FACT SHEET – October 2014
introduction of MOBILE PHONE INTERVIEWING to the NATIONAL VISITOR SURVEY
Background
The National Visitor Survey (NVS) commenced in 1998. Between 1998 and 2013, all interviews were undertaken with respondents on residential fixed line telephones using Computer Assisted Telephone Interviewing (CATI).
The mobile only population has increased substantially in recent years, particularly for younger age groups. Latest estimates indicate that 33% of those aged 15–24 years and 48% of those aged 25–34 years are mobile only.
Consequently, Tourism Research Australia (TRA) introduced mobile phone interviewing as of 2 January 2014 to provide better coverage of the Australian population. This has resulted in a break in series because the travel patterns of people interviewed on mobile phones differ from those of people interviewed on residential fixed lines and these cannot be back-cast reliably by estimation. This means that care should be taken when comparing 2014 NVS estimates with data from previous years.
The early 2014 survey results indicate that data for earlier periods is likely to have under-estimated domestic overnight travel, in particular for overnight trips. This is of particular relevance to the period 2008 to 2013 when growth in mobile phone ownership in Australia was very strong.
HOW DOES THE NEW METHODOLOGY WORK?
The NVS will continue with its annual sample of 120,000. For 2014, this is split evenly between fixed line (50%) and mobile phone (50%) sample, resulting in three phone ownership populations:
§ Fixed line only (reached through the fixed line sample)
§ Mobile phone only (reached through the mobile phone sample)
§ Fixed line plus mobile phone (reached through either sample segments).
The NVS weighting process has been modified to account for coverage of persons with fixed lines and/or mobile phones. For more information on the weighting used under the new NVS methodology, refer to http://www.tra.gov.au/documents/NVS_weighting_technical_fact_sheet.docx.
HOW HAS THE NVS SAMPLE PROFILE CHANGED?
The introduction of mobile phone interviewing provides a sample distribution by age that aligns more closely with that of the Australian population aged 15 years or more. Previously, there was considerable under-coverage of those aged 15–34 years (who are more likely to be mobile only) and over-coverage of people aged 55 years and over (who are more likely to live in fixed line households) (Table 1).
Overall, those interviewed on mobile phones (compared to those interviewed on fixed lines) are more likely to be:
§ aged 15 to 34 years (42% compared with 10%) and less likely to be aged 65 years or more (10% compared with 39%).
§ employed full-time (48% compared with 29%).and less likely to be retired (14% compared with 42%).
§ male (54% compared with 46%).
Table 1: NVS sample profile by age, January to June 2013 and 2014Age group / NVS Jan-June 2013 / NVS Jan-June 2014 / Population
Sample / % / Sample / % / %
15–24 / 4,314 / 7 / 7,271 / 12 / 16.3
25–34 / 4,202 / 7 / 8,095 / 14 / 18.1
35–44 / 8,729 / 15 / 8,534 / 14 / 17.0
45–54 / 10,583 / 18 / 9,871 / 17 / 16.3
55–64 / 12,456 / 21 / 10,748 / 18 / 14.1
65 or more / 18,897 / 32 / 14,632 / 25 / 18.1
Total / 59,181 / 100 / 59,151 / 100 / 100
The introduction of mobile phone interviewing has affected the distribution of the NVS sample, as it is not possible to apply geographic quotas to the mobile sample at this time. However, it is likely that in the future a geographic identifier (such as a postcode) will be attached to mobile numbers to help with region identification.
HOW iS tra quantifying the impact OF MOBILE PHONE INTERVIEWING on survey estimates?
NVS estimates published from 2014 onwards are based on the new methodology (i.e. compiled using survey responses from the fixed line plus mobile phone sample). However, some top-line fixed line only estimates have been derived for comparison purposes.
While the fixed line only estimates are constructed from half the sample collected previously and thus are subject to higher sample error, they provide a useful indicator of what the NVS estimates would have been had mobile phone interviewing not been introduced. The difference between the estimates published by TRA in 2014 and the fixed line only sample estimates will help to quantify the impact from introducing mobile phone interviewing.
how much can be learnt from EARLY 2014 results?
Quarterly NVS survey estimates are subject to greater sample error than rolling annual (year ending) NVS survey estimates. Given this, early quarterly data for 2014 survey results should be viewed as only a useful early indicator of likely broad impacts from the introduction of mobile phone interviewing. A clear picture of the extent to which NVS survey estimates are impacted and how this effect varies for key national, state and regional tourism metrics will take at least 12 months to be known.
broad impacts evident for January to June 2014
key tourism metrics
With the different demographic profile of those interviewed on mobile phones and fixed lines, it is not surprising that January to June 2014 data indicate that the travel patterns of those interviewed on mobile phones and fixed lines are also different.
§ People interviewed on mobile phones reported more overnight trips than those interviewed on fixed lines, including those aged 65 years and over.
§ Overnight trips taken by people interviewed on mobile phones were—on average—shorter than those of people interviewed on fixed lines.
