Understanding the impact of online choice architecture on vulnerable customers

Participants will be going through an online survey experiment involving online marketplaces. Participants will be randomly assigned to different versions of the marketplaces, some of which will feature reference pricing, reference pricing with sensory manipulation or reference pricing with sensory manipulation and discount timers. The study aims to understand how these online choice architecture features influence consumer decision-making.

Participants will be randomly assigned with equal probability to one of four experimental conditions and will remain in that condition for all three shopping tasks (broadband, gym, hotel):

Control: Marketplaces without reference pricing, sensory manipulation, or discount timers.
T1: Main marketplace featuring reference pricing.
T2: Main marketplace featuring reference pricing and sensory manipulation.
T3: Main marketplace featuring reference pricing, sensory manipulation, and discount timers.
For each shopping task, participants will first be presented with the needs of a hypothetical person and then redirected to a realistic replica of a website. This website will also feature a banner that, when clicked, will redirect them to a competitor website offering equivalent products/services but not featuring the specific online choice architecture elements of their assigned treatment condition (i.e., the competitor site is always a 'control' version without these OCAs).

2025-05-30

2025-06-27

- Correct choice: Binary variable for choosing the correct plan/membership/hotel in each of the three marketplaces (broadband, gym, hotel).
- Proportion of correct choices: Proportion of a participant's choices across the three marketplaces that were correct (0, ⅓, ⅔, 1).
- Correct or approximately correct choice: Binary variable for choosing the correct or approximately correct option in each marketplace (where applicable).
- Proportion of correct or approximately correct choices: Proportion of a participant's choices across marketplaces where they purchased the correct or approximately correct choices (0, ⅓, ⅔, 1).
- Decoy choice: Binary variable for choosing the decoy option in each marketplace.
- Proportion of decoy choices: Proportion of a participant's choices across marketplaces where they purchased the decoy (0, ⅓, ⅔, 1).
- OCA-designated choice: Binary variable for choosing an option in each marketplace that receives an Online Choice Architecture (OCA) element in the treatment conditions.
- Proportion of OCA-designated choices: Proportion of a participant's choices across marketplaces that were OCA-designated options (0, ⅓, ⅔, 1).

Correct choice: The products and services chosen by a participant across the three online marketplaces, coded:
As a binary variable, whether the participant chose the correct plan in the broadband marketplace (0 for incorrect)
As a binary variable, whether the participant chose the correct gym membership in the gym marketplace (0 for incorrect)
As a binary variable, whether the participant chose the correct hotel in the hotel aggregator marketplace (0 for incorrect)
Proportion of participant’s choices across marketplaces that were correct - 0, ⅓, ⅔, 1

Alternative definitions of correct choice
As a binary variable, whether the participant chose the correct or the approximately correct choice in the broadband marketplace (0 for incorrect)
Proportion of participant’s choices across marketplaces where they purchased the correct or the approximately correct choices - 0, ⅓, ⅔, 1

Decoy choice
As a binary variable, whether the participant chose the decoy in the broadband marketplace (1 for decoy purchase, 0 otherwise)
As a binary variable, whether the participant chose the decoy in the gym marketplace (1 for decoy purchase, 0 otherwise)
As a binary variable, whether the participant chose the decoy in the hotel aggregator marketplace (1 for decoy purchase, 0 otherwise)
Proportion of participant’s choices across marketplaces where they purchased the decoy - 0, ⅓, ⅔, 1

Choice of OCA-designated product/service
As a binary variable, whether the participant chose a plan in the broadband marketplace that receives an OCA in the treatment conditions (1 if an OCA-designated product is selected, 0 otherwise).
As a binary variable, whether the participant chose a gym membership in the gym marketplace that receives an OCA in the treatment conditions (1 if an OCA-designated product is selected, 0 otherwise).
As a binary variable, whether the participant chose a hotel in the hotel aggregator marketplace that receives an OCA in the treatment conditions (1 if an OCA-designated product is selected, 0 otherwise).
Proportion of participant’s choices across marketplaces that were OCA-designated products/services - 0, ⅓, ⅔, 1.

