked about their access to paid and unpaid maternity leave in their current job. To avoid additional skips for men, a ‘not applicable’ option was provided. However, 1535 men provided answers to these questions, presumably answering whether other employees at their workplace had access to maternity leave. Also, older females selected ‘not applicable’ because they were not planning on using such leave.
The questions were revised in wave 2. |
Watson and Wooden (2002a) p17 |
| Moving house |
In wave 2, we asked movers when they moved to their current address, but did not ask when they left their previous address. For people who move twice in a year, we do not know the exact length of tenure at the former address. The questionnaire was amended in wave 3. |
Watson and Wooden (2004a) p30 |
| Duration of defacto relationship |
In waves 2 and 3, we asked those completing the NPQ how long their most recent defacto relationship started and how long it lasted. This is inconsistent with wave 1, where we asked about the first such relationship and from wave 4 these questions have been reverted to the original ones. |
Watson and Wooden (2004a) p30 |
| Data collection issues |
| Sex |
A small number of individuals had their sex corrected in the next wave (in wave 2, 37 people's sex was corrected).
Note that the latest sex and date of birth is applied back through the earlier waves. This may lead to some subsequentially introduced inconsitencies in the question skips that rely on age or sex. |
Watson and Wooden (2004a) pp.25–26 |
| Date of birth |
A relatively small number of corrections are applied to a person’s date of birth in the next wave. (In wave 2, there were 50 people with a major change to their date of birth and 451 with a minor change. In later years, the number of changes were less and usually to replace dates of birth that were missing for new entrants to the household.)
Note that the latest sex and date of birth is applied back through earlier waves. |
Watson and Wooden (2004a) pp.25–26 |
| Working hours |
In Wave 1, respondents were asked to compare their current hours with those a year ago. 26 cases reported hours a year ago that were inconsistent with their answer of whether they were more or less. The answer to the later was changed to reflect the former.
Similarly, a small number of cases (in wave 1, there were 7) were inconsitent with their answer to whether they wanted to work more or less and the number of hours they wanted to work. Generally the answer to whether they wanted more or less hours was altered.
For those with two jobs, some recorded more hours in all jobs that was less than their main job (in wave 1, there were 13). The hours in all jobs were usually set to –6 (unbelievable value).
For those who work at home, some recorded more hours worked at home than in their main job (in wave 1, there were 33). Where this could not be resolved by looking at the hours worked in all jobs for multiple job holders, the hours worked at home were usually set to –6 (unbelievable value). |
Watson and Wooden (2002a) p19 |
| Interviewer observations |
Interviewes were required to complete observations of the dwelling and of the PQ interview. Unfortunately, not all interviewers completed this. For example, in wave 1, about about 0.1–0.4 per cent of cases had missing values. |
Watson and Wooden (2002a) p20 |
| Mode effects and social desirability / acquiescence bias |
Differences observed are quite small in absolute terms. Items tested:- difference between reported health in PQ and SCQ in wave 1;
- whether responses tempered by presence of other adults during the interview.
|
Watson and Wooden (2002a) pp.21–22 |
| Cross–form comparisons |
| HF and PQ |
Few questions are asked more than once. Proportion of cases where answers differed in wave 1 between HF and PQ:- 10% for long–term health condition;
- 6.1% for labour force status;
- 0.4% for marital status.
Note HF and PQ may be done on different days and answered by different people. Also the questions were not identically worded. |
Watson and Wooden (2002a) p22 |
| Cross–wave inconsistencies |
| Marital status changes |
Respondents are asked whether they changed their marital status since the last wave interviewed. Some report a different status but say there has been no change (for example, there were 258 respondents reporting no changing their marital staus since wave 1 but who had a different status). Most of these errors are recall errors but a small number may also be transcription errors by the interviewer. |
Watson and Wooden (2004a) p27 |
| Address changes |
Address changes can be identified through either a comparison of actual addresses recorded on the HF undertaken by Nielsen or via a question in the PQ. In wave 2, for example, 119 people indicated in their PQ that they had not changed address, but the address recorded was different and 141 people said they had moved, but the HF address was the same. |
Watson and Wooden (2004a) p27 |
| Employment status changes |
Respondents are asked to recall whether they were employed or not at the previous interview. In wave 2, for example, 4.6 of those employed in wave 1 did not recall being employed then and 6.8 per cent of those not employed in wave 1 recalled that they were.
