Skip to content
An official website of the OECD. Find out more
Log In | Register |
Created by the Public Governance Directorate

This website was created by the OECD Observatory of Public Sector Innovation (OPSI), part of the OECD Public Governance Directorate (GOV).

How to validate authenticity

Validation that this is an official OECD website can be found on the Innovative Government page of the corporate OECD website.

Behavioural Insights Projects

Submit a project

Understanding Canadians’ Attitudes and Behaviours around Weather Forecasts and Warnings

General Information

Project description

In 2015 the World Meteorological Organization (WMO) released recommendations to move towards impact-based weather forecasts and warnings. These impact-based warnings will help increase lead-time for decision makers and the public, to ensure they are making the best choices to prepare for different types of weather events. Canada is currently in the process of adopting these recommendations by moving towards an impact-based tiered warning system for severe weather events. To assist in the communication of this new warning system, this study establishes a baseline understanding of Canadians’ use of weather forecasts and warnings.

We will conduct two online experiments. Both experiments will assess how Canadians react to a new style of extreme weather warnings. The first experiment (n = 500) will utilize a discrete choice methodology to assess how different aspects of the warning message impact self-reported intention to act. We will vary the colour of the warning message (yellow, orange, red), the type of weather warning (snowfall, rainfall, severe thunderstorm, wind), how many call-to-action statements are included (1-4), and the framing of the call-to-action statements (standard, gain, loss).

The second experiment (n = 3,000) will be an online randomized control trial to assess how different instructions as well as aspects of a wind warning influence attitudes and behaviours around the warning. For this, participants will be randomly assigned to 1 of 4 instruction manipulations (full instructions, condensed instructions, tagline instructions, no instructions), as well as 1 of 2 call-to-action statement conditions (2 statements, 4 statements). Then each participant will view a total of 4 weather warnings (old style control, yellow, orange, red) and asked to respond to some questions after each warning. This results in 500 participants viewing each wind warning message.

Analysis Plan

Pre-analysis plan: Is there a pre-analysis plan associated with this registration?

No

Hypothesis

Experiment 1: DCE
1. Intention to act will vary by colour, type of warning, framing of call-to-action statement, and the number of call-to-action statements

Experiment 2: RCT
1. Perceived severity of the wind warning will vary based on the instructions viewed, the number of call-to-action statements and warning colour.
2. Perceived likelihood of the wind warning will vary based on the instructions viewed, the number of call-to-action statements and warning colour.
3. Perceived self-efficacy will vary based on the instructions viewed, the number of call-to-action statements and warning colour.
4. Self-reported behavioural intentions will vary based on the instructions viewed, the number of call-to-action statements and warning colour.

How hypothesis will be tested

Experiment 1: DCE
A hierarchical Bayesian multinomial logistic regression will be used to calculate individual raw utility scores for each participant for each attribute and level. These raw utility scores can then be used to calculate reportable measures that answer the research questions, such as feature importance, preference share, and the optimal package.

Experiment 2: RCT
All hypotheses will be analyzed using a mixed-effects linear model. The dependent variable will be changed based on the hypothesis above. The fixed effects will be the independent variables of the instruction manipulation, the number of call-to-action statements, and the warning color. The participant will be included as a random effect.

It may also be possible to analyze hypotheses 3 and 4 using a factorial ANOVA, in this case the instruction manipulation, and the number of call-to-action statements will be included as between-subjects measures, while the warning colour will be inputted as a within-subjects measure.

Dependent variables

Experiment 1: DCE
- The warning chosen for having higher self-reported intention to act

Experiment 2: RCT
- Self-reported severity of the wind storm
- Self-reported likelihood of the wind storm occurring
- Self-efficacy ratings
- Self-reported behavioural intentions

Analyses

See above

Sample Size. How many observations will be collected or what will determine sample size?

3,500 Canadians

Data Exclusion

N/A

Treatment of Missing Data

We will exclude from analysis participants with missing data.

🇨🇦

Canada

Ottawa

Who is behind the project?

Institution: Impact and Innovation Hub (IIU)- Impact Canada
Team: Impact and Innovation Unit (IIU)

Project status:

Pre-registration

Methods

Methodology: Experiment, Online Experiment, Survey
Could you self-grade the strength of the evidence generated by this study?: 7
Data collection: Have any data been collected for this project already?: Some or all of the data have been collected, and the research team has had access

What is the project about?

Policy area(s): Environment, Climate attitudes and behaviours, Climate Change
Topic(s): Decision-making, Technology Adoption

Date published:

12 May 2025

Comments are closed.