9208.0 - Survey of Motor Vehicle Use, Australia, Jul 1998  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 18/02/2000   
   Page tools: Print Print Page Print all pages in this productPrint All

INTRODUCTION

Significant changes in the collection and estimation methodologies were introduced for the 1998 Survey of Motor Vehicle Use (SMVU). These were mainly designed to reduce recall bias, i.e. the error that occurs when people cannot remember whether or when events of a given type occurred, resulting in omitting events, incorrectly placing events in time or reporting events that never took place. This paper discusses the impact of these changes and provides a summary of data quality issues affecting the results of the new methodology. The first section provides information on how the changes to methodology have affected comparisons of results of the current SMVU with those of previous collections. The second section discusses factors related to the data quality of the SMVU.

HISTORICAL COMPARISONS

Due to changes to the methodology, results may differ from those produced from earlier surveys. Users are cautioned against making detailed direct comparisons between the 1998 survey results and those produced from previous surveys.

Factors were produced which may be applied to the 1995 statistics on total distance travelled by type of vehicle to remove the estimated recall bias from those data. The factors were first published in the information paper, Motor Vehicle Use, Australia (Cat. no. 9219.0) and were based on estimated total distance travelled obtained from special surveys conducted for the 12 months ended 30 September 1996.

The factors could only be reliably estimated for total distance travelled by main vehicle types at the Australian level and will not be appropriate to lower level detail, if recall biases vary at this more detailed level. Care should also be taken in applying factors to data from surveys conducted prior to 1995 because while it is recognised that the recall problem was inherent in all previous surveys, there is no reliable measure of how recall biases have varied over time.

The estimated recall bias factors for total kilometres travelled by vehicle type are as follows:-


1. ESTIMATED FACTORS FOR TOTAL KILOMETRES TRAVELLED - 12 MONTHS ENDED 30 SEPTEMBER 1996
Estimated factor
RSE %

    Passenger vehicles (a)
0.91
5
    Motor cycles (b)
0.71
11
    Light commercial vehicles
1.01
7
    Rigid trucks
1.01
5
    Articulated trucks
0.95
5
    Non-freight carrying trucks
0.82
20
    Buses (a)
1.12
6
    Total (a)
0.94
4

(a) The estimated factor is statistically significant at the 10% level.
(b) The estimated factor is statistically significant at the 5% level.



The factors shown above were applied to the preliminary estimates of total distance travelled by type of vehicle for the 12 months ended 30 September 1995. Table 2 shows adjusted statistics with the estimated recall bias removed.



2. TOTAL KILOMETRES TRAVELLED BY TYPE OF VEHICLE - 12 MONTHS ENDED 30 SEPTEMBER 1995

Preliminary estimates
million kilometres
Preliminary estimates
RSE %
Estimated factors
Adjusted estimate
million kilometres
Adjusted estimate
RSE %

    Passenger vehicles
123,691
3
(a)0.91
113,054
6
    Motor cycles
1,526
7
(b)0.71
1,076
13
    Light commercial vehicles
27,751
3
1.01
28,056
7
    Rigid trucks
6,725
2
1.01
6,772
6
    Articulated trucks
5,094
2
0.95
4,860
6
    Non-freight carrying trucks
249
9
0.82
204
22
    Buses
1,479
2
(a)1.12
1,661
6
    Total
166,514
2
(a)0.94
155,683
5

(a) The estimated factor is statistically significant at the 10% level.
(b) The estimated factor is statistically significant at the 5% level.




The following table provides estimates of the total distance travelled by type of vehicle in Australia compiled from the 1998 and previous surveys of motor vehicle use. For 1995, both preliminary estimates from the survey and adjusted data to exclude the estimated recall bias are shown.


3. TOTAL KILOMETRES TRAVELLED BY TYPE OF VEHICLE (a)

1979

million kilometres
1982

million kilometres
1985

million kilometres
1988

million kilometres
1991

million kilometres
1995(b)

million kilometres
1995(c)

million kilometres
1998

million kilometres

    Passenger vehicles
84,872
96,109
106,574
116,640
114,286
123,691
113,054
134,261
    Motor cycles
1,768
2,152
2,276
1,924
1,615
1,526
1,076
1,350
    Light commercial vehicles
15,928
16,951
20,121
21,982
22,814
27,751
28,056
24,958
    Rigid trucks
5,837
8,417
7,627
7,840
6,114
6,725
6,772
6,015
    Articulated trucks
2,607
3,000
3,588
3,836
3,959
5,094
4,860
4,921
    Non-freight carrying trucks
457
237
242
261
201
249
204
175
    Buses
n.a.
n.a.
n.a.
1,433
1,401
1,479
1,661
1,639
    Total
111,469
126,866
140,427
153,915
150,389
166,514
155,683
173,317

          (a) All data except 1998 are for the year ended 30 September 1998. Data for 1998 are for the year ended 31 July.
          (b) Preliminary estimates.
          (c) Adjusted to remove estimated recall bias.