A very detailed analysis is given in Goode (2007).
The majority of mistakes are made by those who change employment states between interviews. Variables significantly associated with making a mistake are being in full time education, the number of children, the time elapsed between interviews (possibly) and the number of jobs reported in the employment calendar. |
Goode (2007);
Watson and Wooden (2004a) p27–28 |
| Calendar matching |
There is a two to six month overlap (or seam) in the activity calendar collected each wave. Of those who had at least one job in the calendar seam, 19 per cent provided job spell information that was inconsistent between wave 1 and wave 2. 1.8 per cent of the jobs matched within 1 month, 0.7 matched within 3 months, 2.1 matched but with an error of more than 3 months and 14.8 per cent had at least one job that could not be matched. |
|
| Comparison with external data |
| General |
Generally, the estimates are quite close for labour market, housing, demographic and health variables. |
Watson and Wooden (2002a) pp.24–26 |
| Income |
Compared to the ABS Survey of Income and Housing Costs, HILDA reports higher wages and salaries, and investment income. |
Watson and Wooden (2004a) pp. 17–21
Note income estimates in Watson and Wooden (2002a) (pp.24–26) are not imputed so not a fair comparison. |
| Wealth |
Comparison with ABS and RBA suggest HILDA slightly understates the volume of financial assets, is much closer to the RBA than the ABS for non financial assets, and is much lower (20 per cent) on debts than the ABS and RBA estimates. |
Watson and Wooden (2004a) pp.22–24 |
| Height and weight |
HILDA compares reasonably well with the ABS National Nutrition Survey but HILDA has a greater proportion of obese people but also lower item non–response. |
Wooden et al. (2008) |
Some more detailed information on the amount of missing income data and the extent of the income imputation is provided below.
Missing Income Data and the Extent of Income Imputation
The proportion of cases with missing income data are provided in Table 39. For most income variables, the proportion of missing income falls each wave. Part of the reason for the decline in the proportion of missing income may be because respondents are becoming more comfortable with the survey. For respondents with item non–response, the variables with the highest proportion of missing cases are still business income, investments and private transfers.
Table 40 shows how much of the mean income was imputed for each wave. For responding people with item non–response, 5.0 percent of total financial year income was imputed in wave 6, compared to 7.9 percent in wave 1. Including the imputed income totals for non-respondents within responding households (but excluding children), the percentage of total financial year income imputed for enumerated persons is 13.4 percent in wave 6.
This shows that while approximately one in ten responding persons are missing some component of financial year income in wave 6, only one twentieth of the mean income comes from imputed values and the remainder is from reported values. At the household level, one in five households is missing some component of financial year income and one seventh of the mean income is from imputed values.
Table 39: Proportion of cases with missing income data, waves 1 – 6
| Variable |
Wave |
| 1 |
2 |
3 |
4 |
5 |
6 |
