Data from: Duration and economic value of a walking-in-nature therapy program: Implications for conservation
Data files
Oct 12, 2025 version files 156.58 KB
-
CHAUVE_1.RMD
26.76 KB
-
Chauvenet_df_final_15Sept2025.csv
125.21 KB
-
README.md
4.61 KB
Abstract
Nature exposure, such as visiting protected areas, provides mental health benefits that reduce healthcare costs and improve productivity, with global values in the trillions. Countries are bringing nature-based programs into mainstream mental healthcare via nature therapies. This study quantifies the scale, duration, and economic value of mental health benefits from a long-established nature therapy program and implications for conservation. Using a Before-After-Control-Impact design, we evaluated a 12-week nature walking program with social mechanisms for therapeutic adherence. Mental health was assessed using the Personal Wellbeing Index, a measure of subjective well-being, with participants and controls from the same subpopulation. Measurements occurred at program start, end, and 12 weeks post-intervention. Economic benefits were calculated using the financial value of quality-adjusted life-years. The nature-based therapy program improved the mental well-being of participants during the program and for at least three months afterwards. While controls showed well-being improvements when they reported having physically exercised (despite not being instructed to), program participants exhibited an additional Personal Wellbeing Index increase of 5.1%. Training in nature was a critical component, leading to the highest increase in mental health benefits and doubling of their duration (up to 12 months). The mean total economic benefit per participant who followed the program design in full was c.AU$4,000. Total economic contribution via mental health, adjusted for socioeconomic and demographic factors, participation patterns, post-program fade-out, and the national number of participants each year, is therefore c.AU$20 million per annum. 5. Mental health benefits of nature visits fade once people stop visiting parks. To maximise their contribution to political and economic support for protected areas, therefore, the focus for future research and practice should be on social mechanisms to promote lifelong park visit habits.
Dataset DOI: 10.5061/dryad.hqbzkh1wb
Description of the data and file structure
We designed a Before-After-Control-Impact experiment.
Both treatment and control participants were recruited within members of a group who self-identify as “wild women” on dedicated social media accounts. These women are affiliated with Coastrek® via its parent company Wild Women on Top (WWoT®), a women’s outdoor hiking and mountaineering organisation. We identified 4000 current and active Wild Women on Top members contactable by email and invited them to take part in this research. Members enrolled to take part in the Melbourne 2022 event formed our treatment group, whereas those who had taken part in previous programs or planned to join future programs but were not signed up to the Melbourne 2022 event, formed our control group.
Our Before-After-Control-Impact design relies on the same individuals completing multiple surveys, and both treatment and control groups were surveyed with the same questionnaire three times at fixed time points. To minimise drop-out, we conducted surveys at only three time-points. The first, T1, was immediately prior to the beginning of the 12-week program, Week 0. Participants can sign up at any time before this, but that is the date at which they are committed and start active involvement. The second, T2, was at the end of the 12-week program, Week 12. That is immediately after the participants had completed the final event of the program (30-60 km team hike along a national park). They were asked to complete this survey within a week of the event. The 12-week interval T1 - T2, the period of the therapy program itself, thus spans the greatest anticipated improvement in wellbeing. The third, T3, Week 24, was 12 weeks after the end of the program, chosen as a fade-out period equal in length to the program itself.
Files and variables
File: Chauvenet_df_final_15Sept2025.csv
Description:
Variables
- ID: Unique identifier for participants.
- PWI: Personal Wellbeing Index score in each surveys at Time 1, 2 and 3.
- time: Continuous time variable representing surveys at Time 1, 2 and 3.
- Relationship.ord: Categorical variable representing marital status of participants.
- children.ord: Categorical variable representing parental status of participants.
- Age.ord: Categorical variable representing age of participants.
- trainteam.ord: Categorical variable representing training team status of participants.
- trainloc.ord: Categorical variable representing training location status of participants.
- exercise.ord: Categorical variable representing training intensity status of participants.
- subcat2.ord: Categorical variable representing treatment (yes) or control (no) status of participants.
- PWI_diff: Proportional change in PWI between T1 and T2, and T2 and T3.
- PWI_lag: Previous PWI.
- Survey.ord: Categorical variable representing surveys at Time 1, 2 and 3.
File: CHAUVE_1.RMD
Description: R code to load and prepare the survey data, and run the two Generalised Additive Models presented in the paper. Code to make predictions with both models, and to perform the economic valuation is also included.
Code/software
Software used: R version 4.4.2 (2024-10-31)
Platform: aarch64-apple-darwin20
Running under: macOS Sequoia 15.6.1
Loaded packages:
- attached base packages: stats, graphics, GrDevices, utils, datasets, methods, base
- other attached packages: itsadug_2.4.1; plotfunctions_1.4; glue_1.8.0; ggtext_0.1.2; AICcmodavg_2.3-3; infer_1.0.7; mgcv_1.9-1; nlme_3.1-167 paletteer_1.6.0; ggnewscale_0.5.0; FactoMineR_2.11; factoextra_1.0.7; effects_4.2-2; carData_3.0-5; ggpubr_0.6.0; rstatix_0.7.2; lubridate_1.9.4; forcats_1.0.0; stringr_1.5.1; purrr_1.0.4; readr_2.1.5 tibble_3.2.1 ggplot2_3.5.1; tidyverse_2.0.0; data.table_1.17.0; gridExtra_2.3; viridisLite_0.4.2; reshape_0.8.9; MuMIn_1.48.4; tidyr_1.3.1; corrplot_0.95; RColorBrewer_1.1-3; ncf_1.3-2; MASS_7.3-64; colorRamps_2.3.4; dplyr_1.1.4; lme4_1.1-36; Matrix_1.7-2; raster_3.6-32; sp_2.2-0
Access information
Other publicly accessible locations of the data:
- Not applicable
Data was derived from the following sources:
- Not applicable
Human subjects data
We received explicit consent from participants to publish the de-identified data in the public domain. Unique identifiers were used to anonymise the responses.
