Introduction
-
time assigned to travel activities;
-
time assigned to non-travel activities, with the activities being subdivided into the categories of unconstrained activities (leisure) and constrained activities (such as personal care);
-
budget assigned to goods consumption being subdivided into constrained and unconstrained goods.
-
travel surveys including information about the characteristics and determinants of travel activities such as trip purpose, start and end time, duration, cost, transport modes, location of origin and destination;
-
time use surveys giving complete information about travel and non-travel activities throughout the day including the types of main and parallel activities, location, start and end time of each activity episode;
-
consumer expenditure surveys dealing with goods consumption and budget assignment in the short and long run.
Current state of survey practice
Travel surveys: the trip-based approach
Time use surveys: the activity-based approach
Comparison of trip-based and activity-based survey approaches
Consumer expenditure surveys
-
Expenditure diaries Respondents report all their actual expenditures for goods and services in diaries, usually over a period of 14 days. Few diaries exist with diary periods of 7 days, 1 or 2 months. Diaries are filled out either for individual persons or for the entire household. Self-administered paper diaries or online diaries are used.
-
Retrospective interviews, questionnaires In most countries respondents also report long-term and regular expenditures retrospectively for the last 1, 3, or 12 months. These expenses serve to correct the diary data for costs which do not occur in the diary period, and they ensure that also seasonal and one-time big-ticket items are included. This is indispensable for the calculation of the total consumer expenditures.
-
Part 1: Private garden or farming products for personal requirements. All home-made agricultural products harvested and consumed within the 14-day period had to be recorded in the diary.
-
Part 2: All expenses made on food and drinks including pet food and visiting a restaurant or cafe. A pre-defined categorization of food groups was used, so that all costs incurred could be stated by choosing the appropriate product and adding its amount and price. Product groups had a high level of detail (e.g. “wholemeal bread” and not just “bread”).
-
Part 3: All other expenses. For all other kinds of expenses two pages per day were provided offering one page with pre-defined categories of products (e.g. personal care, clothes, fuel) and an additional page with open text fields where respondents had to specify the purchased items.
COICOP main divisions | |
---|---|
01. Food and non-alcoholic beverages | 07. Transport |
02. Alcoholic beverages, tobacco and narcotics | 08. Communication |
03. Clothing and footwear | 09. Recreation and culture |
04. Housing, water, electricity, gas and other fuels | 10. Education |
05. Furnishings, household equipment and routine household maintenance | 11. Restaurants and hotels |
06. Health | 12. Miscellaneous goods and services |
[13. Not for private consumption]a |
The Mobility–Activity–Expenditure-Diary (MAED) design
Approach to integrating the three survey traditions
Activity-based approach |
Open time-intervals, pre-defined activity classification All tested designs used pre-defined activity categories and open time intervals—contrary to the HETUS guidelines, which recommend open activity description and pre-defined time intervals. The main ideas behind the open time intervals were (1) to ensure that all trips are reported with correct start and end times, also short trips of less than 10 min; and (2) to reduce the response burden, because our scheme avoids multiple recording of long activity sequences (sleep, work etc.). The idea behind the pre-defined activity categories was to indicate the requested level of detail for the reported activities to the respondents. This should help to reduce unwanted variation in the level of detail, which is the main argument against open time-intervals in the HETUS guidelines |
Accurate separation of travel activities and non-travel activities Literature shows that travel activities and non-travel activities tend to be mingled in activity-based questionnaires (Gerike et al. 2015). We thus tested different designs for motivating respondents to report travel-activities reliably and separately from non-travel activities |
Addresses of visited locations We provided sufficient space to report the complete address for the start and end location of each trip in all versions of the questionnaires. Complete addresses are essential to georeference the locations visited which is a required prerequisite to obtain information on non-selected alternatives in the mode choice models or to add spatial attributes such as the distance to the next public transport stop. The importance of complete and correct locations was additionally emphasised in the instructions |
