Old Chapters > Chapter 8 (old)

Chapter 8 (old)

CHAPTER 8.0 HOUSEHOLD SURVEY INSTRUMENT DESIGN

Note: Significant components of this chapter come from Chapter 2 of NCHRP Report 571, as well as from Section 6.6 of the FHWA Travel Survey Manual. Most material has been reviewed and updated by Peter Stopher, Kay Axhausen, Stacey Bricka and Eric Molin.

Chapter still has many outdated examples of surveys and other materials that need updating to the 21^st Century .

8.1 Drafting and Constructing Household Travel/Activity Surveys

The household survey is probably the best travel survey for obtaining the most detailed data on respondents and their travel patterns. As discussed in Chapter 3.0, the household survey can include almost any type of survey question. In addition, the survey may include either interviews or self-administered questionnaires, or both. Household survey interviews can either be computer-assisted or manual.

The first challenge for the survey team is to determine what data elements are needed from the household travel/activity survey. This determination must be based on the anticipated analyses and the survey goals, but the data elements of other recent household travel/activity surveys will help the survey team narrow its selection. Once the data elements are identified, they need to be developed into survey questions and response categories. At the same time, the survey team needs to identify the different survey instruments and materials that will be needed for the survey effort. The final product of this task will be the survey materials, which will be a combination of the products of the three steps.

The three key steps of drafting and constructing household travel/activity surveys are described below.

8.2 Data Elements for Household Travel/Activity Surveys

In most household survey designs, the survey team is in the position to obtain information on a great number of relevant topics. The analysis and travel demand modeling plans for the survey results and the survey’s overall goals will dictate what specific data elements need to be included, which data elements should be included, and which data elements could be included, if possible.

Unfortunately, in general, the surveyor does not have the luxury of including as many data items as possible in a household travel/activity survey because the length of the survey will affect the quality of responses and the level of non-response. The tradeoff between survey length and response quality is discussed below in the section describing the development of survey questions, but because it is almost always true that users of survey data would like more questions than can be asked, the survey team needs to determine the data elements of the most interest for the survey effort, and prioritize their inclusion in the final surveys.

The remainder of this data elements section describes the most common information sought in household travel/activity surveys. Each survey team will have different data needs from their surveys, but most household travel/activity survey efforts have common concerns. Therefore, reviewing data elements that are frequently collected is a productive exercise.

Two recent research papers provide excellent taxonomies of the recent content of household travel/activity surveys. Axhausen provides a detailed catalog of many recent household travel/activity surveys, including many examples of North American, European, and Australian surveys (Axhausen, 1994). Stecher, Bricka, and Goldenberg provide a breakdown of household survey data categories and elements from recent North American efforts (Stecher, Bricka and Goldenberg, 1995).

Applying these taxonomies, household travel and activity survey data elements can be categorized into the following five categories:

Household Data – Information on the characteristics of the household and on the actual physical property in which the household resides.

Person Data – Demographic, socioeconomic, and employment information for one or more members of the household.
Vehicle Data – Information on the type, ownership, and usage of private vehicles available to household members.
Travel and Activity Data – Diary or recall information about the travel and activities of one or more household members.
Attitudinal, Opinion, Knowledge and Stated Preference Data – Information from respondents that provide surveyors and modelers with the respondents’ views, tastes, and concerns.

For the first four of these categories, recommendations have been developed of the minimum specifications for any household travel survey, which are shown here in Table 8.1. These represent the minimum set of questions recommended in all future household travel surveys. Additional questions may be asked in many surveys, however. For example, income is not included for reasons that are discussed below. It should be noted that many recent surveys would not conform to this specification, which is also indicative of the problems of comparability and completeness in current surveys.

8.2.1 Household Data Elements

Table 8.2 provides a more extensive list of common household data items for household travel/ activity surveys. These data elements may be used to classify respondent households as independent variables in travel demand models, and to compare the sampled households with actual study area households for which Census data are available. It should be noted that the minimum questions labeled H1-H7 in Table 8.1 are included within this lengthier list.

Table 8.1 Recommended Minimum Question Specifications

Category

Ref.

Item

Description

Household

Location

Home address or home position in geographic terms

Type of Building

Detached, semi-detached, terraced, flat, etc.

Household Size

Number of household members

Relationships

Matrix of relationships between all members of the household

Number of Vehicles

Summary of number of vehicles from vehicle data

Housing Tenure

Own or rent status

Re-contact

Willingness to be contacted again for further surveys, etc.

Personal

Gender

Year of Birth

(Preferable to requesting age)

Paid Jobs

Number of paid positions and hours worked at each in the past week

Job Classification

Employee, self-employed, student, unemployed, retired, not employed, etc.

Driving License

Whether a current drivers license is held

Non-mobility

Indication of why no out-of-home activity was performed on a survey day including work-at-home days

Education Level

Highest level of education achieved

Disability

Types of mobility disability, both temporary and permanent

Race^a

Defined as currently measured in the U.S. Census

P10

Hispanic Origin¹

Defined as currently measured in the U.S. Census

Vehicle

Body Type

e.g., car, van, RV, SUV, etc.

Year of Production

Ownership of Vehicle

Household/person, lease, institution

Use of Vehicle

Main user of vehicle

Activity

Start Time^b

Activity or Purpose

Location

Where the activity was performed, unless traveling

Means of Travel

If activity is travel, what mode(s) was used (including specifying if a car passenger or driver)

Mode Sequence

Unless collected as fully segmented data

Group Size

Number of persons traveling with respondent as a group

Group Membership

Number of persons in the group who live in respondent’s household

Costs

Total amount spent on tolls, fares, and respondent’s share

Parking

Amount spent to park

^aAll surveys would use the U.S. Census Bureau definition of Race and Hispanic Origin.

^b Only start time needs to be ascertained in a time-use or activity survey because, by definition, the start time of an activity is the end time of the previous activity. Only the last activity should need an end time. In a trip-based survey, start and end time should be included.

Table 8.2 Household Travel/Activity Survey Household Data Elements

The number of people in the household is a key consideration in estimating household level trip generation rates. The primary challenge for the household travel/activity survey with regard to this data element is to define to the respondent what is meant by the term household. In general, travel survey teams define a household as the total group of people who usually reside at the sample address, regardless of whether they are related to each other. Travel survey teams need to consider how they will account for college students and others whose legal address is at the sample household, but who live elsewhere. In addition, the travel survey team must decide how to handle visitors to sample households, particularly for survey efforts with travel or activity diaries.

The geographic location of the household is usually an essential data element. Any survey effort that will rely on geographic analyses of some type needs to define accurately and precisely the household location. If the location cannot be coded in sufficient detail, the data record usually cannot be used. Hence, the travel survey team needs to determine the level of geographic detail necessary for future analyses (are neighborhood definitions sufficient? are exact addresses needed, are latitude and longitude needed, or can nearest intersections and landmarks be used?). In addition, if interview methods are employed, the travel survey team should determine ways to test the sufficiency of the geographic data as soon as it is collected, and before continuing on with other parts of the survey. If the collected data are unusable, then the interviewer can attempt to clarify the response.

Household income information is commonly used in travel demand models and other survey-related analyses, but the collection of household income data is among the more challenging aspects of household travel/ activity survey instrument design. Household income questions almost always have significant levels of item non-response and refusals. In addition, many travel surveyors have questioned the validity of the self-reported income information, based on comparisons with Census income data and other sources.

The question design issues for this data element are discussed below, but a more basic decision related to household income is how the survey data, which is likely to have problems even with the best designs, may be used for analyses. The non-response workshop of the recent TRB Household Travel Survey Conference recommended that travel modelers recognize the inherent limitations of this data element, and that, at a minimum, they consider combining response categories into a few large categories in an attempt to improve the data reliability.

8.2.2 Person Data Elements

Items P1 through P10 represent the minimum required person data elements, while Table 8.3 shows a more extensive list of common person-based data elements. The person-based data elements for household travel/activity surveys are often used as explanatory variables in trip distribution and mode choice models. In addition, these data are commonly used to compare the survey respondents to the U.S. Census population for the study area. The data items for person-based data are generally straightforward, and often include questions about jobs and workplaces.

8.2.3 Vehicle Data Elements

The recent focus on the interaction between travel demand models and air quality models has led a number of travel surveyors to ask for detailed vehicle information in household travel/activity surveys. It remains to be seen which of the very detailed data elements will prove to have significance in travel demand models, so recent surveys have sought many different elements. Items v1-V4 are considered to be the minimum vehicle details required, but Table 8.4 shows a more extensive list of data elements that have recently been sought. Because a number of the listed data elements require respondents to record information from the vehicles, not all of the data elements are appropriate for all survey methods. For instance, a simple telephone survey could not be used to obtain these data elements.

Table 8.3 Household Travel/Activity Survey: Person Data Elements

Table 8.4 Household Travel/Activity Survey: Vehicle Data Elements

It is often a challenge to define for respondents the vehicles of interest for the survey. In general, survey teams are interested only in vehicles that are registered and operable. The survey team needs to determine whether only vehicles kept by household members should be reported, or whether all vehicles that respondents could use should be reported. Since informal car-sharing and borrowing are common in the U.S., it is important that survey teams define before the survey effort what they will need for their analyses.

Often, vehicle availability data are used in conjunction with U.S. Census data to expand the survey sample. If this is the case, it is imperative that the survey question be consistent with the Census vehicle availability question:

“How many automobiles, vans, and trucks of one-ton capacity or less are kept at home for use by your household?”