When the factors are applied to 1995 survey data collected using the old recall methodology, the estimated total distance travelled by all motor vehicles is revised from 166,514 million kilometres to 155,683 million kilometres, a reduction of 7%. This decrease is mainly due to passenger vehicles, where the total distance travelled is revised from 123,691 million kilometres to 113,054 million kilometres.

It is not possible to obtain similar adjustments for data prior to 1995. As noted above, it should be recognised that the recall problem which the ABS has attempted to overcome by its new collection methodology for the 1998 survey was inherent in all surveys conducted up to 1995, however there is no reliable measure of how recall biases have varied over time. Taking into account earlier studies, it is reasonable to conclude that estimates of total and average distance travelled for passenger vehicles and total vehicles are also likely to have been overstated in surveys before 1995. Similar generalisations cannot be made for other vehicle types. In particular, for commercial vehicles, record keeping practices appear to have improved over time, with a consequent reduction in recall bias.

An historical comparison for total kilometres travelled by passenger vehicles and all vehicles, with the break in series at 1995, is shown in the following graph.



10. The following table provides estimates of the average distances travelled by type of vehicle in Australia. These have been derived by dividing the total distances travelled shown in Table 3 by the number of vehicles estimated to have been registered for use during the relevant year.

4. AVERAGE KILOMETRES TRAVELLED BY TYPE OF VEHICLE (a)
1979

'000
1982

'000
1985

'000
1988

'000
1991

'000
1995 (b)

'000
1995 (c)

'000
1998

'000

    Passenger vehicles
15.1
15.3
15.5
15.8
14.3
14.4
13.1
14.4
    Motor cycles
6.3
6.1
6.5
6.5
5.7
5.2
3.7
4.4
    Light commercial vehicles
17.0
16.9
17.7
18.6
16.9
17.7
17.9
16.3
    Rigid trucks
16.7
19.0
17.9
19.4
18.5
20.0
20.2
17.7
    Articulated trucks
59.3
64.4
72.3
78.7
76.0
87.9
83.9
83.7
    Non-freight carrying trucks
12.9
12.8
11.5
11.3
14.2
15.9
13.0
9.9
    Buses
n.a.
n.a.
n.a.
35.3
33.3
32.5
36.5
30.8
    Total
15.3
15.6
15.8
16.4
14.9
15.2
14.3
14.9

          (a) All data except 1998 are for the year ended 30 September. 1998 data are for the year ended 31 July.
          (b) Preliminary estimates.
          (c) Adjusted to remove estimated recall bias.



When the factors are applied to 1995 total kilometres data, the estimated average kilometres travelled by all motor vehicles in 1995 falls from 15,200 kilometres to 14,300 kilometres, a reduction of 6 per cent. Again, the decrease is mainly caused by passenger vehicles, where the average kilometres travelled is revised from 14,400 kilometres to 13,100 kilometres.

An historical comparison for average kilometres travelled by passenger vehicles and all vehicles, with the break in series at 1995, is shown in the following graph.



Besides average distances based on all registered vehicles referred to above, this publication also contains a number of more detailed average distance travelled tables where estimates relate only to vehicles with a particular use. As stated previously, recall bias adjustment factors are unavailable at the more detailed level. Users are cautioned against applying the same recall bias adjustments to the more detailed tables in order to make historical comparisons.

Furthermore, users should be aware of an additional complication affecting the comparability with previous surveys of the averages per vehicle in tables 3 to 12 and 15 to 17. Because vehicle use throughout the year is not constant, having a shorter reporting period of three months means that the proportion of vehicles which report a particular type of usage will be lower than if the reporting period had been 12 months. Therefore, the estimate of the number of vehicles having a particular type of use in the 12 months will be lower and since the denominator is lower, the average distance travelled for vehicles for the 12 months will be slightly higher than under the old method.