Trip-based approach |
NKD-design Our main idea was to stick as close as possible to the NKD design because it has proven to be successful in reporting travel-related information, but to expand the ‘trip purpose’ section in order to retrieve more detailed information about non-travel activities. We tested different approaches in the pilot studies to include non-travel activities between the trips, before the first and after the last trip, but also for diary days without any trip |
Travel costs Questions about travel costs were included directly in the trip section of the diary from the second pilot study on (public transport ticket, parking ticket etc.) |
Consumer expenditure
|
Reduced level of detail in expenditure categories Consumer expenditure diaries are very detailed with fine subdivisions of product groups. The model of Jara-Diaz et al. (2008) requires first and foremost a distinction between constrained and freely chosen goods; further distinctions may improve the model, but the number of cost categories that the model can deal with is strictly limited. Our classification of reported expenses was based on the main COICOP divisions with 12 categories shown in Table 1. In pilot study 2, pre-defined expenditure categories were tested against an open description of expenses with post hoc classification by the survey team |
Travel costs as an exception According to our specific interest in travel, our final scheme includes more detailed questions on travel costs than usual travel diaries and expenditure diaries. Consumer expenditure surveys do not have special interest in travel; the COICOP division “Transport” is treated like any other product group |
Findings from pilot studies
Pilot study 1 | Pilot study 2 | |||||
---|---|---|---|---|---|---|
Households in % | Overall (n = 300) | TD (n = 100) | AD (n = 200) | Overall (n = 145) | TD (n = 49) | AD (n = 96) |
Recruitment phase
| ||||||
Gross sample size | 100 | 100 | 100 | 100 | 100 | 100 |
Not availablea | 35 | 35 | 35 | 29 | 30 | 29 |
Participation rejected | 27 | 33 | 24 | 42 | 43 | 42 |
Participation agreed: households received questionnaires | 38 (n = 114) | 32 (n = 32) | 41 (n = 82) | 28 (n = 41) | 27 (n = 13) | 29 (n = 28) |
Questionnaire phase
| ||||||
Households returned questionnairesb | 73 | 81 | 70 | 68 | 62 | 71 |
Net response of gross sample | 28 | 26 | 29 | 19 | 16 | 21 |
Pilot study 1 | Pilot study 2 | |||
---|---|---|---|---|
AD (nDiaries = 106; nTrips = 2669) | TD (nDiaries = 50; nTrips = 1329) | AD (nDiaries = 43; nTrips = 991) | TD (nDiaries = 20; nTrips = 487) | |
Occurrences within diaries in % | ||||
Mixture of trips and activities within one line | 35 | _ | 51 | _ |
More than one activity category were chosen | 21 | _ | 53 | 10 |
04:00 a.m.–04:00 p.m. scheme disregarded | 8 | _ | _ | _ |
Inconsistent durations | _ | 8 | _ | _ |
Problems with activities from 04:00 until first trip | _ | 4 | _ | _ |
Return trips missing, legs instead of trips | _ | 2 | 2 | _ |
Occurrences within trips in % | ||||
Missing modes | 4.2 | 1.7 | 0.2 | 0 |
χ2 = 17.04 (p < 0.0001) | G = 1.59 (p = 0.2062)b | |||
Missing addresses | 7.9 | 5.2 (n = 1664)a | 12.7 | 4.9 |
χ2 = 10.3 (p = 0.0013) | χ2 = 18.16 (p < 0.0001) |
The final MAED design
Survey procedure, response rates and incentives
-
People tend to use the same channel for their answer through which they have been contacted (BRAWISIMO 2015). If they receive a written announcement they prefer a written questionnaire.
-
There is some evidence that online questionnaires are filled out with less care and have more missing data, e.g. a high number of missing return trips in travel diaries (Kadan 2015).
-
The survey served as a first feasibility test of the MAED design. The implementation and administration of an additional web-based questionnaire would have been too expensive. In prospective surveys it is yet desirable to offer both a written and online channel to increase response rates.
-
Households with available telephone number (V1) were sent an announcement post card, which notified them that they will be called within the next few days; they were not asked to reply to the announcement.
-
Households without available telephone number (V2) were sent a folding card, which informed them of the survey and the incentive in case of participation. They were asked to reply either by returning the folding card or via SMS or e-mail. Responding households were further asked to state the number of employed household members and a telephone number.