8.2.4 Travel and Activity Data Elements

Household travel/activity surveys commonly collect a great deal of highly detailed data on people’s activities and trips. The minimum specification is shown in Table 8.1, in items A1 through A9. Table 8.5 summarizes many of the most common data elements. Usually, these detailed data are collected by means of diaries that either record all respondent trips over a pre-specified time period or record all the activities that respondents engage in over a pre-specified period. It is likely that all activity-based modeling systems will require the use of activity diaries, but conventional travel demand models can utilize either activity diary data or travel diary data. As discussed above, the trend among recent household travel/ activity surveys has been toward the use of activity diaries.

Table 8.5 Household Survey: Travel and Activity Data Elements

If the survey team chooses to use an activity diary to record respondent activities, a fundamental question that needs to be addressed is whether and how to categorize activities. The first choice for the survey team is whether to include only activities that are performed outside the home, or to include both in-home and out-of-home activities. Recent household surveys have used both techniques.

The next question for the survey team is how to record people’s activities. Because of the wide range of potential responses, the most accurate approach for obtaining the information probably involves the use of open-ended questions, with interviewer probing as necessary. However, this approach increases the burden both on respondents and interviewers. Therefore, survey teams have generally defined activity classification schemes, and have asked respondents to categorize their activities on the basis of those schemes. Table 8.6 shows some of the activity categories that have been used recently in household surveys. More classifications of people’s activities are obtainable from the many time-use surveys conducted since the 1970s. These studies tend to have extremely detailed classification schemes (some with more than 100 categories) for how people spend their time (Szalai, 1972). A recommended set of categories is provided later in this chapter.

Because of the large number of activities that people perform in a typical travel diary period, many household activity survey teams have asked respondents to record only activities that last for more than 30 minutes or activities that require travel (regardless of how long the activity takes). This decision rule limits the reportable activities to a manageable number, but may also dilute the usefulness of the time-use data because many activities may not be reported. Certain types of activities – like meals – that often take less than 30 minutes and do not involve travel are likely to be under-reported.

The level of under-reporting can be limited by having interviewers probe for any at-home activities that could be substituted with activities that take place outside the home, such as eating out/catalog shopping. High quality interviewers are essential for an approach like this, because the probing could greatly increase the data retrieval time if it is not performed efficiently. Another circumstance that time-use and activity-based survey designs must consider is when activities occur simultaneously or when an activity (like eating, reading, or paperwork) is completed in the course of travel.

When trip-based methods are employed, rather than activity-based methods, a key issue in the design of diaries is the definition of a trip. Respondents are likely to define the word, ‘trip’ differently than the survey team members. Therefore, the interviewer or the survey instruments need to explain the term to respondents so that they will answer the question as accurately as possible. In most studies, trips are defined as one-way travel between an origin and a destination other than the origin. This issue is discussed in more detail later in the section on question wording.

Table 8.6 Household Activity Categories for Respondents

8.2.5 Attitudinal, Opinion, Knowledge, and Stated Preference Data Elements

The final data elements that are commonly collected in household travel/activity surveys are the attitude, opinion, knowledge, and stated preference data, including the data elements shown in Table 8.7. These data are not discussed extensively in either the Stecher, Bricka and Goldenberg paper or the Axhausen paper, because of their focus on household surveys that are used for regional travel model development. Because of their length and complexity, household travel and activity surveys that are conducted for purposes of model building should be limited to the collection of data that will be (or could be) used as model inputs. Survey teams should resist the temptation to collect “nice-to-know” information without first understanding exactly how it will be used. Other types of household travel/activity surveys, such as those performed for transit agencies, are more likely to employ the types of data elements listed in the table.

Stated response data have been collected in many different types of household surveys, including those that are used for regional model development. The issues related to these data elements are discussed in Chapter 13.0.

Table 8.7 Household Travel/Activity Survey Attitudinal, Opinion, Knowledge and Stated Preference Data Elements

8.3 Translating Required Data Elements Into Questions and Response Categories

Once the survey team and travel demand modeling staff have established the essential and optional data items to be included in the household travel/activity survey, survey questions need to be developed that will produce the data elements accurately and reliably.

A survey question should be included in a household travel/activity survey interview or questionnaire if the following are true (Backstrom and Hursh-Cesar, 1981):

The information obtained from the question is relevant to the models being developed or refined, or to other anticipated analytical efforts or policy definition.
The question and response categories are expected to be valid measures of the modeling variables.
The responses can be coded meaningfully for modeling analyses (where appropriate).
Analysts, interviewers (if any), and respondents agree unambiguously on the meaning of the question and response categories.
The question and response categories have no wording problems.
The wording of questions and responses is the same or equivalent to any measure from other surveys that will be used in the modeling work.
Response categories exhaust all meaningful answers that can be anticipated.
Response categories are meaningful and understandable to respondents.
(For interviews) the questions and response categories are easily learned by interviewers.

In addition, each survey question should be reviewed in terms of its effect on the overall survey quality. The following should also be true:

The benefits in the survey analysis from the question outweigh its costs in terms of survey length, respondent burden, and increased non-response.
The information gained from the question is more useful than the information that would be gained from other questions that will not be on the survey.
The question does not provoke respondents to be hostile to the survey effort or to question the goals of the surveying agency.

In short, a question and its response categories should provide unambiguous, accurate, reliable, and usable information without affecting the overall validity of the survey effort.

To meet these requirements, the survey team must successfully perform three tasks in translating data elements into actual survey questions:

Determine exact question content and the forms of the questions;
Develop the wording for questions and response categories; and
Determine how questions should be sequenced.

These tasks are described below.

8.3.1 Determining the Form and Content of the Questions

The survey team needs to operationalize the survey’s required data elements by defining in more detail what data are needed and by determining the appropriate question forms to obtain the data. The list of needed data elements that the survey team assembles while analyzing the proposed analyses based on the survey data will include a number of items that will lead directly and logically to the development of one or more survey questions without much effort. For instance, if one of the identified needed data elements is whether a person has a valid drivers’ license, the data could be collected by simply asking the respondent.

On the other hand, for some data elements there will not be a clear set of survey questions. These data elements require that the survey team improve its data definitions and consider more carefully what survey information is required.

For instance, one data element that is often sought in household travel/activity surveys is whether an individual had a vehicle available for a particular trip from home to work. The most straightforward and commonly employed approach to obtaining this data element is to ask, “did you have a vehicle available for this trip?” or some variation. However, as Axhausen points out, this simple question is riddled with ambiguity (Axhausen, 1994). Does answering yes to the question mean that there was a vehicle at home that the person could have driven, or does it mean that, if necessary and with advance planning, the person could have arranged to take a vehicle, perhaps by changing how other household members travel, or does it mean something else?

Given this problem, the survey team might want to consider the development of a series of questions to determine how the household allocates vehicle usage and how mode choice decisions are made, or they might want to consider asking questions that would allow them to trace the usage of all household vehicles throughout the travel period in question. Before deciding how best to proceed the survey team should re-examine the expected analyses that will rely on these data elements to find out the best approach.

As the survey team evaluates the household travel/activity survey’s needed data elements, they should also consider the types of survey questions with which the data elements are best matched with. In Chapter 3.0, different kinds of survey questions, including factual, behavioral, test-of-knowledge, attitudinal, opinion, and stated response questions, were defined. However, in developing the survey questions it is useful to classify questions differently, according to the question form. Figure 8.1 shows a classification scheme for the forms of survey questions. As the top of the figure shows, survey questions can be either open-ended, which allow respondents to reply to the questions freely, or closed-ended, which provide respondents with preset response categories. Open-ended questions are sometimes useful, particularly when:

The survey team is uncertain about the possible range of responses to a question;
The question has so many potential responses that providing categories would be infeasible;
The survey team needs to have precise information on how respondents think about something;
The survey team would like to provide respondents with the opportunity to sound off about the survey topic(s); or
The survey team is seeking verbatim remarks to complement the statistical analyses.

Geographic questions are almost always open-ended, because locations that individual respondents refer to cannot be foreseen by the survey team. In addition, questions that ask for respondents’ opinions are generally open-ended.

However, almost all household travel/activity surveys rely to a large degree on closed-ended questions for most data elements, because both data collection and survey processing are greatly facilitated by forcing responses into a small number of categories. Closed-ended questions allow respondents to simply pick a reply, rather than to form one from scratch. In addition, the possibility of respondents providing inappropriate answers or of misinterpreting questions is limited by providing preset responses. Finally, closed-ended questions are much easier to code than verbatim respondent replies.

Because of these advantages, many interview questions that appear to be open-ended are actually made closed-ended by having interviewers categorize respondents’ remarks into preset divisions.

Figure 8.1 Survey Question Forms

Closed-ended questions are generally one of four types, nominal, ordinal, interval, or ratio. This classification is useful in question design, because it is usually easy to determine which of the four categories best addresses particular needed data elements. Because there are relatively few ways to phrase survey questions in each of the four categories, the question construction of other survey questions in the category can often be used as a guide in converting the data element to a useful survey question.

8.3.2 The Wording of Questions and Response Categories

Before developing the wording for any questions, the survey team should understand how the survey process and specific questionnaire wording can bring inaccuracies and biases into the survey results. Two basic survey question problems can harm the survey effort:

Item non-response; and
Inaccurate replies (response errors).

To minimize the effects of these problems, it is first important to understand what possible motivations respondents might have to be less than forthcoming or to mislead the surveyor. There are four general reasons why respondents can provide incomplete or inaccurate information in surveys:

The respondent does not know the answer to a question or questions;
The respondent cannot remember the answer to a question or questions;
The respondent misunderstands the question; and
The respondent has some motivation not to be totally forthcoming.

While it is not possible for survey teams to eliminate these issues, it is clear from recent household travel/activity survey efforts that good question design can certainly reduce the number of response errors significantly.

8.3.2.1 Categories for Minimum and Other Questions

For the minimum questions suggested in the section 8.2, a number are subject to considerable variability in the categories that might be used to record the data. Proposing a set of standardized categories for those questions and also for one or two other questions that are likely to be included in many surveys would seem useful in pursuing the goal of improving comparability, accuracy, and quality. The variables selected for standardization of categories are type of dwelling (H2), relationship (H4), housing tenure (H6), education level attained (P7), disability (P8), race (P9), vehicle body type (V1), vehicle ownership (V3), trip purpose (A2), means of travel (A4), fuel type, and employment status. Most of the other questions either do not require categories or the categories are self evident and already standardized (e.g., gender [P1]).

To devise recommended standardized categories, definitions of seven international statistical agencies for the specific variables were looked at and compared with the seven data sets examined as well as two other survey definitions. The results are shown in Table 8.8. Where less detail is desired, the primary categories are recommended to be used and these should suffice in the majority of surveys; where more detail is required, the secondary categories should be used. It is also possible to selectively add secondary categories—for example, housing tenure might be used with Categories 1, 2, 31, and 32, if the additional level of detail were desired only for Category 3. Similar mixing of category levels could be used with any of the variables in Table 8.8—for example, splitting rail (A4-71) into different submodes in an area such as New York may be desirable.

Standardization of the activities to be included in designs that provide an activity list would also be useful and has been requested by some professionals involved in transportation surveys. The grouping of activities into common trip purpose–related categories would also be a useful element of this aspect of standardization.

Table 8.8 Recommended Standardized Categories

Variable

Primary Category

Code

Secondary Category

Code

Type of Dwelling (H2)

Single family house detached

Single family house attached

Townhouse

Row house

Duplex

Triplex/fourplex

Apartment/mother-in-law suite

Apartment/condominium

Condominium

Rented apartment

Mobile home/trailer

Mobile home

Trailer/camper

Dorm/group quarters

Dormitory

Hostel

Nursing home

Military barracks

Hotel/motel

Other

Relation-ship (H4)

Self

Spouse/partner

Husband/wife

De facto husband/de facto wife

Son/daughter

Natural son/daughter

Adopted son/daughter

Stepson/stepdaughter

Son-in-law/daughter-in-law

Father/mother

Natural father/mother

Adopted father/mother

Stepfather/stepmother

Father-in-law/mother-in-law

Brother/sister

Natural brother/sister

Adopted brother/sister

Stepbrother/stepsister

Brother-in-law/sister-in-law

Grandfather/grandmother

Paternal grandfather/grandmother

Maternal grandfather/grandmother

Grandchild

Grandson

Granddaughter

Other relative

Male

Female

Not related

Boarder

Housemate/ room mate

Other non-relative

Housing Tenure (H7)

Own

Owned with mortgage

Owned without mortgage

Rent

Rent paid

Occupied without rent

Provided by job/military

Provided by job

Provided by military

Education Level (P10)

No school completed

Elementary school

Preschool/nursery

Kindergarten–4th grade

High school

5th–8th grade (junior high)

9th–12th grade (no diploma)

High school diploma

College/university

Some college but no degree

Associate degree in college

Bachelor’s degree

Post graduate studies

Some graduate school, no degree

Master’s degree

Professional school degree

Doctorate degree

Disability (P11)

Difficulty standing

Difficulty climbing

Visually impaired/blind

Hearing impaired/deaf

Require wheelchair

Require cane/walker

Other (specify)

Race (P12)

White (alone)

Black/African American (alone)

American Indian/Alaskan Native (alone)

American Indian

Alaskan Native

Asian (alone)

Asian Indian

Chinese

Filipino

Japanese

Korean

Vietnamese

Other Asian

Native Hawaiian or Pacific Islander (alone)

Native Hawaiian

Guamanian or Chamorro

Samoan

Other Pacific Islander

Some other race (alone)

Two or more races

Vehicle Body Type (V1)

Auto

Van

Recreational vehicle (RV)

Utility vehicle

Pick-up truck

Other truck

Motorcycle

Other (specify)

Vehicle Ownership (V5)

Household member owned or leased

Employer owned or leased

Other (specify)

Trip Purpose (A2)

Home

Home – domestic activity

Home – paid work

Work and work-related

Main job

Other job

Volunteer work and community services

Looking for work

Education/childcare

Attendance at childcare

Attendance at school

Attendance at college

Eating out

Restaurant/café

Fast food

At friends’ home

Personal business/medical

Availing of/shopping for administrative services

Availing of/shopping for professional services

Availing of/shopping for government/public services

Availing of/shopping for personal services

Availing of/shopping for medical and health care services

Shopping

Purchasing food and household supplies (groceries)

Purchasing clothes, shoes, personal items

Purchasing household appliances, articles, equipment

Purchasing capital goods (cars, houses etc.)

Comparison shopping

Window shopping

Social/recreational

Communication/ correspondence

Socializing activities

Participating in religious/community/cultural events/activities

Visiting entertainment and cultural venues

Indoor and outdoor sporting activities

Games/hobbies/arts/ crafts

Print/audio/visual media

Accompanying others/travel related

Accompanying children to places

Accompanying adults to places

Pick up or drop off other people/get picked up or dropped off (private car, car/van pool, shuttle/limousine)

Activities related to bus, public transit and group rides (except car/van pool and shuttle/limousine)

Change travel mode

Other (specify)

Not further defined (n.f.d.)

Means of Travel (A4)

Car/van/truck driver

Car driver

Van driver

Truck driver

Car/van/truck passenger

Car passenger

Van passenger

Truck passenger

Motorcycle/moped

Motorcycle

Moped

Scooter

Bicycle

Walk/wheelchair

Walk

Skate/roller skate/ roller board

Motorized wheelchair

Non-motorized wheelchair

Bus/school bus

Regular bus

Intercity bus

Express bus

School bus

Train

Trolley/streetcar

Taxi/shuttle

Taxi

Shared-ride taxi/jitney

Commuter van/shuttle bus: employer paid

Commuter van/shuttle bus: pay fare

Dial-a-Ride

Shuttle/limousine

Other (specify)

Fuel Type

Gasoline

Diesel

LPG/LNG

Dual Fuel

Other (specify)

Employment Status

Full-time

35–45 hours

46–55 hours

Greater than 56 hours

Part-time

Less than 20 hours per week

Greater than 20 hours per week

Retired

Semi-retired

Full-time homemaker

Unemployed seeking employment

Unemployed not seeking employment

Full-time student

Child not in school/infant*to be specified if skip mechanism not in place

Volunteer work (unpaid)

The remainder of this section discusses common survey question wording problems, and then discusses question wording issues for particular household travel/activity survey questions, including diaries.

8.3.2.2 Survey Wording Problems

Each question should be tested by the designer for the potential problems listed above. To minimize the chances that a respondent will not know, remember, understand, or be willing to answer survey questions, the survey team should seek to avoid questions that fall into three broad categories:

Confusing questions;
Ambiguous questions; and
Loaded questions.

Confusing questions are questions that mix up respondents in some way. Ambiguous questions are questions which not everyone would agree mean the same thing. Loaded questions are questions that suggest to respondents that certain responses are preferable to others. Loaded questions are usually of the most concern on attitude, opinion, and stated response questions. Since household/activity questions use these types of questions infrequently, they are probably less of an issue than misperceived or ambiguous questions.

Tables 8.8, 8.9, and 8.10 provide examples of survey questions that have these problems. Although these examples are contrived, they illustrate the many ways question wording can lead to response errors. It is fairly easy to identify questions that could be potential problems in many recent household travel/activity survey materials. In most of these cases, the survey teams probably considered alternative wording but found that the alternatives introduced wording problems of other types. Ultimately, the household travel/activity survey team needs to use its judgment, experience, and the results of carefully-designed pretests to make final decisions about survey wording.

Table 8.8 Examples of Confusing Survey Questions

Table 8.9 Examples of Ambiguous Survey Questions

Table 8.10 Examples of Loaded Questions

As a general rule, the wording of questions and response categories on surveys should be aimed at the respondent audience. For household travel/activity surveys, this means that the wording needs to be designed for a broad audience with a wide-range of reading and comprehension abilities, and with differing levels of interest in transportation. The challenge for the survey team is to word the survey as simply as possible without boring more advanced respondents.

The following question-writing principles are suggested to achieve this goal:

The household travel/activity survey should be understandable by the average fourth grader. Once a preliminary set of survey questions is developed, the survey team may want to try administering the questions to some children to see whether any questions are confusing for them.

The household travel/activity survey should hold the interest of intelligent adults who are not involved in transportation planning or market research. The travel survey team may also want to test this principle by administering the survey to non-technical friends and acquaintances. These people can indicate where the survey is tedious and whether certain questions feel condescending.

