RDP 2017-04: How Australians Pay: Evidence from the 2016 Consumer Payments Survey Appendix A: Survey Methodology

Mary-Alice Doyle, Chay Fisher, Ed Tellez and Anirudh Yadav

July 2017

The fieldwork for the 2016 CPS was conducted by the research firm Ipsos on behalf of the Bank in November 2016. The survey consisted of three parts: a pre-diary questionnaire about the demographic characteristics of respondents; a seven-day payments diary; and a post-survey questionnaire about respondents' automatic payment arrangements and their preferences and attitudes about different payment methods. To encourage participation and engagement with the survey, respondents received a gift card on completion of the three components.

The survey was delivered online for most respondents but to ensure the sample was broadly representative of the Australian population, participants without internet access were recruited by telephone to complete a paper-based survey. The overall response rate was good, resulting in a final sample of 1,510 respondents, of which 1,388 completed the survey online and 122 completed the paper-based survey (Table A1). Respondents recorded a total of around 19,500 transactions in the survey, comprising day-to-day payments, transfers to family or friends, automatic payments and cash top-ups (Table A2).

Table A1: 2016 CPS Response Rates
Per cent of payments
  Number recruited Number of completed
responses		 Response rate
(%)
Online respondents 1,972 1,388 70
Offline respondents 264 122 46
Total 2,210 1,510 68

Source: RBA calculations, based on data from Ipsos
Table A2: Total Transactions Recorded in 2016 CPS
  Number Value
($)
Day-to-day payments 16,838 1,381,275
Transfers to family or friends 346 81,205
Automatic payments 1,174 204,214
Cash top-ups 1,108 224,436
Total 19,466 1,891,130

Source: RBA calculations, based on data from Ipsos

A.1 Survey Instruments

Pre-diary questionnaire

The demographic information collected in the pre-diary questionnaire was mostly the same as that collected in 2013. Demographic variables included age, sex, personal and household income, family status, household size, location (capital city or rest-of-state), employment status, occupation, and education level. A full list of debit and credit cards held by the respondent was collected, with the respondent also identifying their primary debit and credit card. Respondents also provided information about how they usually paid off their credit card debt (i.e. whether they paid off their debt every month or whether they let part of the balance roll over from month to month).

Payments diary

The payments diary was very similar to that used in 2013 to ensure comparability of data across surveys. In the diary, respondents recorded details about every transaction they made for a week, excluding automatic payments (which were recorded in the post-survey questionnaire). These details included the value, payment method, channel (e.g. online or in-person) and type of merchant. For card transactions, online respondents selected the specific card they used (from the list of cards they provided in the pre-diary questionnaire), and were asked to indicate whether they inserted the card into the reader or tapped/waved a physical card or their mobile phone over the reader. Respondents also recorded the dollar value or percentage amount of any card surcharges that they paid. Along with their payments, respondents were asked to include details of cash top-ups, including the value of any ATM fee paid (rather than indicating whether or not they paid an ATM fee, as in 2013). The full list of fields used in the 2016 diary is set out in in Table A3.

As part of the fieldwork for the 2016 CPS, Ipsos recruited a separate sample of 299 respondents to complete an online three-day diary instead of the week-long diary. Responses to the three-day diary were not included in any of the results presented in this paper, but will be used by the Bank to evaluate the case for shortening the duration of the diary in future surveys.[34]

Table A3: Fields in the 2016 Payments Diary
Payments
Date Payment purpose:
Day-of-week 1 – Supermarket/bottle shop
Payment amount 2 – Small food store
Card surcharge paid (dollar/per cent amount) 3 – Electrical/furniture
Payment method: 4 – Other retailer
1 – Cash 5 – Takeaway/fast-food
2 – Debit/credit card(a) 6 – Café/restaurant
3 – Personal cheque 7 – Pub/bar
4 – BPAY 8 – Petrol/service station
5 – Bank transfer 9 – Transport
6 – PayPal 10 – Leisure/sports/entertainment
7 – Gift/prepaid card 11 – Holiday travel
8 – Other 12 – Household bills (paid at post office)
Card action: 13 – Household bills (not paid at post office)
1 – Tap/wave card on or near card reader 14 – Post office (excluding household bills)
2 – Tap/wave mobile phone on or near card reader 15 – Medical/health
3 – Insert card and press ‘CR’ button 16 – Services
4 – Insert card and press ‘CHQ’/‘SAV’ button 17 – Transfer to family member or friend
Payment channel: 18 – Transfer within own accounts
1 – In person 19 – Cash deposit
2 – Internet (PC/tablet) 20 – Other
3 – Internet (mobile phone)  
4 – Telephone call  
5 – Mail  
Cash top-ups
Date Source of cash:
Day-of-week 1 – ATM
Cash top-up amount 2 – eftpos cash-out
ATM fee paid (dollar amount) 3 – Over the counter at a bank branch
Total value of banknotes in wallet after top-up 4 – Other
Notes: (a) Online respondents selected the specific card they used (from the list of cards they provided in the pre-diary questionnaire); offline respondents selected from: debit card; MasterCard/Visa credit card; and American Express/Diners Club

Post-survey questionnaire

In the post-survey questionnaire, respondents were asked to record details of any automatic payments that occurred during the diary week, referring to their latest bank statements. This question was worded slightly differently from previous waves, to include both ‘pull’ payments (direct debits), and ‘push’ payments (recurring ‘pay-anyone’ payments set up by the respondent). In previous waves, the survey has only asked respondents to record direct debits.

In 2016, qualitative questions in the post-survey questionnaire focused on consumers' preferences and attitudes towards different payment methods. Questions from the 2013 survey regarding attitudes to the use of different payment methods and the use of cash and cheques, and response to a hypothetical surcharge were included. Questions on use and attitudes towards mobile wallets were included for the first time, as were a range of questions on consumers' use, repayment habits and holding decisions for their credit cards.

A.2 Survey Sample and Weighting

The overall sampling process for the 2016 CPS was similar to that used for the 2013 survey. To ensure the survey sample was broadly representative of the Australian population, there were recruitment targets covering key demographic variables: age, sex, household income, location (i.e. capital city or regional area), credit card ownership, and household internet access. Recruitment targets for most demographic variables were based on data from the Australian Bureau of Statistics (ABS); data on credit card ownership were obtained from the Household, Income and Labour Dynamics In Australia (HILDA) Survey. Ipsos recruited most online participants from its proprietary online panel, while offline respondents were recruited via random digit dialling. Recruitment for the survey commenced on 10 November 2016, with the final dataset comprising responses collected between 14–30 November.

Due to different response rates across the demographic categories for which recruitment targets were set, we constructed survey weights so that the final (weighted) dataset aligns with population benchmarks. Weighting ensures that the survey data can be used to make inferences about the Australian population. We used the iterative proportional fitting procedure, or raking, to calibrate the survey weights, which is common in large social science surveys. The same procedure was used for the 2013 CPS and has been used for payment surveys in Canada and the United States (Henry, Huynh and Shen 2015; Angrisani, Foster and Hitczenko 2016). We implemented the raking algorithm using the ‘ipfweight’ program in Stata 13. As a cross-check, we also calibrated weights using the ‘survey’ package in R, which produced nearly identical results. Table A4 presents the unweighted sample distribution for selected demographic variables, alongside the population distribution and the mean weight for respondents in each group.

Table A4: 2016 CPS Sample Distribution and Mean Weights
  Unweighted sample
proportion (%)		 Population proportion
(%)		 Mean weight
Age(a)
18–24 9 12 1.41
25–34 17 19 1.11
35–44 16 18 1.10
45–54 14 17 1.21
55–65 17 15 0.85
65+ 27 19 0.73
Gender(a)
Female 53 51 0.96
Male 47 49 1.04
Location(a)
Regional 33 33 1.01
Capital city 67 67 1.00
Household income quartile(a,b)
1st 29 25 0.87
2nd 32 23 0.72
3rd 25 26 1.02
4th 14 26 1.88
Credit card ownership(c)
Yes 72 54 0.76
No 28 46 1.62

Notes: (a) Population proportion based on data from ABS
(b) In the CPS, respondents were asked to select their household income from a range (e.g. under $7,799, $7,800–$19,999, $20,000–$29,000, etc); to ensure comparability with these ranges, we derive our estimates of population household income quartiles from ABS Cat No 6523.0 Table 1.3, where income is reported in similar ranges; accordingly, the constructed population income quartiles differ slightly from 25 per cent each
(c) Population proportion based on data from HILDA Survey Release 14.0

Source: RBA calculations, based on data from ABS, HILDA Survey Release 14.0 and Ipsos

Footnote

Central banks in other jurisdictions conduct similar payment surveys over various durations, including one day (e.g. Netherlands), three days (e.g. Canada and the United States) and one week (e.g. Germany). [34]