Brief Mindfulness, Detached Reappraisal, and Additional Investment in a Supply Chain Project Task: A Randomized Experiment

# Brief Mindfulness and Supply-Chain Investment Decisions Replication package for: > Wang, S., & Tang, X. *Brief Mindfulness, Detached Reappraisal, and Additional> Investment in a Supply Chain Project Task: A Randomized Experiment.* A pre-registered randomized experiment in which 356 managers each made 12cost-control decisions in six supply-chain project scenarios. This repositorycontains the de-identified data, the experimental materials, and the Ranalysis pipeline that reproduces every published table and figure. --- ## Design - **Intervention** (between-participants): 12-minute brief mindfulness audio vs. duration-matched active control (narrated geological description).- **Framing** (between-participants): controllable vs. uncontrollable loss framing of a project overrun.- **Sunk cost** (within-participant): high vs. low versions of each of six scenarios → 12 decisions per participant. **Samples.** Behavioral analytic *n* = 334 (A1 = 83, A2 = 85, B1 = 83, B2 = 83);physiological analytic *n* = 320. **Primary outcome.** Normalized additional investment, `D = a / Cmax` (Equation 3of the paper). **Primary model.** Linear mixed-effects model with factorial fixed effects, acentered presentation-position covariate, random intercepts and sunk-costslopes for participants, and random intercepts for scenarios (Equation 6).Estimated by REML with Satterthwaite degrees of freedom. --- ## Repository layout ```├── README.md <- this file├── LICENSE <- MIT License├── requirements.R <- install helper├── run_all.R <- master reproduction script├── data/ <- 11 CSV tables + data dictionary│ ├── data_dictionary.md│ ├── participants.csv│ ├── decisions.csv│ ├── regulation_probes_long.csv│ ├── physiological_baseline.csv│ ├── physiological_decisions.csv│ ├── questionnaires_FFMQ.csv│ ├── questionnaires_PANAS.csv│ ├── questionnaires_state_mindfulness.csv│ ├── credibility_expectancy.csv│ ├── active_control_recall.csv│ └── scenario_parameters.csv├── materials/│ └── Experimental_Materials_English.pdf├── models/│ ├── fit_models.R <- fits the mixed-effects models│ ├── fitted_estimates.rds <- estimates used by reporting layer│ ├── primary_fit.rds <- Equation (6) lmer object│ ├── logistic_fit.rds <- logistic part of two-part model│ ├── beta_fit.rds <- beta part of two-part model│ └── ev_gap_fit.rds <- expected-value-gap model├── scripts/│ ├── 00_helpers.R <- shared paths, loaders, formatting│ ├── 01_sample_characteristics.R <- Table 3│ ├── 02_cell_summaries.R <- Table 4│ ├── 03_primary_model.R <- Table 5, simple slopes, Table S1 (c)│ ├── 04_two_part_and_ev_gap.R <- Section 4.3│ ├── 05_regulation_measures.R <- Table 6│ ├── 06_moderated_mediation.R <- Table 7│ ├── 07_physio_descriptives.R <- Tables S4, S5│ ├── 08_equivalence_bayes.R <- Table S7│ ├── 09_supplementary_S1.R <- Tables S1 (a, b), S2│ └── 10_figures.R <- Figures 5, 6, 7, 8└── results/ ├── tables/ <- CSV copies of every reported table └── figures/ <- PNG + PDF for every reported figure``` --- ## Quick start ### 1. Install R and packages **Debian / Ubuntu**: ```bashsudo apt-get install -y r-base r-cran-lme4 r-cran-lmertest r-cran-emmeans \ r-cran-boot r-cran-ggplot2 r-cran-dplyr r-cran-tidyr \ r-cran-glmmtmb``` **CRAN**: ```rinstall.packages(c("lme4","lmerTest","glmmTMB","emmeans","boot", "ggplot2","dplyr","tidyr"))``` Or