Almost none of the household travel/activity survey respondents will have a background in transportation planning or surveys, and many will have little or no interest in the subjects. Survey team members need to be extremely careful about projecting their level of knowledge and interest onto potential respondents. In particular, the transportation planning field is full of jargon and expressions that are not obvious to non-planners. The survey team needs to be very careful with the use of many words, including:

“Trip”
“Journey”
“Travel”
“Activity”
“Origin”
“Destination”
“Mode”
“Trip purpose”
“Bus”
“Shuttle”
“Transit”
“Transfer”

This is not to say that such terms should not be used. Rather, the terms are often central to the information that is being sought, and can be absolutely necessary in many cases. The survey team needs to be aware, however, that these and other terms do not always mean the same thing to everyone, and for some people, they will mean nothing. The survey team needs to make sure that the terms are either defined for respondents, or that the context in which they are being used does not allow for ambiguity in their meaning.

In addition, marketing research and survey analysis terms and jargon should be avoided in surveys, because some respondents will resent being part of an experiment. Among the words to watch for in this regard are:

“Questionnaire”
“Research”
“Data”

8.3.2.3 Question Wording Issues for the Household Travel/Activity Survey

In household travel/activity surveys, a few question types are known to be problematic. Three particular question types are discussed here:

Questions about household income;
Questions about personal property; and
Activity and travel diary questions.

8.3.2.4 Questions About Household Income

Household income questions usually have the highest levels of non-response of all the survey questions. Recent survey efforts have reported item non-response rates for household income questions of more than 10 percent. Unfortunately, the people who refuse to answer income questions are usually not representative of the whole population. Research indicates that those who refuse to answer income questions are more likely than the population as a whole to have higher income levels (Richardson, Ampt and Meyburg, 1995). In addition, some recent U.S. survey efforts, such as a recent statewide effort in New Hampshire, have found that households in the lowest income categories are less likely to complete these surveys. This may be due to either a higher income non-response rate for this group or to an overall higher unit non-response. Therefore, analyses of income data with significant non-response are likely to be biased.

Many surveyors believe that the income question has a high response error level, as well. Many respondents who are unwilling to provide accurate income information sometimes make up an answer, rather than simply refusing to answer it. Also, some respondents will not know their total household income or will be confused about which types of income to include in their estimates. It is difficult to determine the magnitude of the response error, because it is impossible to tell valid responses from invalid ones.

Because of the perception of response error, some surveyors believe that there is a practical maximum number of income categories beyond which the data are probably too inaccurate and the question either takes too long in an interview or takes up too much space on the mail survey form. These practitioners believe the survey team should limit the number of income categories to between 8 and 12.

It is generally accepted by travel surveyors and market researchers that the household income question should, if possible, be the last question on the survey. If the question is asked earlier in the survey, the likelihood of the respondent not completing the survey is increased. In addition, once a respondent refuses to answer a question, the likelihood that he or she will refuse to answer others, as well, increases. For some survey efforts, such as those where more than one household member is interviewed, asking income as the very last question is not always feasible. In these cases, survey teams should attempt to sequence the income question after other demographic questions and after the collection of any other descriptive information that may be used in survey expansion.

A number of wording and questionnaire design techniques have been tried to reduce the level of non-response and improve the quality of data from income questions on household travel/activity surveys. Unfortunately, many of the techniques contradict each other, and their success may be specific to certain survey populations. Since careful comparisons are not usually made, it is impossible to say how effective each is.

If the survey team identifies the need for a survey question about household income, they should consider evaluating alternative wording and questionnaire designs as part of the survey pretest to identify the best solution for their area. The form of the income question can be varied, so that pretest respondents receive different questions. The preliminary pretest results may indicate that one question form is superior to others. Alternatively, if a survey team is using one or more focus groups for developing the questionnaire, the focus group participants can be asked to assess the relative invasiveness of different question forms.

Some recent household travel/activity survey teams using telephone data retrieval have reported some success with asking for household income information in a series of choices, rather than by listing all the categories.

The question is structured, as follows:

For 2009, will your household’s total income from all sources, before taxes and any other deductions from pay be less than $55,000, or $55,000 or more?

Ask all with household income less than $55,000: Will it be under $40,000, or $40,000 to $55,000?

Ask all with household income $55,000 or more: Will it be more than $55,000 and less than $75,000, or $75,000 or more?

The simple choice questions continue until the desired level of categorization is achieved. In this type of question structure, the first query ($55,000 in the example) is usually set near the median household income, or a little lower. This approach can be useful, because in some cases, partial data can be collected from people who would not have responded to the usual income question. On the other hand, the question lengthens the interview at a point where most respondents really would like the interview to be over.

Travel surveyors have tried different approaches of leading into the income question. Some surveys have re-stressed confidentiality and the study goals before asking the question. The household income question on the 1994 Boise survey is worded:

Now a question just for statistical and travel forecasting purposes, we need to know your total household income before taxes. I will read several ranges to you. Please stop me when we reach the right one (interviewer then reads categories in ascending order).

Other travel surveyors believe the best approach to asking income is to include the question at the end of a series of short factual demographic questions. The income question is asked in the same quick way as the other demographic questions. The hope is that the respondent will sense that the income question is just another question that will be used to differentiate groups of people, and that the respondent will simply fall into the rhythm of answering questions.

In telephone-mail-telephone surveys, some surveyors believe that difficult questions, like income, should always be deferred until the retrieval call. On the other hand, some telephone-mail-telephone surveys have asked the income question at the end of both the recruitment interview and the data retrieval interview. Surveyors have found that some people who refuse to supply the information in the recruitment (which is a “cold call”) will answer the question during the data retrieval (presumably, they have been convinced of the survey’s legitimacy).

Similarly, some household travel surveys have asked all adult members of the household to answer the question, in the hopes that if one household member is reluctant to give out the information, others may not be. Of course, methods that seek the information more than once may lead to consistency questions if the survey team ends up with more than one household income estimate for a household.

8.3.2.5 Questions about People’s Property

Usually the only questions on the household travel/activity survey that have significant non-response problems are those that ask about household income. However, as household travel/activity surveys become more detailed, a few other questions need to be carefully worded and presented. In particular, questions about people’s property that might interest enterprising thieves are likely to be a problem for an increasing number of respondents.

It has become fairly common to collect detailed household vehicle information on household travel/activity surveys. While most respondents will recognize a question about the number of vehicles available to the household as having valid transportation planning use, most will not be familiar enough with air quality analyses to understand the need to know the make, model, and year of the vehicle. Untrusting respondents, who question the legitimacy of the survey effort, may feel that they are contributing to some car thief’s shopping list. Some recent survey efforts have asked respondents to record the vehicle identification number (VIN) of all their vehicles. Because the most widely-known use of this serial number is to track stolen vehicles, this question may raise the suspicions of respondents even more.

As issues involving telecommuting become more important in transportation planning, it is likely that more and more household travel/activity surveys will also seek information about people’s ownership of computers, fax machines, and other equipment. Combined with travel and activity diary data and detailed addresses, this dataset would be ideal for thieves, and it is likely that respondents will recognize this fact.

To limit the non-response on questions of these types, survey teams should:

· Put the questions near the end of survey instruments;

· Explain to respondents what the data are to be used for; and

· If possible, ask only in the second or third contact with the respondents, so that they are more comfortable with the survey’s legitimacy.

8.3.2.6 Activity and Travel Diaries

Diary design is an extremely important element of questionnaire development because response errors in the form of unreported trips are common, and are almost always a serious problem for those who analyze the survey data.

Types of Diaries

Over the past 20 years, household travel/activity surveys have used several types of diaries for which Axhausen has developed the following typology:

· Stage-Based Diaries – Treat the travel on a single mode (and the associated wait time) as a building block to construct the whole trip (data are gathered on the basis of each trip stage or segment);

· Trip-Based Diaries – Establish the trip as a whole and then disaggregate them into stages, if necessary (data are gathered on the basis of the whole trip);

· Activity-Based Diaries – Focus on activities and then collect trip details to and from the activities (data are gathered on the basis of trip-end activities); and

· Half Tour-Based Diaries – Collect information for travel between home and the farthest point of a trip chain, and then fill in information on individual trips in the chain (data are gathered on the basis of the key stop in a trip chain).

Note: The whole of this section needs updating to surveys conducted in the past decade rather than surveys that are in most cases around 20 years old and represent outdated designs. However, this is a major undertaking and requires inputs from people in the US who are familiar with recent surveys, e.g., Dave Levinson. Maybe Johanna Zmud and Laurie Wargelin could be asked to update this entire section.

Stage-based diaries have been used in a 1991 mail survey in Chicago, in a 1990 telephone-mail-telephone household survey in the Bay Area, and in a 1993 telephone-mail-telephone survey in Tucson. Figure 8.2 [PS3] shows the example page from the Chicago survey. In this diary, information on each trip stage is recorded in one column. The Chicago survey is stage-based, because one of the valid answers for the question, “Why did you go to this destination?” is to “change type of transportation.” Several columns of information might be needed to describe a single trip.

Figure 8.3 shows the Bay Area’s stage-based diary design. Actually, the diary instrument used for this survey was a simple memory jogger. Respondents recorded a minimum amount of information about the stages of the trips they made in the memory jogger, and then the telephone retrieval call was used to obtain details about the trip stages. Again, the respondent is asked to record information about each stage of his or her trips.

Figure 8.4 shows the Tucson survey diary’s example, which illustrates yet another diary format for stage-based reporting. In this diary, the trip stages are recorded in the numbered rows, labeled with “Then I went to.”