Main MAED survey | |||
---|---|---|---|
Households in % | Overall (n = 4997) | V1 (tel) (n = 1942) | V2 (no tel) (n = 3055) |
Recruitment phase
| |||
Gross sample size | 100 | 100 | 100 |
Not availablea | 39 | 88 | |
Participation rejected | 33 | 1 | |
Participation agreed: Households received MAEDs | 17 (n = 865) | 28 (n = 535) | 11 (n = 330) |
Questionnaire phase
| |||
Households returned MAEDsb | 63 | 62 | 64 |
Net response of gross sample | 11 | 17 | 7 |
Usable net response of gross sample after validation of MAEDs | 10 (n = 490) | 15 (n = 299) | 6 (n = 191) |
-
The amount varied between 30 and 60 Euros. 60 Euros were too much; the response rates and data quality did not increase accordingly. 30 Euros were sufficient for the simpler diary without expenditures in pilot study 1. The expenditures caused a considerable extra effort so that 40 Euros seemed to be the best compromise.
-
We also tested a payment in advance. The motivating effect was indeed stronger than the after-payment of the same amount. Nonetheless we decided on the after-payment to the credit of a higher data quality: respondents were better motivated to answer our validation calls before they received the payment.
Quantitative results of main survey
MAED survey | Statistics Austria National Census 2013 | NTS 2013 | Austrian Time Use Survey 2008/09 | |
---|---|---|---|---|
Households | 490 | 2,006,004 | 10,490 | 3060 |
Employed persons | 748 | 4,019,408 | 17,013 | 4546 |
Gender | ||||
Male | 50.0 | 53.3 | 53 | 50.0 |
Female | 50.0 | 46.7 | 47 | 50.0 |
Age | ||||
15–19 | 2.3 | 5.0 | 0.9 | 3.5 |
20–29 | 6.8 | 19.5 | 13.6 | 17.1 |
30–39 | 18.7 | 22.6 | 19.1 | 26.8 |
40–49 | 35.7 | 29.1 | 31.7 | 30.1 |
50–59 | 31.9 | 20.0 | 31.2 | 19.1 |
60+ | 4.6 | 3.8 | 3.5 | 3.4 |
Employed | 88.7 | 88.8 | n.d. | 89.1 |
Self-employed | 11.3 | 11.2 | n.d. | 10.9 |
Compulsory education | 2.7 | 17.8 | 5.9 | 11.7 |
Apprenticeship, vocational school | 36.0 | 50.9 | 48.3 | 60.2 |
High school | 24.3 | 15.9 | 20.2 | 14.5 |
College, university | 37.0 | 15.4 | 25.6 | 13.6 |
MAED survey | Statistics Austria National Census 2013 | NTS 2013 | Austrian Time Use Survey 2008/09 | |
---|---|---|---|---|
Household size
| ||||
1 person | 14.5 | 30.2 | 13.3 | 15.1 |
2 persons | 29.4 | 23.1 | 30.0 | 27.0 |
3 persons | 22.0 | 19.0 | 24.0 | 22.7 |
4 persons | 27.1 | 18.2 | 22.6 | 24.6 |
> 4 persons | 6.9 | 9.6 | 10.1 | 10.5 |
Urban | 24.1 | 33.5 | 26.7 | 26.8 |
Intermediate | 28.2 | 29.9 | 27.9 | 28.2 |
Thin | 47.8 | 36.7 | 45.4 | 45.0 |
Target region
| ||||
Eastern Region | 33.9 | 44.1 | 47.5 | 30.2 |
Upper Austria | 23.1 | 16.9 | 5.6 | 15.0 |
Styria | 18.2 | 13.8 | 21.9 | 10.4 |
Salzburg | 6.9 | 6.4 | 4.7 | 15.7 |
Carinthia | 5.1 | 6.2 | 4.2 | 9.0 |
Tyrol, Vorarlberg | 12.9 | 12.7 | 16.0 | 19.7 |
Mobility
MAED survey | NTS 2013 | χ2 | p value | |
---|---|---|---|---|
n = 748 persons n = 18,203 trips | n = 9436 persons n = 57,044 trips | |||
Share of mobile persons
| ||||
Working day | 0.97 | 0.91 | 162.1 | p < 0.0001 |
Saturday | 0.89 | 0.82 | 15.9 | p < 0.0001 |
Sunday | 0.71 | 0.69 | 1.82 | p = 0.257 |
Number of trips per mobile person
| ||||
Working day | 3.98 | 3.39 | 8.09 | p < 0.0001 |
Saturday | 3.74 | 3.40 | 1.78 | p = 0.074 |
Sunday | 2.85 | 3.15 | − 1.43 | p = 0.154 |
Distance of trips (km) | ||||
Per trip | 12.1 | 14.9 | − 8.06 | p < 0.0001 |
Per day | 45.9 | 51.9 | − 3.47 | p = 0.0005 |
Duration of trips (min) | ||||
Per trip | 23.9 | 25.6 | − 3.98 | p < 0.0001 |
Per day | 90.8 | 89.5 | 0.56 | p = 0.576 |
Mode choice
| ||||
