Generate Comprehensive Scientific and Agronomic Report for Experimental Trials
Source:R/experiment_report.R
experiment_report.RdSynthesizes experimental topology, statistical modeling decisions, assumption audits, ANOVA hypothesis tests, biological rankings with effect sizes, and design relative efficiency into a complete, publication-ready Markdown or text narrative.
Arguments
- fit
An object of class `"agri_fitted_model"` returned by
fit_experiment.- post
Optional object of class `"agri_posthoc"`. If `NULL`, computed automatically.
- efficiency
Optional object of class `"agri_design_efficiency"`. If `NULL`, computed automatically.
- output_file
Optional character path to save the Markdown report (e.g., `"experiment_report.md"`).
Examples
data(wheat_splitplot, package = "agriDesignR")
fit <- fit_experiment(
data = wheat_splitplot,
response = "grain_yield",
main_plot = "temperature",
sub_plot = "genotype",
block = "block"
)
rep <- experiment_report(fit)
#> Note: adjust = "tukey" was changed to "sidak"
#> because "tukey" is only appropriate for one set of pairwise comparisons
print(rep)
#> # Scientific & Statistical Report: grain_yield
#> **Generated by**: `agriDesignR` | **Date**: 2026-08-26
#>
#> ## 1. Experimental Topology & Modeling Strategy
#> The experiment was conducted under a **Split-Plot Design (Parcelas Divididas)** evaluating response variable `grain_yield` across treatment factor(s) `temperature, genotype`. A **LMM (Linear Mixed-Effects Model)** was fitted using the `lme4` computational engine (gaussian distribution family). Random effects were structured as `(1 | block) + (1 | block:temperature)`, properly partitioning variance components according to the physical field/greenhouse hierarchy.
#>
#> ## 2. Model Diagnostics & Assumption Verification
#> Residual assumptions were systematically audited. Normality of residuals was evaluated via Shapiro-Wilk test (W = 0.98, p = 0.6964), indicating that the normality assumption was **PASSED**. Homogeneity of variance across treatment groups showed a maximum-to-minimum variance ratio of 21.36x (Status: **ACCEPTABLE**). Minor to moderate deviations were detected; if necessary, 'remedy(fit)' can be utilized to evaluate Box-Cox or variance weighting structures.
#>
#> ## 3. Analysis of Variance & Significance
#> ```text
#> Type III Analysis of Variance Table with Kenward-Roger's method
#> Sum Sq Mean Sq NumDF DenDF F value Pr(>F)
#> temperature 359.95 359.95 1 3 75.346 0.003217 **
#> genotype 1584.58 396.15 4 24 82.923 1.086e-13 ***
#> temperature:genotype 238.75 59.69 4 24 12.494 1.238e-05 ***
#> ---
#> Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
#> ```
#>
#> ## 4. Agronomic Performance Ranking & Effect Sizes
#> Comparison family: treatment = temperature x genotype, conditioned by none, adjustment = tukey, alpha = 0.05, estimand = estimated marginal means. Multiple pairwise comparisons with compact letter displays identified distinct statistical tiers. The top-performing treatment level was **`1 : 4`** with an estimated mean of 65.71 (+13.76% relative to the baseline `Control_22C : G1`, Cohen's d = 1.32).
#>
#> ```text
#> Rank temperature genotype Mean_Estimate SE Delta_vs_Ctrl Percent_Gain
#> 1 Control_22C G4 65.710 2.273 7.947 13.76
#> 2 Control_22C G2 62.648 2.273 4.885 8.46
#> 3 Control_22C G5 58.880 2.273 1.117 1.93
#> 4 Control_22C G1 57.763 2.273 0.000 0.00
#> 5 Heat_36C G4 56.985 2.273 -0.778 -1.35
#> 6 Control_22C G3 53.998 2.273 -3.765 -6.52
#> 7 Heat_36C G2 44.330 2.273 -13.433 -23.25
#> 8 Heat_36C G5 42.733 2.273 -15.030 -26.02
#> 9 Heat_36C G1 39.630 2.273 -18.133 -31.39
#> 10 Heat_36C G3 30.130 2.273 -27.633 -47.84
#> Cohen_d Group
#> 1.32 d
#> 0.81 ad
#> 0.19 ab
#> 0.00 ab
#> -0.13 abd
#> -0.63 be
#> -2.23 ce
#> -2.50 c
#> -3.01 c
#> -4.59 f
#> ```
#>
#> ## 5. Experimental Design Relative Efficiency (RE)
#> Relative efficiency analysis demonstrated an RE of **208.23%** compared to a completely randomized design (CRD). [OK] Highly Effective Blocking: The blocking factor 'block' increased experimental precision by 108.23% compared to an unblocked DCA (CRD). A CRD would have required 2.08x more replications per treatment to detect the same effect size.
#>