The stage-based design is most useful for survey efforts where the survey team has identified the need for path choice and sub-mode choice information. These diaries readily provide information both on the number of modal transfers and their locations. Because respondents record each stage of a trip, the stage-based designs tend to require more from respondents, and therefore, more space on the questionnaires than the other designs.

Figure 8.3 The Bay Area’s Stage-Based Diary

Figure 8.4 Tucson’s Stage-Based Diary

Figure 8.5 shows a trip-based diary used in a 1990/1991 British survey, the Sainsbury’s Swindon Survey. Trip-based diaries, like this one, rely on respondents to characterize their main mode of travel or provide respondents with more exhaustive mode lists which incorporate sub-modes.

A recent U.S. example of a trip-based diary is the ongoing panel mail survey being conducted by New York MTA. The instruction page for this diary survey is shown in Figure 8.6. In this diary, respondents are asked to provide the origin and final destination of trips and to record the travel modes that they used in order of usage. This panel survey effort is being used primarily to track people’s travel choices with respect to MTA services and to measure the MTA’s market share in different markets, not for the development of regional travel models. Therefore, stage-based information is not considered to be necessary.

In the past few years, the activity-based diary has become the predominant form of diary in the United States. Recent telephone-mail-telephone household travel/activity surveys in Portland, Detroit, and New Hampshire (to name a few) have used activity-based diaries. Figure 8.7 shows a portion of the activity diary for the 1994 Research Triangle Activity and Travel Survey.

Figure 8.8 shows the diary for a 1994 Detroit survey, and Figure 8.9 shows the activity diary for the 1994 Portland survey. Respondents in the Detroit survey were instructed to treat travel mode changes as destination activities. This is the activity-based counterpart to stage-based diary design. All the trip stage information is collected for detailed modeling analyses, but at the cost of asking respondents to record more details and of needing much thicker diary booklets. The Portland diary was used to record all activities, both at-home and outside-the-home. The question “Where did your activity take place?” is needed to identify multiple activities at the same location. Note that the activity-based diaries collect as much, or more, travel data than the other “Travel diaries.” The survey team is not limited to activity-based analyses by selecting to use activity-based diaries.

The final type of diary, the half-tour based approach has not been widely applied in the U.S., but might be useful for certain special types of analyses. This diary approach seeks detailed information on the primary trip within a trip chain, and then asks for less-detailed information about the other stops on the trip chain (usually limited to the number of stops). This type of diary might be particularly useful in the analysis of intercity and long-distance trips. As an example of this diary approach, Axhausen provides the diary from a recent Canadian fuel-usage survey, shown in 8.10.

Figure 8.5 An Example of a Ttip-Based Diary from the United Kingdom

Figure 8.6 An Example of a Recent U.S. Trip-Based Diary

Figure 8.7 An Activity Diary from a Recent North Carolina Research Triangle Survey

Figure 8.8 Activity Diary from a Recent Detroit Survey

Figure 8.9 Portland’s Activity-Based Diary

Figure 8.10 An Example of a Half-Tour Based Diary

Selecting the Best Type of Diary

There are good reasons to use each type of diary, so survey teams need to decide which approach to follow based on their data needs. An increasing number of survey teams are choosing to ask people to record activities, rather than trips, primarily because there is strong evidence to suggest that diaries that focus on activities, rather than trips, measure travel more completely than travel diaries. Based on analysis of the recent Boston Household Survey, Stopher concludes that: 1) activity diaries appear able to capture non-home-based trips better than travel diaries; and 2) overall trip rates per person and per household from the activity diary are significantly higher than most travel diaries measure (Stopher, 1992). Jones found that asking about activities, rather than trips on either diary surveys or recall surveys results in improved trip-rate estimates, compared to trip-based approaches (Jones, 1985[PS4] ).

Researchers and surveyors also seem to be selecting stage-based designs more frequently, whether the diary is travel-based or activity-based. This is probably a function of two things:

· Data analyses are increasingly requiring more detailed information; and

· Asking respondents to remember each part of their trips can sometimes help them remember brief stops that they would have otherwise forgotten.

Level of Detail for Diary Questions

As the example forms show, most household travel and activity surveys seek detailed information about people’s travel. This means that surveys that are to be mailed back must be carefully designed to obtain all the necessary information. However, if the diary data are to be retrieved by telephone, the survey team has two options:

· Provide complete diary forms, similar to those used in mailback surveys, from which respondents can simply read their answers; or

· Provide simplified diary forms with which respondents record key information about trips or activities, and then are asked to provide more detailed facts during the retrieval call.

With the first option, respondents are not surprised in the data retrieval call by questions that they were not expecting. In the second option, they are, to some extent. This is important because of the extreme length of diary retrieval call. If respondents are already dreading having to supply a large amount of trip or activity information, suddenly being asked “additional” questions may cause them to refuse to complete the survey. For instance, respondents to a recent NPTS pretest objected to the additional questions[PS5] .

Another advantage of providing the complete diary forms is that they are in a form that can be used if they are mailed back. If a respondent is unwilling to complete the data retrieval telephone call, the survey team may want to ask them to mail the completed diary instead. If a shortened, or abridged, diary form is used, this is not an option. In addition, with a complete diary, the quality of information collected by proxy (where one household member provides another household member’s diary information) is improved.

On the other hand, complete diary forms are more likely to appear complex or confusing, and may discourage respondents before they even begin to complete them. A complete diary form will usually have questions that will not be answered by a particular respondent. For example, questions about transit trips would not be answered by someone who never used transit. In addition, with the complete diaries, it is less likely that respondents will be willing to complete the diaries as they perform the activities and make the trips throughout the period, as requested. Instead, they may try to complete the form at one or a few times only, raising the likelihood that trips will be forgotten.

To alleviate this problem, some survey teams have provided memory joggers, similar to the Bay Area and Tuscon forms shown in Figures 8.3 and 8.4, as well as diaries, so that the respondents could record less detailed information with the memory joggers throughout the diary period, and then fill-in the more detailed diaries from them. The memory joggers are much easier for respondents to carry with them, and so it is believed that respondents are more likely to record trips as they occur, but respondents have found the duplication of effort required with this method to be burdensome.

Both the complete diary form approach and the simplified diary approach are currently being used in the U.S. The relative importance of the different advantages and disadvantages cited above do not seem to support the recommendation of one approach over the other. The survey team can select the best approach for the particular survey population under study through pretesting or through focus groups with potential respondent groups.

Additional Diary Question Design Issues

Often, one or more household member is unavailable or unwilling to complete the survey diaries, so survey teams must develop a set of procedures for accepting or not accepting proxy reports. Usually, proxy reports for children under age 14 are deemed acceptable. However, surveyors have differed on how to address the potential need for proxies for older household members. Most surveyors try to avoid proxy reports for adults. Survey teams have specified that interviewers attempt to reach these respondents as many as four times. If these individuals are still not reached or persuaded to provide the information, some surveyors have sought proxies from within the household. Others have deemed the household contact as incomplete, because proxies consistently report lower numbers of trips.

One final diary design consideration is that some travel surveyors have found that CATI data retrieval can enhance the completeness and accuracy of the diary data. CATI systems can identify missing links in people’s trip patterns (indicating a misrecorded or forgotten trip or activity) and check for data anomalies, such as a person not ending up at home by the end of the period. A common diary completion error is to not record the final trip to home on a day. The CATI system can look for such curiosities, and instruct the interviewer to confirm that the data are correct.

8.3.3 Standard Question Wordings

To permit comparisons across surveys conducted in different locations at different times, it is essential that certain key questions be asked in the same manner. It is also important that the question wording or response definitions in a local survey be consistent with the wording (and definitions) used in a national survey or census, especially for variables that may serve as the basis for sampling, expansion, and checking for bias. Again, the focus of this issue is the minimum questions proposed in Section 8.1.1, but with the addition of some other questions that are frequently used in travel surveys, such as income. Only those questions where the wording is not necessarily self-evident and where variations that could affect the responses given are offered for standardization in Table 8.11.

Table 8.11 Recommended Standardized Question Wordings

Question

Recommended Question Wording

Household Size (H3)

“Including yourself, how many people live at this address? Please do not include anyone who usually lives somewhere else or is just visiting, such as a college student away at school. (If further clarification is needed, include infants and children, live-in domestic help, housemates, roomers.)”

Number of Vehicles (H6)

“How many vehicles are owned, leased, or available for regular use by the people who currently live at this address? Please be sure to include motorcycles, mopeds and RVs.” (As clarification, regular use means “are in working order.”)

It is recommended that travel surveys include a separate question regarding the availability of bicycles for daily travel:

“How many bicycles in working condition are available to members of your household for use in their daily travel?”

Owner or Renter Status (H7)

“Do you own or rent your home?

1 Own/buying (e.g. paying off a mortgage)

2 Rent/lease or

3 Provided by job or military”

Gender (P1)

“Are you (is this person) male or female?”

Disability (P11)

A question should be asked about disabilities that impact travel.

“Do you have a disability or condition that has lasted 6 or more months and which makes it difficult to go outside the home alone—for example, to shop or visit a doctor’s office?”

Activity or Trip Purpose (A2)

For work or work-related activities:

· Volunteer work should be specifically excluded from the definition;

· The clarification should be added that work means work for pay or profit; and,

· Questions should be asked about a second job.