Shopping around: visiting competitor website
Shopping around: purchasing and finishing journey on competitor website

Shopping around: visiting competitor website
As a binary variable, whether the participant visited the alternative broadband marketplace (1 if they visited the alternative, 0 otherwise)
As a binary variable, whether the participant visited the alternative gym marketplace (1 if they visited the alternative, 0 otherwise)
As a binary variable, whether the participant visited the alternative hotel aggregator marketplace (1 if they visited the alternative, 0 otherwise)
Proportion of tasks where the participant visited the alternative marketplace - 0, ⅓, ⅔, 1

Shopping around: purchasing and finishing journey on competitor website
As a binary variable, whether the participant purchased and finished the broadband journey in the alternative broadband marketplace (0 if they finished the journey in the main marketplace)
As a binary variable, whether the participant purchased and finished the gym journey in the alternative gym marketplace (0 if they finished the journey in the main marketplace)
As a binary variable, whether the participant purchased and finished the hotel aggregator journey in the alternative hotel aggregator marketplace (0 if they finished the journey in the main marketplace)
Proportion of tasks where the participant purchased and finished the journey in the alternative marketplace - 0, ⅓, ⅔, 1

Participants will be randomly assigned to one of four experimental conditions. Each participant will complete three online shopping tasks. The study will investigate how variations in online marketplace design influence consumer choices. We will also examine if effects differ for pre-identified vulnerable consumer groups.

The experiment will be conducted via an online survey, coded in Qualtrics, with participants recruited through the panel provider Dynata. The survey will begin with a welcome screen detailing eligibility, the right to withdraw, and that compensation depends on completing the survey and passing attention/comprehension checks.

Next, participants will answer demographic questions (age, gender, education, UK region, local urbanisation rate). This will be followed by pre-experimental questions on their finances (income, employment, benefit receipt), mental health (MHI-5 inventory, diagnosis, disability), shopping habits (online shopping frequency, stress levels, bargain hunting), financial literacy (Lusardi and Mitchell's 2008 3 question inventory), and time discounting preferences (2 standard switching decision task questions).

After answering these questions, participants will receive instructions for the main experimental tasks. They will be asked to complete three shopping tasks: selecting a broadband deal, a gym membership, and a hotel stay for a hypothetical individual with predefined needs. Participants will be instructed to choose the cheapest option that meets all the specified needs of that person. Comprehension of these instructions will be tested with one or two comprehension checks; a second question will appear only if the first is answered incorrectly; failure to answer at least one of the comprehension checks correctly will result in the participant being screened out.
Participants will be randomly assigned with equal probability to one of four experimental conditions and will remain in that condition for all three shopping tasks:

Control: Marketplaces without reference pricing, sensory manipulation, or discount timers.
T1: Main marketplace featuring reference pricing.
T2: Main marketplace featuring reference pricing and sensory manipulation.
T3: Main marketplace featuring reference pricing, sensory manipulation and discount timers.
These conditions were selected for their policy relevance. Participants will not be aware of their assigned condition.
For each shopping task, participants will first be presented with the needs of a hypothetical person, and then redirected to a realistic replica of a (1) broadband, (2) gym, (3) hotel aggregator website (i.e., main marketplace). This website will also feature a banner that, when clicked, will redirect them to a competitor website offering equivalent products/services but not featuring reference pricing, sensory manipulation or discount timers.

Each of the 3 tasks is expected to take approximately 3 minutes (9 minutes total). Following the three marketplace tasks, participants will answer three simple geography trivia questions as an attention check.
A post-experimental survey will then assess their marketplace task experience: stress levels, post-task mood, task difficulty per marketplace, decision confidence, value-for-money perception, feelings of deception, noticing reference prices, and the influence of reference prices.

Qualtrics built-in randomiser, with equal probability of being assigned to each experimental condition

Individual-level randomisation to one of four groups (control, treatment 1, treatment 2, treatment 3), which occurs as participants enter the online marketplace section of the survey. Each participant goes through 3 marketplace tasks, while staying in the same group (i.e., 3 observations per participant).

No

8,000 respondents

For participant-level proportion outcomes, there will be 8,000 observations. And there will also be 24,000 task-level observations (8,000 respondents * 3 tasks each). Of the 8,000 respondents, at least 1,000 of them will meet fixed financial insecurity criterion (household annual income of less than £15,000) and at least 1,000 UK of them will meet characteristics of poor mental health (MHI-5 score of less than or equal to 68).

2,000 completing respondents in control group
2,000 completing respondents in treatment group 1
2,000 completing respondents in treatment group 2
2,000 completing respondents in treatment group 3