Experimental Design Kernel (EDK Build 20260306.05)
The data you upload is processed entirely within your browser. No information is ever sent to or stored on our servers, ensuring your work remains completely private on your own device.
SCEA ExpiDesK (Statistical Computing for Experimental Applications - Experimental Design Kernel) serves as the foundational engine of the SCEA suite, specifically engineered to automate and standardize the core workflows of agricultural and biological research. By integrating robust frameworks for Completely Randomized Design (CRD) and Randomized Block Design (RBD), the kernel allows researchers to seamlessly transition from initial layout planning to sophisticated statistical analysis within a single interface. ExpiDesK ensures mathematical precision in treatment allocation and variance partitioning, providing a reliable computational backbone that transforms raw experimental data into actionable scientific insights with professional-grade accuracy.
This tool performs a complete Analysis of Variance (ANOVA) for Completely Randomized Designs (CRD) and Randomized Block Designs (RBD), with selectable modes for factorial, pooled, split-plot, and strip-plot workflows. It calculates key statistics, checks for overall significance, and performs post-hoc pairwise comparisons using Fisher's LSD or Tukey's HSD to identify which specific treatment groups are different from each other.
The app uses a single input mode: multi-variable long format CSV.
Required Structure| Treatments | Replication | Growth | Yield |
|---|---|---|---|
| Control | R1 | 10.5 | 50.2 |
| Control | R2 | 11.2 | 53.1 |
| Fertilizer A | R1 | 15.1 | 75.8 |
| Fertilizer A | R2 | 14.9 | 77.4 |
Data transformation is used to meet the assumptions of ANOVA. The app can suggest a transformation automatically, or you can apply one manually:
1/x or Math.log(x+1)).
Transformations are applied only when the raw data violate ANOVA assumptions such as normality or homogeneity of variances. Use transformations in the following cases:
Corrections are used for specific data types, mainly binomial/proportion data derived from counts (e.g., 5 infected out of 10). A correction prevents extreme values like 0% or 100% from biasing the analysis.
p' = (x + 0.5) / (n + 1)
p' = 1 / (4 * N)p' = 1 - (1 / (4 * N))
Summary: Use corrections only for count-based proportions with extreme values—never for continuous measurement-based percentages.
Could not find the required 'q' value in the internal table. Please provide it manually.
Manage standard and custom data transformations. Use 'x' for the data value, 'y' for the back-transformation, 'n' (replication count for the treatment), and 'N' (total observations) for advanced formulas.
| Name | Forward Formula | Backward Formula | Action |
|---|
One or more validation checks failed for your current data and settings.
Transformation Requirements: