Final Paper Instructions

This assignment guides you through creating a research project repository and developing your final paper proposal.

1 Step 1: Create a Private GitHub Repository

Go to GitHub and log in
Click the “+” icon in the top-right corner and select “New repository”
Name your repository (e.g., SNR690-final-paper-yourname)
Add a brief description of your project
Important: Select “Private” to keep your work confidential
Check “Add a README file”
Click “Create repository”

2 Step 2: Clone Repository to RStudio

In your new GitHub repository, click the green “Code” button
Copy the HTTPS URL
Open RStudio
Go to File > New Project > Version Control > Git
Paste the repository URL
Choose a local directory and click “Create Project”

3 Step 3: Modify the README

Edit your README.md file to include the following sections:

3.1 Objective

What do you aim to accomplish with this research?

3.2 Research Question

What specific question are you trying to answer?

3.3 Variables

List your response variable(s) and predictor variable(s)

3.4 Experimental Design or Data Structure

Describe your data collection approach or existing dataset structure

3.5 Analysis Method

What statistical method(s) will you use?

3.6 Main Result

What do you expect to find or demonstrate?

3.7 Logical Alignment

Do these components align logically? Explain briefly how your methods will answer your research question.

3.8 Expected Timeline

Create a week-by-week plan from Week 6 through Week 14 outlining your anticipated progress on the final paper. For each week, write a brief description of what you plan to accomplish (e.g., data cleaning, running analyses, drafting the methods section). This timeline should be realistic and account for your other coursework and obligations. You will use this as a self-assessment tool throughout the semester: revisiting and updating it as you progress.

3.9 Usefulness for Your Career or Degree

Briefly describe how this project connects to your thesis or your broader academic or professional goals. How does working through this research question and analysis contribute to your development as a researcher or practitioner in your field? Does it help you with your thesis/dissertation? This section should demonstrate that the project is not just an academic exercise, but an opportunity to apply appropriate quantitative approaches to a problem that is meaningful and relevant to your own research trajectory

5 Step 5: Write a method justification section

On the same document, write a brief justification for your chosen method(s). This should start with a sentence like this: - I chose X methods because … - A key limitation of my analysis is … so we …

It should be two or so paragraphs.

Your paragraphs can address:

Writing Checklist

⬜ What is your response variable and what distribution does it follow?
⬜ What is your data structure (independent, nested, repeated, blocked)?
⬜ Why is a simpler analysis insufficient?
⬜ Why is a more complex analysis not warranted?
⬜ What assumptions does your method make? List ALL OF THEM!

And more, depending on your project. The goal is to justify your method choice and acknowledge its limitations. And to be 100% sure about it befor you go into data cleaning and analysis.

6 Step 6: Check your assumptions

Based on the assumptions you wrote, check your assumptions. You can do this by hand, on a piece of paper or on your project. list every assumption your chosen model requires. Then rate each one:

Assumption	Status
(e.g., independence of residuals)	🟢 Checked: and it holds
(e.g., normality of random effects)	🟡 Haven’t checked yet: but I could by next class
(e.g., no spatial autocorrelation)	🔴 Can’t fully check: need to acknowledge it and discuss it

The goal is not to have all green. Red assumptions are ok! they are honest limitations (all studies have them) that belong in your discussion section.

7 Example Project Template

Objective: Assess the impact of habitat restoration on native bird species richness in riparian zones

Research Question: Does riparian habitat restoration increase native bird species richness compared to unrestored sites?

Variables: - Response: Bird species richness (count) - Predictors: Restoration status (restored/unrestored), time since restoration (years), habitat area (hectares)

Experimental Design or Data Structure: Observational study with 20 restored and 20 unrestored riparian sites across Arizona, sampled over 3 breeding seasons

Analysis Method: Generalized Linear Mixed Model (Poisson regression) with site as random effect to account for repeated measures

Main Result: Restored sites are expected to show significantly higher species richness, with effect size increasing with time since restoration

Logical Alignment: Yes - the GLMM approach handles count data (species richness) and accounts for site-level variation while testing the restoration effect. The temporal component allows assessment of restoration trajectory, directly addressing whether and how restoration increases bird diversity.

Expected Timeline: - Week 6: Finalize research question analysis plan - Week 7: Clean data - Week 8: Run preliminary analyses - Week 9: Interpret results and create figures - Week 10: Draft methods section - Week 11: Draft results section - Week 12: Draft introduction and discussion - Week 13: Revise and finalize paper - Week 14: Submit final paper

Usefulness for Your Career or Degree: This project directly informs my thesis research on habitat restoration and biodiversity. It allows me to apply advanced statistical methods to real-world ecological data, enhancing my quantitative skills and contributing to a publishable paper that will strengthen my CV and support my future academic career.

Method Justification: I chose a Generalized Linear Mixed Model with a Poisson distribution because my response variable, bird species richness, is a count variable that typically follows a Poisson distribution. A simpler analysis, such as a t-test comparing restored and unrestored sites, would not account for the repeated measures across multiple seasons or the potential non-independence of observations within sites. On the other hand, a more complex model, like a zero-inflated model, is not warranted given that preliminary data exploration did not indicate an excess of zeros in species richness.