| Responding Persons (non-zero cases only) |
| Current income (per week) |
| Wages and salaries (main job) |
4.6 |
3.1 |
2.8 |
2.7 |
2.4 |
2.2 |
| Wages and salaries (other jobs) |
16.7 |
13.9 |
13.2 |
13.0 |
12.9 |
11.1 |
|
3.2 |
2.0 |
2.0 |
1.8 |
1.6 |
1.0 |
| Financial year income |
| Wages and salaries |
7.9 |
6.9 |
5.5 |
3.8 |
4.5 |
4.6 |
| Aust govt pensions |
2.1 |
2.1 |
1.3 |
2.0 |
1.4 |
1.0 |
| Foreign govt pensions |
0.5 |
2.7 |
0.0 |
0.5 |
2.4 |
0.5 |
| Business income |
29.1 |
28.6 |
27.4 |
19.4 |
21.7 |
18.6 |
| Investments |
| Interest income |
19.5 |
18.6 |
13.9 |
11.0 |
11.3 |
12.8 |
| Dividends and royalties |
14.6 |
14.5 |
11.9 |
9.2 |
10.2 |
11.3 |
| Rent income |
20.3 |
14.7 |
14.9 |
11.3 |
10.5 |
10.3 |
| Private pensions |
6.3 |
4.7 |
3.3 |
4.1 |
4.9 |
3.9 |
| Private transfers |
8.0 |
23.1 |
15.8 |
14.4 |
21.2 |
13.4 |
| Total FY income |
15.7 |
14.9 |
12.1 |
9.6 |
10.7 |
10.3 |
| Windfall income |
| Windfall income |
4.0 |
2.8 |
3.2 |
2.7 |
2.1 |
4.6 |
| Enumerated Persons (zero and non–zero cases, excluding children) |
| Current income (per week) |
| Wages and salaries (main job) |
10.0 |
8.6 |
7.9 |
8.3 |
7.3 |
7.0 |
| Wages and salaries (other jobs) |
8.4 |
7.6 |
7.0 |
7.5 |
6.6 |
6.3 |
| Benefits |
8.6 |
7.5 |
7.0 |
7.3 |
6.4 |
6.1 |
| Financial year income |
| Wages and salaries |
12.1 |
10.9 |
9.6 |
9.0 |
8.7 |
8.6 |
| Aust govt pensions |
8.3 |
7.7 |
6.8 |
7.5 |
6.4 |
6.1 |
| Foreign govt pensions |
7.7 |
7.0 |
6.4 |
6.9 |
6.0 |
5.8 |
| Business income |
10.3 |
9.6 |
9.0 |
8.7 |
8.0 |
7.4 |
| Investments |
| Interest income |
12.0 |
11.2 |
9.5 |
9.3 |
8.6 |
8.9 |
| Dividends and Royalties |
11.5 |
10.7 |
9.4 |
9.0 |
8.4 |
8.4 |
| Rent income |
9.2 |
8.3 |
7.8 |
7.8 |
6.9 |
6.8 |
| Private pensions |
8.0 |
7.3 |
6.6 |
7.1 |
6.3 |
6.1 |
| Private transfers |
7.9 |
7.6 |
7.0 |
7.3 |
6.8 |
6.2 |
| Total FY income |
21.3 |
20.1 |
17.2 |
15.3 |
15.5 |
15.1 |
| Windfall income |
| Windfall income |
7.9 |
7.2 |
6.7 |
7.1 |
6.2 |
6.2 |
| Households (zero and non–zero cases) |
| Current income (per week) |
| Wages and salaries (main job) |
14.2 |
12.3 |
11.2 |
12.2 |
10.9 |
10.4 |
| Wages and salaries (other jobs) |
11.9 |
10.8 |
10.0 |
10.8 |
10.0 |
9.3 |
| Benefits |
12.1 |
10.6 |
9.9 |
10.6 |
9.6 |
8.9 |
| Financial year income |
| Wages and salaries |
17.0 |
15.7 |
13.8 |
13.0 |
12.8 |
12.8 |
| Aust govt pensions |
11.6 |
10.8 |
9.6 |
10.7 |
9.5 |
8.9 |
| Foreign govt pensions |
10.6 |
9.8 |
8.9 |
9.8 |
8.9 |
8.4 |
| Business income |
14.4 |
13.3 |
12.7 |
12.3 |
11.7 |
10.7 |
| Investments |
21.2 |
19.8 |
16.9 |
16.2 |
15.6 |
16.0 |
| Private pensions |
11.3 |
10.2 |
9.3 |
10.2 |
9.4 |
8.8 |
| Private transfers |
10.9 |
10.8 |
9.8 |
10.6 |
10.0 |
9.1 |
| Total FY income |
29.4 |
28.0 |
24.0 |
21.8 |
22.3 |
21.5 |
| Windfall income |
| Windfall income |
10.9 |
10.0 |
9.3 |
10.2 |
9.1 |
9.1 |
Table 40: Mean financial year income ($) (including imputed values) and proportion of mean income imputed, waves 1 – 6 (weighted)
| Variable |
Wave |
| 1 |
2 |
3 |
4 |
5 |
6 |
| Responding persons |
| Wages and salaries |
| Mean |
20,580 |
21,058 |
21,791 |
22,707 |
24,333 |
26,297 |
| Proportion Imputed |
5.9 |
4.6 |
3.7 |
2.7 |
3.2 |
3.5 |
| Total income |
| Mean |
27,317 |
28,335 |
29,204 |
30,680 |
33,025 |
35,672 |
| Proportion Imputed |
7.9 |
6.9 |
5.9 |
4.5 |
5.0 |
5.0 |
| Enumerated persons |
| Wages and salaries |
| Mean |
21,079 |
21,689 |