§ With trip length shorter, overnight expenditure per trip was lower but expenditure per night nonetheless higher for people interviewed on mobile phones than fixed lines.
§ There was little difference in the day trip rate between people interviewed on mobile phones and those interviewed on fixed lines. The average trip expenditure for a day trip appears to be slightly higher for landline users; however, this pattern cannot be determined until a full year of data is available due to seasonal patterns around business travel.
§ People interviewed on mobile phones were almost twice as likely to report an outbound trip as those interviewed on fixed lines.
Table 2: Key tourism metrics, year ending June 2014
Metric / Fixed line only / Published / Difference (%)Overnight trips (million) / 76.3 / 79.1 / 3.7
Visitor nights (million) / 292.2 / 295.0 / 1.0
Overnight trip expenditure ($ billion) / 52.8 / 53.3 / 1.0
Average nights (number) / 3.8 / 3.7 / -2.7
Average expenditure per trip ($) / 691.9 / 673.8 / -2.6
Average expenditure per night ($) / 180.6 / 180.7 / 0.1
Day trips (million) / 164.8 / 165.2 / 0.3
Day trip expenditure ($ billion) / 18.1 / 18.2 / 0.5
Average expenditure per trip ($) / 110.1 / 110.3 / 0.2
differences in travel characteristics
Many of the differences in travel characteristics between those interviewed on mobile phones and fixed lines can be explained by the higher propensity for mobile phone ownership in the younger age groups. Below are some of the early indications of changes in travel characteristics evident in the early 2014 data.
§ People on mobile phones are more likely to be taking overnight trips to visit friends and relatives and slightly more likely to be conducting business. Landline phone respondents are more likely to have taken a holiday. This is consistent with previous trends showing no growth in overnight domestic holidays for persons aged less than 45 years (highest mobile phone ownership group), but large increases in outbound holidays. The increasing trend of growth in those travelling to visit friends and relatives (VFR) has continued, while business has recovered from the effects of the global financial crisis.
§ People on mobile phones have shown higher participation in active and outdoor activities and had participated in more activities generally (10%).
§ Early data shows more spend is occurring on items such as alcohol, take away meals, shopping, entertainment and local transport.
§ Mobile phone owners are more likely to be back packing.
§ Mobile phone owners are three times more likely to be Fly-in/Fly/out (FIFO); 1.5% compared to 0.5% (total 1%).
fly-in fly-out (fifo) travel
The NVS has always reported some FIFO travel, but this has been under-estimated for two reasons:
· The FIFO population is very mobile and not easily reached on residential fixed lines.
· An indeterminable number of FIFO trips were previously classified as out-of-scope by NVS interviewers because they involved travel between two places of usual residence.
Given the potential for FIFO travel to crowd out leisure travel in many parts of Australia, TRA has included three new questions in the NVS in 2014 to better identify FIFO workers and distinguish between their work and non-work related travel. While it will take time to assess how effective these questions have been, early data for 2014 indicate an increase in FIFO travel with the majority reported by people interviewed on mobile phones.
is there an impact on year ending june 2014 survey estimates?
There is an impact on year ending June 2014 survey estimates from the introduction of mobile phone interviewing, but this is diluted by the presence of data for the September and December 2013 quarters collected from fixed line interviews only. The impact on year ending estimates will build as September and December quarter 2014 data progressively replace data for the same quarters in 2013. Year ending December 2014 survey estimates will provide the first annual data where the impact of mobile phone interviewing is fully integrated.
has mobile phone interviewing IMPROVED THE accuracy OF SURVEY ESTIMATES?
The introduction of mobile phone interviewing has improved the age and gender distribution of the NVS sample. This has contributed to a more evenly distributed spread of weights being applied to the sample than before. For further information on weighting used under the new NVS methodology, refer to http://www.tra.gov.au/documents/NVS_weighting_technical_fact_sheet.docx.
The early data from 2014 indicates that the new methodology has reduced the relative standard errors (RSEs) of NVS estimates and this will contribute to an improvement in survey accuracy and stability. TRA will be recalculating and publishing the RSEs attached to different sized NVS estimates collected under the new survey methodology in conjunction with the December 2014 NVS release.
will tra be continuing to assesS the impact of mobile phone interviewing on the nvs?
As data for September and December quarters 2014 are processed, TRA will continue to analyse the impact of mobile phone interviewing, and consequently update this fact sheet with each release.
By the December 2014 NVS release, TRA expects to have a much clearer picture of the extent of the impact and how this varies across a broad range of data items. While this will not eliminate the break in series, the information will enable informed and appropriate comparison of NVS data collected under the current and previous survey methodology.
is further change to the nvs sample methodology likely?
TRA monitoring of phone ownership patterns will be on-going. With mobile phone ownership continuing to grow and the number of fixed line only households contracting quite quickly, it is likely that there will be increases to the current 50% share of sample interviewed on mobile phones in future years. It is also possible that the distribution of NVS sample by origin area will be adjusted at some time in the future.