When asking for activities, at a minimum include a category “Other at-home activities.” Advanced practice is to ask separately for activities that could be performed either at or away from home, such as meals, work, shopping (using the Internet).

Number in Traveling Party (A6)

“Including yourself, how many people were traveling with you? How many of these were household members?”

If computer-assisted telephone interviewing (CATI ) is used, it is suggested that the follow-up question regarding number of household members only be asked when the household size is greater than one.

At a minimum, the number in the traveling party should be asked whenever a private car, van, or truck is the mode of travel.

Income

“Please stop me when I get to the category that best describes the total combined income for everyone living at this address for last year:”

Income response categories should match the start and end points used by the U.S. Census, although collapsing across income categories is acceptable.

8.3.4 Sequence of Survey Questions

In conjunction with developing the wording for the survey questions, the survey team needs to determine where and when each question will be asked. The survey team must decide:

· In which survey instrument each question will be asked;

· The order of the questions; and

· The length of the survey.

For the simple mail and simple telephone survey, the first choice is easy. There is only one place the survey question can be – in the mailed survey, or in the telephone interview. However, for the more complicated survey methods, the survey team has the choice of placing questions in the recruitment call, or in the mailed survey, or (for the telephone – mail – telephone) the data retrieval call, or in some combination of the instruments.

Usually, the recruitment call is designed to be short. The call needs to achieve the following:

1. Screen potential respondents to ensure they are in the survey population of interest;

2. Gain the cooperation of potential respondents;

3. Obtain information necessary for the next phases of the survey (mailing address, number of diary instruments needed, best time for data retrieval call, etc.); and

4. Obtain respondent information, so that the survey team can weight the survey results to account for people who were recruited, but who do not complete the survey.

For the recruitment interview, the general order of the questions should be close to this sequence. The recruitment call should pique the interest of respondents, and make them look forward to the mailing, so if time permits, interesting attitudinal and opinion questions could be included. For the most part, the recruitment call should avoid questions that will make respondents uncomfortable. As noted above, it is generally not a good idea to ask about people’s property in these calls. Asking household income questions on the recruitment call has both positive and negative points. By asking the question, the survey team can gain an important variable for non-response weighting, but they do so at the risk of alienating respondents before they have completed their tasks. On balance, it is not recommended.

For survey methods with mailed data retrieval, the mailed survey instrument usually contains most (or all) of the key survey questions, plus the diary (if one is being used). As discussed below, the travel/activity diary and associated materials are generally separate documents from the rest of the survey questions, and usually do not need to be sequenced with the other questions. The other questions should be ordered according to the following principles (Dillman, 1978):

· Order the questions in descending order of how important respondents are likely to perceive them. Begin with questions that are clearly related to the survey topic (recent travel information) and then move on to questions which are only tangentially related (demographics).

· On printed survey materials, organize the questions into related groupings, perhaps separated with titles, “About your travel choices,” or “about your household.”

· Within each group, sort the questions by the tasks they ask of respondents. For instance, all “yes” or “no” questions in one sequence and all the scale questions in another.

· Build ties between the question groupings, so that one flows into the next. If a mail survey has several question groupings which do not seem to relate to one another, the respondents are more likely to feel burdened, and will more easily become confused.

· Put questions that are possibly objectionable to respondents at the end of each grouping of questions.

· Always put demographic questions last, with household income questions at the very end.

If the survey team has special concerns about how two different questions will interact with each other, it makes the most sense to remove one from the mail survey and ask it in the recruitment or data retrieval interview. There is no way to prevent respondents in a mail survey from using (or misusing) information from one question in another, regardless of whether the two questions are near each other, or in which order they appear.

The sequence of the data retrieval call will depend on how much data are to be retrieved directly from mailed materials, and how much will be from new questions to respondents. Most data retrieval calls are designed simply to obtain the information that the respondent has recorded. These calls will follow the mail survey and diary questions precisely, perhaps asking a question or two along the way to clarify or probe responses. It is usually a good idea for the survey team to ask for a few details about the information collected from the mailing, simply to keep respondents actively involved in the interview.

Since both survey cost and non-response levels are related to the length of the interviews and the size of the mail survey, questionnaire length is very important. In general, recruitment calls are brief, between eight and 15 minutes. Recruitment call times for recent [PS6] household travel/activity surveys have been:

<!--[if !supportLists]-->· <!--[endif]-->Portland, 1995	8-9 minutes
<!--[if !supportLists]-->· <!--[endif]-->Beaumont, TX, 1993	7 minutes
<!--[if !supportLists]-->· <!--[endif]-->North Carolina Research Triangle, 1994	10 minutes
<!--[if !supportLists]-->· <!--[endif]-->Baltimore, 1993	9 minutes
<!--[if !supportLists]-->· <!--[endif]-->NPTS, Pretest 1994	12 minutes

Interviewers and respondents exchange information during this call, so it is more likely to be interesting for respondents.

Data retrieval calls are substantially longer. Retrieval calls for one-day diary surveys usually average 30 to 50 minutes per household. Two-day diary survey retrieval average times range from 40 to 75 minutes per household. Since these figures represent averages, larger households have had to have been on the phone for very long periods of time. In some recent surveys, data retrieval has been spread over two or three calls. Data retrieval calls are also less interesting for respondents than recruitment calls since they usually are simply being asked to read their written response. The effect of survey length on non-response is highly variable, depending on the level of interest the respondent has in the survey topic, but most household travel/activity surveys are well into the range of being too long for a substantial number of people. Besides the obvious approach of cutting survey questions out, two procedures have been used to shorten data retrieval calls:

· Move questions from the retrieval interview into the recruitment interview; and

· Split the sample so that respondents are only asked a subset of questions. Different respondents are asked different questions so all the questions are asked of at least some of the sample.

Both of these approaches have clear limitations, however. Recent survey efforts that have tried to ask too much of respondents in the recruitment interview had lower respondent cooperation rates, and one effort ran into public relations problems because respondents were uncomfortable with the level of detail they were being asked to provide in the initial call. Splitting the sample is usually very difficult for household travel activity surveys, because almost all the retrieval time is spent on the diary information, which cannot be split practically.

The survey team must ensure that every effort has been made to keep the respondents interested in the survey, and to provide an organized interview as soon as possible.

8.4 Survey Instruments and Materials

Because household travel/activity surveys are generally used to collect a wide range of detailed data, the survey team is likely to need to develop a number of different survey instruments and materials. Table 8.12 summarizes the most common survey materials. As the table indicates, the key questions in designing any of these materials are:

· Simply put, what is the purpose of the survey material or instrument?

· Who are the “users” of the survey material or instrument?

It is important to remember that the answer to the latter question usually includes more than one group, who sometimes require very different design decisions. Designing survey materials with only respondents in mind can lead to interviewer errors and coding, editing, and cleaning problems. These problems will almost certainly show up in the survey cost and scheduling.

The following sections describe the development of the survey materials used in the most common household travel/activity survey methods.

8.4.1 Materials for Any Survey Method

8.4.1.1 Pre-Notification Letter, Brochure, Postcard, or Interview Script

The primary purpose of pre-notification is to inform potential respondents of the upcoming survey and to persuade them to participate. Pre-notification is in effect a sales effort. The survey team wants to convince a potential respondent that the household travel/activity survey effort is:

· Legitimate;

· Important;

· Worth looking for in their mail; and

· Worth their time and effort to complete.

Table 8.12 Survey Materials Commonly Used in Household Travel/Activity Survey

To convince respondents of these facts, the following design guidelines are suggested:

· The pre-notification phone call should be made, or the letter, brochure, or postcard should be sent so that it arrives three or four days before the survey materials.

· If pre-notification is by mail, the document needs to be short and to the point. It should convey its message to the reader in less than a minute.

· The document should look official. If possible, it should be printed on letterhead or have the name of the sponsoring agency or agencies prominently displayed.

· The document can either be from the agency sponsor or another interested agency that might be more recognizable to respondents.

· If possible, the document should be signed by a recognizable public figure or an elected official.

· The document should explain the reasons for the survey in plain language, and stress the confidentiality of the survey effort.

· The document should provide the name and telephone number of someone who can answer questions about the survey and confirm its validity. In most cases, almost no one will call the number, but if potential respondents see that it is offered, it will help them believe in the validity of the survey.

· For those people who do call the number, it is best if the number is for someone at the sponsoring agency, rather than at a private market research firm. Some survey contractors can provide a phone number that can be covered by an answering machine with the name of the agency on the message. Representatives of either the survey firm or the sponsoring agency can return the few calls that will come in. It is useful to draft a list of the most frequently asked questions with the response for use by the sponsoring agency and/or the contractor. Whoever receives the call will answer respondent questions in a consistent manner.

A secondary use of the pre-notification material is to use it to identify bad addresses for which mail is undeliverable. To do this, the pre-notification letter must be sent by first-class mail with a return address (many surveyors believe all materials should be sent by first-class mail to separate the materials from junk mail). Mailing the pre-notification material by this method helps the survey team to identify the percentage of bad addresses in the sampling frame, and allows the team to save some postage costs by sending the survey materials only to valid addresses.

Figure 8.11 shows an example pre-notification letter from the 1996 Oregon (statewide) Travel [PS7] Behavior Survey.

