Tag: Surveys


An Artwork by Cheri Samba is at least a description of using general measures (bow and arrow, skiet-rekker) to target something specific (mosquitos).

Is the National Household Travel Survey a statistical analysis of household travel? Or is it an analysis of household travel within transport theory and practice? The very general don’t do well when specificity is important.

These two questions are important for two reasons: (1) transportation surveys vary with methodology and intent and (2) relatable travel details are crucial for more specific evaluation. This short comment on the NHTS 2018 reflects on the second point with a focus on the “main mode” concept in the NHTS 2013 survey and travel literature.

“Main Mode”

The glossary in the NHTS 2013 describes the main mode as the “highest” mode of travel used in a hierarchy: (1)train, (2) bus, (3) taxi, (4) car driver and so on. Throughout the survey this hierarchy takes various forms. One may ask how the “height” of a mode is determined; or if an individual used a taxi to join a car pool and was a driver that week; or if the taxi takes longest end of the trip but the respondent also used a bus. This “height” bar is consistent with vehicle size maybe, but not with travel theory.

Trip Chain illustration from the 2011 Massachusetts Travel Survey

The mere fact that trips may be multimodal in various ways and compositions of time, continence, purpose and price in each mode. The main mode may be described in terms of length of time; price; most important leg of the trip (i.e. no other mode penetrates the area) — among others. Illustrated above is a tour to and from home and work, respectively. A trip is a part (leg) of a journey. A journey is like traveling from home to work. A travel survey should capture the touring patterns on a national level — in a way that is affordable to the public sector; administratively viable from a data collection and training perspective; and lastly it should communicate national transport policy goals. The theoretical details matter or else Cheri Samba’s work will go unnoticed. 😂

Understanding the Details

In transportation research, the details are starting to matter more and more. Whether it is observing travel over time; travel schedules; understanding preferences and ratings; or even language and other cognitive generators (I.e behavior). Capturing the right sequence of this behavior is also important — as respondents indicate. And finally, using the correct mix of measures to evaluate true attitudes.

Source of the Painting Above. Book: Africa Explores 20th Century African Art, 1991.

In the NHTS2018, it’s rather clear that the survey is designed descriptively and not systematically within transport theory and practice.

The time frame is a 7 day reflection and respondents seem to describe travel demographics. Travel demographics are descriptive indicators of travel; not measures of cause and effect by intent, only through manipulation. What is collected is largely a statistical description of travel. For instance in the trip specific section (5 and 6), main mode is specified in business trips(5) as the one spent the most time in. In the section on other travel patterns(6) it is called “main mode” without definition. Statistically users of the survey know which modes people used but from a mobility and access perspective the respondents make trips to make these business and ‘other’ trips. It is virtually clear we don’t know how business, overnight and day trips manifest — which modes feed intra and interregional migration? We won’t know this on a national level.

Shots of Some of the Questions that emphasize “main mode”.

Interestingly, work and education trip specific sections of the survey offer an apparently hierarchical approach to identifying the main mode of travel. These sections have a stronger blanket approach and are interpreted hierarchically based on the survey design and not the respondents indication. The “main mode” concept dissipates into the order of the transport modes: if Metrorail is 01 and commuter taxi is 06 and a bus (03) is used in the final leg then the main mode will be 01, Metrorail. This is statistically sensible but in practice it dilutes the overarching mobility picture: which mode is used first? How much time is spent in each? Is there a transfer? How long is this transfer(a component of waiting time)? The questioning in the NHTS 2013 treats travel behavior as if it were a demographic exercise, rather than a structured chain of choices that individuals make based on their past and current circumstances.

Shots of some of the questions that apply the heirachical approach to estimate “main mode”.

Way Forward

Without forging, understanding the trip chain is a rather important for travel planning. Using a statistically prudent avenue to estimate main mode in the work and education trip sections only shows main modes within a predestined hierarchy — not the respondent specific main mode by some measure(i.e. Travel time). With the advent of improving integrated transport systems the main mode methodology used for business and other (overnight and day) trips excludes the possibility of capturing multimodal travel. Without mentioning the inconsistency around the “main mode” concept within the survey, the trip chain and how it manifested are not part of the travel survey. This is not because such measures are not important to household travel, but because the NHTS 2013 survey appears to be a statistical investigation of travel demographics by intent. If the next survey intends to translate mobility and access policies and strategies in South Africa then it may need to turn to a travel theory based approach to the survey — which is statistically more interesting.