The airline industry has suffered deep declines as a result of the COVID-19 pandemic. In addition to governments placing restrictions on air travel, many passengers are afraid of flying due to anticipated health risks. Consumer research can help airlines find perferred seating configurations. Will air travelers pay for social distancing?

We weigh competing objectives when booking airline flights online. Nonstop flights are more convenient than one-stop or multi-stop flights. Amenities differ across airlines and seating sections within planes, with first-class seating being better than business class, and business class better than coach. Just as ticket prices vary across travel and seating options, they vary across the dates of travel and days of the week, with holidays more expensive than non-holidays and weekdays more expensive than weekend days. There are many factors to consider when making airline travel plans, and there are trade-offs between travel benefits and ticket prices.

Suppose a traveler must fly from Los Angeles to New York City for a family emergency. Getting to the destination quickly is essential, so the traveler will be choosing a nonstop flight. This is a 2,500-mile flight that takes five hours.

A typical Boeing 737-800 is about twelve feet wide in the main cabin. There are fourteen rows of economy coach seats with six seats per row, three on each side of the aisle. When all seats are available for booking, there are 14 x 6 = 84 economy coach seats with rows numbered 20 to 33. To accommodate travelers’ preferences for social distancing, suppose airlines are evaluating three reduced seating configurations: two of three seats available, one of three seats available, and one of six seats available. This figure shows various reduced seating configurations:

Suppose prospective travelers are shown the ranking task shown in this figure:

Those familiar with experimental design will note that seating position (window, middle, or aisle) is inextricably associated (confounded) with seating configuration. Neither reduced seating configuration has middle seats, and the one-out-of-every-six-seats-available configuration has no aisle seats. It is not possible to construct a completely-crossed design involving price, seating position, and seating configuration, which presents a problem for estimating part-worths associated with individual attributes. Alternatively, we can think of seating position/configuration as a single attribute with six levels.

It may be better to think of seating-position/seating-orientation as six separate treatments combinations. Alternatively, we could characterize each configuration by its minimum distance between pairs of passengers.

Those familiar with experimental design will note that this ranking task has four seating options (all seats available and three reduced seating configurations) and four price levels ($175, $350, $525, and $700). A completely-crossed 4-by-4 design would have sixteen treatment combinations. The eight items in the ranking task correspond to cells in a balanced fractional factorial design with each seating option and each price level represented twice.

Consider one traveler’s responses to the ticket ranking task, as shown in this figure:

As the traveler ranked airline tickets, she no doubt weighed ticket prices against a desire to maintain social distance between herself and fellow travelers. Ranks reflect preferences. What do this traveler’s ranks suggest about her preferences? How much is she willing to pay to have more distance between herself and others?

If the woman were traveling between cities less distant from one another, say between Washington D.C. and New York City, then she might consider driving rather than flying. Ranks or choices across modes of transportation reflect a person’s willingness to fly.

How could we design a choice-based conjoint study to evaluate willingness to fly? What type of analysis would be needed to reveal differential preferences across groups of travelers identified by age or income level?