8.4.1.2 Thank You Card

From a myopic data collection point-of-view, once the survey team has been able to retrieve complete and seemingly valid data from a respondent, further contact with that respondent is superfluous. However, in some instances, it is politically advantageous for an agency to send a postcard thanking the respondent households for their efforts. These cards build a sense of good will between the respondents and the agency, which might be valuable in future planning efforts that are increasingly relying on citizen participation.

The cards also help to confirm for respondents that the survey effort was legitimate and that their information was valuable. A survey team that is considering the possibility of using the household travel/activity survey as part of an ongoing panel design should certainly consider the use of thank you cards.

8.4.1.3 Follow-Up Letters and Postcards

Usually, if a potential respondent household fails to respond to the initial mail survey, they are sent one or more reminder letters or postcards. These materials are designed to:

· Remind the respondent households who have completed the survey, but not yet returned it to mail it back;

· Re-stress the importance of the survey effort and the importance of the particular household receiving the letter or card; and

· Provide potential respondents with another opportunity to complete the survey by assigning new diary periods, and/or by inviting them to contact the survey team by phone.

The first set of follow-up letters or cards stress the reminder message. Often, a simple postcard is sent that states that the respondent household’s survey has not yet been received, and asks the household please make sure that they do not forget to send it.

With each successive follow-up round, the content of the letters and/or cards shifts away from the reminder message toward a more persuasive message. Later follow-up mailings tend to be either letters stressing the importance of the household’s participation or letters combined with a new set of survey materials. Richardson, Ampt, and Meyburg suggest that each successive follow-up mailing be sent in different color and style of envelope so that they are less easily dismissed by potential respondents.

Figure 8.12 shows an example follow-up letter from the 1991 CATS [PS8] household travel mail survey.

8.4.2 Materials for Surveys with a Mail Component

8.4.2.1 Envelope for the Survey Mailing

The first challenge in getting households to respond to the survey is to get one of the household members to open and read the survey package. A very high percentage of direct mail is thrown away without ever having anyone open it. Unfortunately, mailed household travel/activity surveys generally look very much like direct mail when they show up in people’s mailboxes.

The travel survey team needs to take steps to get their envelope opened. One simple step for survey efforts with pre-notification or where the mailing is taking place after the respondent has been contacted by phone is to describe the envelope to the respondent before it is mailed to them, “You will be receiving the mail survey in a large blue envelope with our return address on it.”

In addition, Dillman suggests the following steps to separate the mailing from junk mail (Dillman, 1978):

· Use as small an envelope as possible to limit postage costs and to avoid the “bulk” image;

· Type the name (use specific people’s names, rather than only the family name whenever possible) and address directly on the envelope, rather than on address labels;

· Do not embellish the envelope with messages, like “dated materials inside” or “immediate reply requested”; and

· Use actual stamps, rather than metered or bulk rate mail, even though postage costs will be higher (not all surveyors see this as a justified expense, particularly if respondents have been pre-notified of the effort).

The design of return envelopes for mailback questionnaires is less important than the survey package envelope, because the respondent would use it only if she or he has decided to complete the survey. The important design considerations for these envelopes include the following:

· The envelopes should be provided;

· The envelopes should be prepaid through the use of business reply mail[PS9] ; and

· The return address should be for the sponsoring agency, if possible. If the replies are going to be sent to a survey contractor, respondents should already be aware of the name of the firm from the cover letter and other survey materials or interview contacts. Unless it is absolutely necessary, envelopes should not be sent to addresses outside the study region.

8.4.2.2 Cover Letter for the Survey Mailing

The cover letter for the survey mailing has been shown to be an important survey element. It must perform the same functions as the pre-notification letter, plus introduce the attached survey materials. The cover letter should be no more than one page, and should be both motivational and informative.

Figure 8.11 Example Pre-Notification Letter for the Oregon Statewide Activity Survey

Cover letters should have the following general outline (Dillman, 1978):

· Top of Page – Official letterhead of the sponsoring agency;

· Date – Exact (and correct) date of transmittal;

· Address – Name and address, similar to any business letter;

· Salutation – Specific salutation (e.g., Dear Ms. Thompson:), if possible (sometimes not possible because only the family name is known or because the person’s sex is not obvious from his or her name), otherwise no salutation at all;

· Paragraph 1 – 1) topic of the study; and 2) social usefulness of the study;

· Paragraph 2 – 1) why recipient is important to the study; and 2) who in the household should participate;

· Paragraph 3 – 1) promise of confidentiality; and 2) explanation of privacy procedures and serial numbers on survey forms;

· Paragraph 4 – 1) usefulness of the study results; and 2) explanation of incentive (if one is being employed);

· Paragraph 5 – What to do if questions arise;

· Closure – Thank you;

· Signature – Business letter-type signature block, with typed name and title under the signature, and with an original handwritten signature in blue ink[1].

If possible, the signature should be from a well-known elected official. Recent household surveys have had cover letters signed by Governors and U.S. Senators. Of course, when higher profile people are used for the signature, the ability to provide an actual handwritten signature is eliminated, although a digitally scanned signature in color can be quite as effective.

Figure 8.13 provides an example of a recent household travel/activity survey cover letter. While this letter does not follow the suggested format with precision, it covers the main points that need to be included, and appears to be an effective communication tool.

With address-based sampling, it is possible to individualize the cover letters so that the study’s benefits can be defined specifically for the individual receiving the letter. In an ongoing [PS10] Bay Area survey, this individualization is being done as follows:

“Dear <<respondent name>>:

Do traffic conditions on <<main corridor near respondent’s home>> sometimes concern you? What about the number of parking spaces at the <<nearest transit station>> station?”

Individualization can also take place later in the cover letter.

“You are important to the success of this study – no matter how much or how little you travel. You may be the only household on <<street>> that has been randomly selected for participation in the study.”

8.4.2.3 Survey Fact Sheet

The design of the cover letter involves a tradeoff between supplying more information and the need to keep the letter short and punchy so that it gets read, instead of just skimmed. Sometimes, when survey teams feel it is necessary to provide additional information about the survey, they include a one-page fact sheet or pamphlet. Figure 8.14 shows the survey fact sheet from a recent[PS11] household activity survey.

8.4.2.4 Survey Questionnaire

The development of mail survey questionnaires and the mail component of telephone-mailout-mailback surveys requires the survey team to consider both the extremely difficult wording issues discussed above and issues of survey layout, as well. The quality of the layout of self-administered survey will affect:

· Overall non-response rates;

· Item non-response rates;

· Response quality;

· Survey coding quality; and

· Data entry and editing efficiency.

·  

· Figure 8.12 Example Follow-up Letter for a Chicago Mail Survey

Figure 8.13 Cover Letter for the Mailing of a Boston Area Household Activity Survey

The importance of questionnaire layout on self-administered survey materials is illustrated by the results of an actual household products panel survey effort in 1980 (Mayer and Piper, 1982). Figure 8.15 shows two mail survey questions that were asked of the same 4,000 respondents about the same products within a year of each other. The only difference between the surveys in which the two questions were offered was the questionnaire layout.

The question on the left produced results which were consistent with the expectations of the study sponsor. The question on the right produced the dramatically different results shown in the figure. These results were found to be invalid in clarification telephone calls. After discovering the problem, the survey team theorized that the line on which the other brand was to be entered was too close to the box around the question, and that respondents counted from the bottom to check their brand.

It is fortunate (or perhaps unfortunate) for the survey team that the study sponsor happened to be the manufacturer of one of the two products in question. The survey team had a strong sense of the relative market shares of the two products from previous waves of the panel, so the problem was discovered and rectified. For travel surveys, there are generally no recent survey data to test the validity of questions, and so survey teams need to be as careful as possible in designing the layout of questionnaires.

Fowler provides the following guiding principles for developing self-administered questionnaires (Fowler, 1988):

1. A self-administered questionnaire should be self-explanatory. Reading instructions should not be necessary, because they will not be read consistently.

2. Self-administered questionnaires should be restricted to closed-end answers whenever possible. Checking a box or circling a number should be the only task required. When respondents are asked to answer in their own words, the answers are usually incomplete, vague, or difficult to code, and therefore are of only limited value as measurements.

3. The question forms in a self-administered questionnaire should be few in number. The more the questionnaire can be set up so that the respondent has the same kinds of tasks and questions to answer, the less likely it is that respondents will become confused; also, the easier the task will be for the respondents.

4. A questionnaire should be typed and laid out in a way that seems clear and uncluttered. Photoreduction, or other strategies for putting many questions on a page, actually reduces the response rate compared with when the same number of questions are spaced more attractively over more pages.

5. Skip patterns should be kept to a minimum. If some respondents must skip some questions, arrows and boxes that communicate skips without verbal instructions are best.

6. Provide redundant information to respondents. If people can be confused about what they are supposed to do, they will be.

Figure 8.14 An Example of a Form to Provide Additional Information on the Survey to Respondents

Figure 8.15 An Example of the Importance of Layout on Self-Administered Questionnaires

Because household travel/activity surveys often collect very detailed information from all household members, they can become quite repetitive and boring for respondents. Survey designers have sought ways to use graphical and tabular formatting to both improve the presentation of the survey materials and to ease the burden on respondents.

The tabular and graphical formats are used both on mailed survey instruments and on interviewer scripts. The formats improve the appearance and user-friendliness of the survey forms for respondents and interviewers, because they help organize the information logically and unambiguously. On the other hand, the formats tend to require shortened forms of questions and the informal wording of questions. The survey team needs to be especially careful when using tables or graphics that they do not introduce any of the wording problems discussed above.