22,307 |
23,055 |
24,805 |
26,850 |
| Proportion Imputed |
15.1 |
14.9 |
13.8 |
12.8 |
12.3 |
12.1 |
| Total income |
| Mean |
27,836 |
28,958 |
29,803 |
31,072 |
33,497 |
36,276 |
| Proportion Imputed |
16.2 |
16.4 |
15.6 |
14.0 |
14.1 |
13.4 |
| Households |
| Wages and salaries |
| Mean |
42,368 |
43,471 |
44,776 |
46,404 |
49,874 |
53,934 |
| Proportion Imputed |
15.1 |
14.9 |
13.8 |
12.8 |
12.3 |
12.1 |
| Total household income |
| Mean |
55,949 |
58,042 |
59,821 |
62,541 |
67,350 |
72,868 |
| Proportion Imputed |
16.2 |
16.4 |
15.6 |
14.0 |
14.1 |
13.4 |
Missing Wealth Data and the Extent of Wealth Imputation
The proportion of cases with missing wealth data are provided in Table 41. This table has two columns for each wave to highlight the proportion of respondents who answered the wealth question with a wealth band.38 Wealth bands are strictly adhered to in the imputation of any wealth value (that is the imputed value must fall within the reported band) and greatly improve the quality of imputation. Treating cases where a wealth band is available as missing unfairly over represents the missingness problem so both situations have been provided. Missing cases for responding person and household level wealth items are reported as a proportion of non–zero cases only to more clearly show how much of a problem missing data is. However, not all missing cases required a non–zero impute (most cases do but for some it is unknown if they have the asset or debt and they can receive a zero impute) so the proportions give are a slight overestimation.
For most wealth variables, the proportion of missing income falls between wave 2 and wave 6. Part of the reason for the decline in the proportion of missing wealth may be because respondents are becoming more comfortable with the survey. In some situations where a wealth band option has been introduced, or an existing wealth band has been continued, there has been an increased proportion of missing values (when counting the wealth band as missing data). For respondents with item non–response, the variables with the highest proportion of missing cases are superannuation for retirees and those not retired. At the household level the largest amount of missingness is for trust funds, life insurance, business debt and business value. Each of the household level items are for situations where only a small amount of households actually have the asset or debt so the actual number of cases to be imputed is quite small.
Treating wealth band information as a response, nearly 39 percent of wave 2 households have some component of net worth missing. In wave 6 this proportion has dropped to 29 percent.
Table 42 shows the proportion of cases with missing home value which has generally declined over time.
Table 43 and Table 44 give the weighted mean wealth value (including imputed values) along with what proportion of the mean is attributed to imputed values. For all of the household wealth totals presented in Table 43, there has been a drop in the proportion imputed between wave 2 and wave 6. Home value (in Table 44) showed a small decrease in how much the mean was imputed between waves 2 to 5.
Comparing the table of missing values against the weighted means show that despite nearly 45 percent of households in wave 6 missing some component of net worth only 9.1 percent of the mean net worth wealth value was contributed by imputation.