Public transport | 10.9 | 11.9 | 7.36 | p = 0.007 |
Car | 69.5 | 70.1 | 0.44 | p = 0.507 |
Bicycle | 5.8 | 5.5 | 3.25 | p = 0.072 |
Walk | 13.8 | 12.5 | 4.61 | p = 0.032 |
Time use
Consumer expenditures
Category | Classification | Sources |
---|---|---|
Housing | Committed | H |
Food | Committed | D |
Accommodation and restaurants | Non-committed | D |
Clothing | Non-committed | D, H |
Furnishing, household equipment | Committed | D, H |
Health | Committed | D, H |
Travel | Committed | D, H |
Electronics and communication | Non-committed | D, H |
Leisure, recreation, culture | Non-committed | D, H |
Education | Committed | D, H |
Services | Committed | D, H |
Financing | Committed | D, H |
Insurance | Committed | D, H |
Savings | Non-committed | H |
Other | Non-committed | D, H |
-
Method 1 is based on a source selection method described by Creech and Steinberg (2011) in the Consumer Expenditure Survey Anthology by the U.S. Bureau of Labor Statistics. The Consumer Expenditure Survey (CE) also consists of two instruments, a diary survey for all expenses incurred over a 2-week period and an interview survey that captures expenses for a recall period of 3 months or longer. In order to deal with the significant amount of overlap and to select an appropriate source, the Personal Consumption Expenditure (PCE) estimate produced by the Bureau of Economic Analysis (BEA) is used as reference for a comparison. A Mean Squared Error (MSE) is calculated by adding the variance of the CE data to the squared difference between the mean of the CE data and the PCE estimate. This is performed for both CE sources and the source with smaller MSE is chosen for each expenditure category. We adapted the method for a comparison with the Austrian Consumer Expenditure Survey estimates. The expenses stated in the household questionnaire have consistently smaller variations than those in the diary, so that the decision was always in favour of the household source. However, the smaller variation does not necessarily indicate a higher reliability; it results from the fact that the household section comprises averaged estimates, whereas the diary comprises actual costs with higher variation but lower risk of biased perception.
-
In Method 2 we did not select a particular source but calculated the weekly mean values (MV) of diary and household expenses for each overlapping category. This method has two advantages: (1) it reduces the number of zero spendings due to mutual completion; (2) it avoids inconsistencies, if the diary includes expenses during the reporting week, whereas the household section states no spendings for the same category.
Conclusions and outlook for further research
-
The overall diary structure resembles a conventional travel diary, which structures the day by trips; non-travel activities are nested within the trips.
-
Non-travel activities are reported in open time intervals and pre-defined activity categories, although the HETUS guidelines recommend pre-defined intervals and open activity descriptions. This alteration was necessary to keep the response burden at a reasonable level; for the same reason we omitted the parallel activity description, which is a serious downside of this simplification.
-
Questions on expenditures are placed within different sections of the diary to achieve an intuitive and self-explanatory scheme: travel expenses in the trip section, expenses related to non-travel activities in the activity section, long-term expenses in the household questionnaire.