Experimental Design and Data Analysis

Pairwise Comparisons: Witchweed and Fertilizer Interaction

• Using anova models to produce interaction plots to guide recommendations for fertilizer or weed control.
• Tukey's tests can be performed to further analyze combinations.
• P-values are adjusted to account for family-wise error rates due to multiple comparisons.
• Factorial designs are powerful as they address multiple research questions, reduce type one errors, and provide a better understanding of mechanisms.

Understanding Main Effects and Interactions

• Grasp the concept of main effects and interactions.
• Learn to interpret ANOVA tables by starting from the highest order interaction term; if it's not significant, move up to the next higher orders or main effects.
• Avoid just listing significant findings from an ANOVA table when a significant interaction is present, focus on what's truly important.

Checking Model Assumptions with Residuals

• Always check model assumptions for ANOVA.
• Use residuals from models with graphical approaches like residuals versus fitted plots and QQ plots.
• Residuals vs Fitted plots: testing equal variance assumption.
• QQ plots: testing normality assumption.
• Residuals are essential for testing assumptions; using raw data will result in deductions.

Interpreting Confidence Intervals in E-Means Plots

• Confidence intervals in e-means plots help visualize differences.
• Overlapping confidence intervals indicate non-significance.

Addressing Assumption Violations in ANOVA

• Real-world data often violates normality assumptions.
• ANOVA is robust against departures from normality, especially in balanced designs (equal replication across treatment levels).
• Equal variance assumption is more critical than normality assumption.
• If transformations (like log transformation for count data or arcsine for proportion data) fail to improve assumptions, acknowledge the limitation and interpret results cautiously.

Transformations in Data Analysis

• Try a log transformation for count data.
• Use a square root transformation for a harsher effect.
• Use arcsine transformation for proportion data.
• Avoid excessive transformations to maintain data interpretability.
• If assumptions fail, acknowledge it and proceed with caution.
• Cite papers supporting the robustness of ANOVA against deviations from normality.

Kruskal-Wallis Test

• Kruskal-Wallis test, is an alternative for simple one-way ANOVA designs.

Second Project Introduction

• The second project involves using real-world data from a published paper to conduct a full scientific paper analysis.
• Download the pre-processed data from Canvas, not directly from the paper.
• Choose one of the provided data sheets or create your own data set.
• Formulate a scientific question and conduct a complete analysis.

Scientific Questions and Experimental Design

• Develop a question, determine the experimental design, treatment design, and model equation.
• Use ANOVA and multiple regression techniques to answer the questions.
• Design questions that can be answered using these techniques.
• Focus on residuals, not raw data.

Scientific Writing Guide

• Follow a set formula for writing scientific papers.
• Refer to first-year notes on scientific writing.

Data Set Metadata

• Use the metadata sheet in the Excel file to understand the data set variables and measurements.
• Metadata helps ensure proper data usage and interpretation when using datasets collected by others.

Template and Formatting

• Use the provided Quattro or R Markdown template for the report.
• Ensure "embed resources" is set to true in Quattro to render plots.

Utilizing AI Tools

• Use AI tools cautiously for coding and plot improvements.
• Avoid using AI for interpretation or analysis.
• Acknowledge AI usage in the acknowledgement section, including prompts and verification methods.

Referencing and Literature

• Format references consistently, following a specified style.
• Cite primary literature (journal articles, textbooks).
• Avoid web pages as primary sources unless complementing primary sources.

Appendix

• Use the appendix for supplementary information, such as assumption testing plots.
• Label figures in the appendix as "Figure A1", "Figure A2", etc.

Model Assumptions Testing Location

• Test model assumptions in the methods section.
• Report whether assumptions were met or violated, and reference appendix plots.

Post-Hoc Tests

• Post-hoc tests can be conducted after the ANOVA.