Table 41: Proportion of cases with missing wealth data including and excluding wealth band responses, waves 2 and 6
| Variable |
Wave 2 |
Wave 6 |
| inc. bands |
excl. bands |
inc. bands |
excl. bands |
| Responding Persons (non–zero cases only) |
| Joint bank accounts |
9.8 |
– |
6 |
– |
| Own bank accounts |
5 |
– |
3 |
– |
| Superannuation, retirees |
20.1 |
– |
19.7 |
12.2 |
| Superannuation, not retired |
17.3 |
10.7 |
27.5 |
13.6 |
| HECS debt |
11 |
– |
8 |
– |
| Joint credit card debt |
10.1 |
– |
7.6 |
– |
| Own credit card debt |
3.6 |
– |
3.1 |
– |
| Other Debt |
2.4 |
– |
1.8 |
– |
| Enumerated persons (zero and non–zero cases) |
| Joint bank accounts |
11.3 |
– |
8.4 |
– |
| Own bank accounts |
9.8 |
– |
7.9 |
– |
| Superannuation, retirees |
8.0 |
– |
7.0 |
7.0 |
| Superannuation, not retired |
17.1 |
13.3 |
23.0 |
14.4 |
| HECS debt |
7.8 |
– |
6.4 |
– |
| Joint credit card debt |
7.7 |
– |
6.2 |
– |
| Own credit card debt |
8 |
– |
6 |
– |
| Other Debt |
7.5 |
– |
6.2 |
– |
| Household wealth items (non–zero cases only) |
| Children’s bank accounts |
6.0 |
– |
5.0 |
– |
| Business value |
20.1 |
– |
17.5 |
7.8 |
| Cash investments |
11.6 |
– |
12.3 |
7.1 |
| Equity investments |
15.3 |
– |
13.3 |
4.4 |
| Collectibles |
14.0 |
– |
15.1 |
8.1 |
| Other property value |
4.6 |
– |
0.5 |
– |
| Life insurance |
24.9 |
– |
28.5 |
16.9 |
| Trust funds |
35.7 |
– |
35.8 |
26.7 |
| Vehicles: Value |
2.3 |
– |
1.5 |
– |
| Business debt |
22.9 |
– |
11.6 |
8.1 |
| Home Value |
7.6 |
– |
4.2 |
– |
| Home: All debt |
5.4 |
– |
4.2 |
– |
| Other property: Debt |
7.1 |
– |
5.9 |
– |
| Overdue bills: Debt |
– |
– |
2.2 |
– |
| Household totals (zero and non–zero cases) |
| Financial Assets |
36.3 |
31.6 |
40.6 |
24.7 |
| Non–Financial Assets |
10.9 |
– |
7.5 |
5.3 |
| Total Assets |
41.0 |
36.6 |
43.8 |
27.5 |
| Financial Liabilities |
15.1 |
– |
12.3 |
12.2 |
| Net Worth |
43.0 |
38.9 |
44.9 |
29.4 |
Table 42: Proportion of households with missing home value data, waves 1 – 6
| Variable |
Wave |
| 1 |
2 |
3 |
4 |
5 |
6 |
| Home value (households) (non–zero cases only) |
| Home value |
5.9 |
7.6 |
5.6 |
4.0 |
3.3 |
4.2 |
Table 43: Mean wealth value ($) (including imputed values) and proportion of mean value imputed, waves 2 and 6 (weighted)
| Variable |
Wave |
| 2 |
6 |
| Households |
| Financial assets |
| Mean |
152,081 |
219,158 |
| Proportion Imputed |
18.8 |
18.5 |
| Non–financial assets |
| Mean |
315,398 |
507,143 |
| Proportion Imputed |
7.8 |
4.3 |
| Total Assets |
| Mean |
467,473 |
726,301 |
| Proportion Imputed |
11.0 |
9.0 |
| Total Liabilities |
| Mean |
65,485 |
113,661 |
| Proportion Imputed |
6.0 |
6.0 |
| Net Worth |
| Mean |
401,980 |
612,640 |
| Proportion Imputed |
12.3 |
9.0 |
Table 44: Mean home value ($) (including imputed values) and proportion of mean value imputed, waves 1 – 6 (weighted)
| Variable |
Wave |
| 1 |
2 |
3 |
4 |
5 |
6 |
| Households |
| Home Value |
| Mean |
200,168 |
232,206 |
271,135 |
298,830 |
315,108 |
345,298 |
| Proportion Imputed |
4.9 |
7.3 |
5.2 |
3.3 |
3.2 |
4.0 |
Endnotes:
| 38 |
A wealth band is two values which the respondent believes their actual value to be within. The bands differ between some variables. Back to where you were |
|
top of page
![The University of Melbourne [logo]](/hilda/template-assets-custom/images/custom_logo_crest_comb.gif){kind=logo}