just: ```bashRscript requirements.R``` ### 2. Reproduce every result ```bashgit clone <this-repository>cd <this-repository>Rscript run_all.R``` Expected runtime: around 20–30 seconds on a modern laptop. ### 3. Run a single step Each script is self-contained once `models/fitted_estimates.rds` exists: ```bashRscript scripts/03_primary_model.R``` --- ## Software Analyses were conducted in **R 4.4.1** with: - `lme4` 1.1-35.5- `lmerTest` 3.1-3- `glmmTMB` 1.1.9- `emmeans` 1.10.4- `boot` 1.3-30- `ggplot2`, `dplyr`, `tidyr` for data wrangling and plotting Tests are two-sided; the nominal threshold is .05; confidence intervals are95% unless stated otherwise. --- ## Expected outputs Tables (in `results/tables/`, CSV): | File | Reproduces ||---|---|| `Table3_sample_characteristics.csv` | Paper Table 3 || `Table4_cell_summaries.csv` | Paper Table 4 || `Table5_primary_model.csv` | Paper Table 5 || `Table6_regulation_measures.csv` | Paper Table 6 || `Table7_moderated_mediation.csv` | Paper Table 7 || `Table_TwoPart_logistic.csv`, `Table_TwoPart_beta.csv` | Section 4.3 two-part model || `Table_EV_gap.csv` | Expected-value-gap interaction || `TableS1a_exclusions.csv` | Supplementary Table S1 (a) || `TableS1b_excluded_vs_retained.csv` | Supplementary Table S1 (b) || `TableS1c_sensitivity.csv` | Supplementary Table S1 (c) || `TableS2_scenario_parameters.csv` | Supplementary Table S2 || `TableS4_probe_descriptives.csv` | Supplementary Table S4 || `TableS5_physio_descriptives.csv` | Supplementary Table S5 || `TableS7_equivalence_bayes.csv` | Supplementary Table S7 | Figures (in `results/figures/`, PNG and PDF): | File | Reproduces ||---|---|| `Figure5_manipulation_checks.*` | Paper Figure 5 || `Figure6_contrasts.*` | Paper Figure 6 || `Figure7A_cell_means.*` | Paper Figure 7, Panel A || `Figure7B_distribution.*` | Paper Figure 7, Panel B || `Figure8A_regulation_std.*` | Paper Figure 8, standardized panel || `Figure8B_regulation_resp.*` | Paper Figure 8, respiration panel || `SuppFig_cell_bars.*` | Supplementary bar chart | --- ## Equations - **Equation (1)** — Expected net value of continuing: `Vc = p · R − Ca`.- **Equation (2)** — Expected-value rule: continue iff `p · R − Ca > S`.- **Equation (3)** — Normalized allocation: `D = a / Cmax`.- **Equation (4)** — Expected loss: `L* = (max_a Vc(a) − Vc(a*)) / span`.- **Equation (5)** — DR composite: equally weighted mean of three standardized components `(z(DR_self) + z(−SCL_change) + z(ln_RMSSD_change)) / 3`, then standardized again against pre-decision observations in the physiological sample.- **Equation (6)** — Primary linear mixed-effects model: ``` D_ijk = β0 + βM·M + βF·F + βSC·SC + βMF·M·F + βMSC·M·SC + βFSC·F·SC + βMFSC·M·F·SC + βpos·pos_c + u_part + u_sc_slope + u_scenario + ε ``` Factors are coded ±0.5; `pos_c` is the centered presentation position. --- ## Analysis seed The random seed used in sampling steps is `20250425`. It is set at the top ofevery script that resamples (bootstrap intervals, model fitting). Fitted-modelobjects are serialized to `models/*.rds` for later inspection. --- ## Citation Please cite the paper when using these data or code. ## License MIT License. See `LICENSE`. ## Ethics The study was approved by the Institutional Review Board of the School ofEconomics and Management, Beijing Jiaotong University. All participants wereadults in managerial positions and provided written informed consent.