Still, if the tables and graphics are carefully-designed, their advantages far outweigh their potential disadvantages. Survey teams should seek opportunities to present the survey questions in clear, logical, and visually interesting formats. The widespread availability of desktop-publishing and graphics software greatly enhances the ability of survey teams to use special formats.

Survey teams should consider the use of colored paper and different ink colors to enhance the attractiveness of the survey materials, to improve the organization of the different survey materials in the survey packet, and to aid in data entry. Using different colored paper for each survey form makes it easier for respondents to keep them straight, and provides a useful mechanism for simplifying written instructions or data retrieval calls. For instance, an interviewer could say, “Now I would like you to read me the answers you filled out on the pink form.”

Using ink colors other than black has a few advantages, as well. First, the forms can be made more visually interesting for respondents. In addition, since many respondents will fill out the forms in black ink, or pencil using some other color for questions will make it easier for data entry specialists to pick up the answer, or for the respondents, themselves, who could be asked to read what they wrote. The recent Research Triangle survey used green ink to improve the data entry.

8.4.2.5 Household and Vehicle Forms

One common approach for improving the layout of survey questionnaires in household travel/activity surveys is to separate out groups of related questions into stand-alone survey forms. Travel diary data are almost always recorded in separate diary booklets, and, increasingly, survey teams are using household and/or vehicle forms, that seek the same type of information for all members of the household and all the vehicles available to the household.

Figures 8.16 and 8.17 show examples of different formats used to collect household and vehicle data from respondents. Each form is designed to be a separate one- or two-page instrument. In surveys that use such forms, respondents are generally asked to complete several different forms. For instance, a household survey might include a household form, a vehicle form, a travel diary, and a stated response exercise, which are all separate forms. This design is appealing to respondents, because it logically categorizes the questions, it improves the visual layout of the questions, and it allows them to feel a sense of accomplishment as they finish off individual forms.

8.4.2.6 Travel or Activity Diary

Virtually all recent household travel/activity surveys seeking diary information have presented the diary questions in an easy-to-complete tabular format, rather than as a series of unformatted questions. Diaries are usually organized in booklets, one per household member, that record trips or activities, one (or a few) per page.

The several diary pages shown in Figure 3.2 and in Figures 6.4 through 6.11 illustrate the many different diary layouts. The Chicago (Figure 3.2), and Ontario (Figure 6.11) surveys collect information about each activity or trip in a single column.[PS12] This columnar format was first developed for a 1973 West German trip-based diary survey. The design has come to be known as the “KONTIV” format, and has been used in many surveys since[PS13] .

Figure 8.16 Household Form from a Recent[PS14] Boston Survey

Figure 8.17 An Example Vehicle Information Form

The Bay Area (Figure 6.4), Tucson (Figure 6.5), and New York MTA (Figure 6.7) [PS15] diaries collect trip information in rows. This format is easy for respondents to follow, but it is limited in the amount of data it can collect. The Bay Area format is actually a memory jogger, in which only the key trip details are collected. The telephone data retrieval call collected the more detailed information.

The row-wise format uses space efficiently. The MTA diary questionnaire was produced in a small booklet measuring only 8.5 inches by 3.5 inches. This convenient size makes it easier for respondents to carry the questionnaire with them as they travel.

The other diary pages shown, the British diary (Figure 6.6), the Research Triangle diary (Figure 6-8), the Detroit diary (Figure 6.9), and the Portland diary (Figure 6.10)[PS16] use a diary format for which each activity or trip is recorded on a single page of the diary booklet. This format is the most graphical and readable of the three. It is especially well-suited for surveys relying on mailback data retrieval. The British survey was a self-completion instrument and, therefore, required the added clarity of the format. In addition, although the Detroit survey used telephone data retrieval, the form used for this survey was based on an earlier activity-based diary that appeared in a telephone-mailout-mailback survey in Boston.

8.4.2.7 Memory Jogger

Because of the level of detail to be reported, household travel/activity survey diaries usually need to be several pages. Although survey designers have produced the diaries in attractive booklets, it is not generally convenient for respondents to carry the booklets with them during the diary period. Survey designers suspect that a high proportion of respondents complete their diaries at the end of the diary period, rather than during it as they asked to do. This can lead to recall problems and inaccurate trip or activity reports.

To reduce this problem, past survey teams sometimes included a memory jogger, a one page trip or activity record where only the most important details of travel are recorded. The memory joggers were designed to be more convenient for respondents to keep with them as they travel around, so it was believed that respondents would be more likely to use them.

For most of these studies, at the end of the diary period, the respondent was asked to copy the information from the memory jogger into the diary and to supply the missing details that the diary required. Unfortunately, it has been found that in many cases, respondents complete either the memory jogger or the diary, but not both. Therefore, it is recommended that memory joggers not be combined with complete diaries.

As discussed previously, for some studies with telephone data retrieval, a simplified diary is the only travel information the respondent is asked to complete. Details of the trips and activities on the simplified diaries are obtained directly through the telephone interview. In these designs, past examples of memory joggers are likely to be more useful in terms of instrument design than complete diaries.

Figure 8.18 shows an example memory jogger from the recent [PS17] Boston survey.

8.4.2.8 Reminder Card

Survey teams that ask respondents to complete diaries or to record certain information on particular days often include a card reminding respondents of their day-of-record. The respondent is encouraged to hang the card in some central location, such as on the refrigerator, to remind themselves (and other household members if data are being sought from entire households) that they need to perform their survey duties that day.

Figure 8.19 shows a reminder card from a recent survey effort.

Figure 8.18 An Example Memory Jogger

Figure 8.19 An Example Diary Reminder Card

8.4.2.9 Cover Letter for Follow-Up Survey

In general, the cover letter for a follow-up mailing of a household travel/activity survey needs to cover the same points as the cover letter for the original mailing. However, this letter should stress the motivational points made in the earlier letter, rather than the informational points. The key message of this letter is that the sponsoring agency needs this household’s travel information.

8.4.2.10 Follow-Up Survey Materials

In general, the survey materials sent in follow-up mailings are the same as for the main mailing. However, the survey team needs to ensure that any dated materials in the original mailing are changed to reflect the follow-up mailing dates. Of particular importance in this regard are any dates associated with travel or activity diary periods. Usually, survey teams establish new diary periods for follow-up survey respondents. If this is the case, the survey team needs to ensure that none of the materials sent in the follow-up reference the old date.

8.4.3 Materials for Surveys with an Interview Component

Even though respondents never see any of the survey materials related to the telephone survey, the design of these materials can have a dramatic impact on the quality of the overall survey effort. Household travel/activity surveys often involve very long and involved telephone inter-views. It is essential that these interviews go as smoothly as possible, and for this to happen, the interview forms need to be designed to relieve the interviewer of as much burden as possible (Lavrakas, 1987).

If a survey contractor is used on the survey, the manager will be able to convert the survey team’s questionnaire into a usable interview script for either a PAPI or CATI technique. If no survey firm is involved, the survey team should work closely with the telephone interviewers prior to any data collection to establish the most useful presentations of questions and responses. In almost all cases, surveys that do not use survey contractors will use PAPI techniques, so issues that the survey team should decide on with the interviewers are (Lavrakas, 1987):

· Standard questionnaire typeset conventions (e.g., questions are typed in capital and lower case letters, instructions are in capital bold letters, etc.);

· Techniques for recording answers;

· Rules for recording open-ended items;

· Standard skip pattern conventions (e.g., color-coded responses and questions); and

· The use of tables and matrices to record data items.

REFERENCES (to be consistently formatted by Alyson Clarke)

Axhausen, K.W. Travel Diaries: An Annotated Catalogue, University of London Centre for Transport Studies Working Paper, November 1994.

Backstrom, Charles and Gerald Hursh-Cesar. Survey Research, 2nd edition, John Wiley & Sons, 1981, pp. 119-122.

Dillman, Don. Mail and Telephone Surveys: The Total Design Method, John Wiley & Sons, New York (1978), pp. 123-125, 165-175.

Fowler, Floyd J., Jr. Survey Research Methods Revised Edition, Sage Publications (1988) p. 102.

Jones, Peter. Summary of the Diary Surveys Workshop in Ampt, E.S., Richardson, A.J., and Brög, W. (1985) New Survey Methods in Transport, VNU Science Press: Utrecht, The Netherlands, pp. 36-39.

Lavrakas, Paul. Telephone Survey Methods, SAGE Publications, 1987, pp. 142 146.

Mayer, Charles S. and Cindy Piper. A Note on the Importance of Layout in Self-Administered Questionnaires, Journal of Marketing Research, Vol. XIX (August 1982), pp. 390-391.

Richardson, A.J., Elizabeth Ampt, and Arnim Meyburg. Survey Methods for Transport Planning, Eucalyptus Press, Melbourne 1995, p. 303.

Stecher, Cheryl C., Stacey Bricka, and Leslie Goldenberg. Travel Behavior Survey Data Collection Instruments, Resource Paper for Household Travel Surveys: New Concepts and Research Needs Conference, Irvine, CA (March 1995).

Stopher, Peter R. Use of an Activity-Based Diary to Collect Household Travel Data, Transportation Vol. 19 (1992), pp. 159 176.

Szalai, A. (ed.). The Use of Time. Mouton (The Hague), 1972.

[1] For large survey efforts, a scanned signature would be required to provide an “original-like” signature.