|Page tools: Print Page Print All RSS Search this Product|
8 In addition, to help correct for under coverage on the population frame, new articulated vehicles registered after 31 October 1999 were progressively added to the population frame during the enumeration period and supplementary samples drawn. Adjustments were also made to the estimation process to account for new and re-registered articulated vehicles not given a chance of selection in the survey.
9 In order to encourage record keeping, confirm ownership and to update contact details, owners of articulated vehicles selected in the survey received early advice about their inclusion in the survey. Respondents were contacted one month prior to the beginning of the reporting period for which their vehicle was selected. At this time respondents were asked to return a questionnaire reporting selected vehicle characteristics. Respondents were also advised that they would receive a follow up questionnaire during the next month seeking details about the use of the vehicle over a specific two week reporting period. Examples of the main items requested in the second questionnaire were included together with an information sheet about the survey.
10 When the questionnaires for each reporting period were returned to the ABS they were checked for completeness and accuracy and, where possible, follow-up contact was made with owners to resolve reporting problems. Missing data items were clerically imputed.
11 Where the selected vehicle owner had not owned the vehicle for the whole fortnightly survey period, the details provided for the period of ownership were adjusted to give a two week equivalent, except where the vehicle was deregistered, in which case only the use up to deregistration was included.
12 Estimates were produced for each of the 26 fortnightly reporting periods. These estimates were then aggregated to produce final estimates of freight movements for the survey reference period.
Rail, sea and air
13 Collection of rail, sea and air freight movements has been conducted as a census since June quarter 1994. Although the methodology used for the FMS 2000 does not differ significantly from that which was first adopted in 1994 for the 1994-95 Freight Movements Survey, some changes have occurred in the scope and coverage rules and in the data items collected.
14 Questionnaires were despatched quarterly to all rail and air freight operators and port authorities. Respondents were asked to provide information on the origin (rail and air only), destination, commodity type, weight and method of transportation. Major providers with large volumes of data supplied information on computer disk.
15 Data were aggregated to produce quarterly totals and these were further aggregated to produce annual figures relating to the survey reference period.
RELIABILITY OF ESTIMATES
16 Since estimates of freight moved by road are based on a sample survey, rather than a complete enumeration, the data are subject to sampling variability. That is, they may differ from results that would have been obtained had all articulated vehicles been included. For more information on sampling error, see Technical Note 1. As the rail, sea and air freight components are fully enumerated these data are not subject to sampling error.
17 Sampling error is not the only type of inaccuracy which affects data reliability. Other types of error, referred to as non-sampling error, can be present in any type of collection, whether it be a complete enumeration or a sample survey. For example, non-sampling error can occur because of non-response to the survey, imperfections in reporting by providers, definition or classification difficulties, or errors in transcribing and processing data. While the effects of non-sampling error are not quantifiable, every effort is made to minimise the impact through the design and testing of questionnaires and the use of efficient operating procedures.
18 A potentially important factor relating to non-sampling error is the response rate achieved. The response rates for each of the transport modes included in the survey were:
Response and non-response for road
19 Non-response to the road component of the FMS 2000 predominately occurred because the ABS was unable to trace the selected vehicle or the form was not able to be completed. This was despite the ABS making all reasonable efforts to maximise response rates. Where appropriate, mail reminders and telephone follow-up were used to attempt to contact initially non-responding vehicle owners.
20 A large non-response increases the potential for non-response bias. Non-response bias occurs if the usage patterns of the non-responding vehicles differ significantly from the usage patterns of the responding vehicles. For the FMS 2000, adjustments were applied at the stratum level to the 'weights' (the factors which expand the sample data to obtain estimates for the population) to allow for differing levels of non-responding vehicles.
E1 RESPONSE AND NON-RESPONSE FOR ROAD, BY CATEGORY
21 Due to the non-response level of 22%, there is a potential for non-response bias in the survey estimates. To gauge the impact of this potential bias, a 'what if' analysis was conducted on the survey data. The analysis considered what would happen to the estimates of total tonnes carried and total tonne-kilometres travelled if the non-respondent vehicles had a similar usage pattern to two categories of responding vehicles:
22 To estimate the low and high usage patterns, survey responses were ordered from lowest to highest for each of total tonnes and total tonne-kilometres. For each of the usage categories, different percentage cut off points were considered, from the lowest/highest 20% (the approximate non-response rate) to the lowest/highest 40% of usage. There was little difference between the estimates produced based on these percentage cut offs.
23 Whilst it is extremely unlikely that all non-responding vehicles would have similar usage patterns to the lowest/highest 40%, the table below compares the estimates which would have been obtained if either were the case. The actual estimates obtained, where non-respondents are estimated using the respondent mean, are included for comparative purposes.
E2 'WHAT IF' ANALYSIS FOR ROAD ESTIMATES
Imputation for road
24 The road component of the FMS 2000 measured freight carried by all articulated vehicles during the reference year. Because selections were taken from vehicles registered some time before the beginning of each collection period, adjustments were made to the survey estimates to account for the use of vehicles which were not in the population from which the survey sample was taken, but which were registered some time during the survey reference period. This involved two groups of vehicles:
25 At the Australian level, the adjustment for vehicles being re-registered accounted for approximately 2% of total tonne-kilometres, while the effect of adjusting for new vehicles was estimated to be 6% of total tonne-kilometres.
26 Newly registered vehicles were accounted for in two ways: the addition of newly registered articulated vehicles to the population frame from which the sample was drawn each fortnight, and imputation for vehicles that did not have a chance of selection.
27 A sample of new vehicles was added to the existing sample from the eighth fortnightly cycle of the FMS 2000. As a result of including this sample, the level of adjustment required for new articulated vehicles which were missing from the population was reduced for the remainder of the survey. The adjustment for vehicles not given a chance of selection was based on averages from the newer vehicles responding to the survey.
OTHER DATA QUALITY ISSUES
28 An investigation was conducted into the possible effect of under-reporting on the survey estimates for the road freight component. All 26 cycles were examined to determine if there was a difference between the number of trips reported in the second week when compared to the first week of the reporting period. As a result of the findings, adjustments were made to the estimates. Further detail on these adjustments can be found in Technical Note 2.
COMPARISON WITH SURVEY OF MOTOR VEHICLE USE
29 Survey estimates for road freight movements from the FMS 2000 and the ABS Survey of Motor Vehicle Use (SMVU) are not fully comparable. One principal difference relates to the survey reference periods; the FMS 2000 was conducted from 1 April 2000 to 31 March 2001, whereas the SMVU was conducted from 1 November 1999 to 31 October 2000.
30 Another major difference relates to the respective methods used in each survey to produce estimates for tonne-kilometres travelled.
31 FMS respondents are asked to record distance travelled and weight carried for each laden trip made during a fortnightly reporting period. The total tonne-kilometres travelled estimate is derived from summing the product of distance travelled by weight carried.
32 The SMVU estimate of total tonne-kilometres travelled is derived from the product of total laden business distance travelled by the average load carried per laden trip reported over the quarterly survey period.
ABS DATA AVAILABLE ON REQUEST
33 As well as the statistics included in this publication, the ABS has unpublished data available for a charge. Inquiries should be directed to the National Information and Referral Service on 1300 135 070.
RELATED PUBLICATIONS AND PRODUCTS
34 Users may also wish to refer to the following publications and products which contain information relating to motor vehicles in Australia:
Where figures have been rounded, discrepancies may occur between the sums of component items and totals. Unless otherwise stated, percentages are based on stated or classifiable responses.
These documents will be presented in a new window.