Each group is assigned ONE dataset. Your goal is to investigate the research question using visualization, reasoning, and appropriate statistical methods.
Tasks: 1. Visualize the data before any formal analysis 2. Identify dependent and independent variables 3. State null and alternative hypotheses IN WORDS 4. Choose and justify an appropriate statistical approach 5. Evaluate assumptions and limitations 6. Interpret results in environmental context
The goal is reasoning — not naming a test.
Context: Urban heat islands are driven by impervious surfaces. Vegetation may mitigate heat through shading and evapotranspiration.
Metadata: - Vegetation_Percent: Percent land area covered by vegetation - Surface_Temperature_C: Surface temperature (°C)
Research Goal: Determine whether vegetation cover is associated with surface temperature.
Variables: - Independent: Vegetation_Percent - Dependent: Surface_Temperature_C
Assumptions to Consider: - Are both variables continuous? - Does the relationship appear linear? - Are residuals evenly scattered? - Are outliers influential?
Common Pitfalls: - Treating correlation as causation - Ignoring non‑linear patterns - Overinterpreting small effect sizes
Context: Human activity often alters nutrient levels downstream.
Metadata: - Location: Sampling category (Upstream, Downstream) - Nitrate_mgL: Nitrate concentration (mg/L)
Research Goal: Evaluate whether nitrate concentrations differ by sampling location.
Variables: - Independent: Location (categorical) - Dependent: Nitrate concentration
Assumptions to Consider: - Are observations independent? - Is the response approximately normally distributed per group? - Are group variances similar?
Common Pitfalls: - Treating groups as ordered when they are categorical - Ignoring unequal variance - Failing to visualize group overlap
Context: Nutrient enrichment increases algal productivity but may plateau or interact with other limiting factors.
Metadata: - Fertilizer_Level: Treatment category (Low, Medium, High) - Algal_Cover_Percent: Percent surface coverage by algae
Research Goal: Assess whether algal cover differs across fertilizer treatments.
Variables: - Independent: Fertilizer_Level - Dependent: Algal_Cover_Percent
Assumptions to Consider: - Independence of observations - Normality within groups - Similar variance among treatments
Common Pitfalls: - Running multiple pairwise tests without a global question - Ignoring variance differences - Concluding “which group is best” without post‑hoc reasoning
Context: Rainfall contributes to streamflow, but responses vary due to watershed characteristics and storm intensity.
Metadata: - Rainfall_mm: Rainfall amount (mm) - Streamflow_m3s: Stream discharge (m³/s)
Research Goal: Examine whether rainfall and streamflow are associated.
Variables: - Independent: Rainfall_mm - Dependent: Streamflow_m3s
Assumptions to Consider: - Linearity of relationship - Absence of extreme leverage points - Independence of observations
Common Pitfalls: - Assuming causation from association - Ignoring nonlinear or threshold behavior - Overinterpreting correlation strength
Context: Habitat type affects species richness via disturbance and resource availability.
Metadata: - Habitat: Habitat category (Forest, Grassland, Urban) - Species_Richness: Number of observed species
Research Goal: Determine whether species richness differs among habitat types.
Variables: - Independent: Habitat - Dependent: Species_Richness (count data)
Assumptions to Consider: - Is the response continuous or discrete? - Are distributions symmetric? - Are variances similar?
Common Pitfalls: - Applying methods that assume normality to count data - Ignoring skewness or zero inflation - Overstating ecological mechanisms without evidence
Context: Air quality regulations aim to reduce particulate pollution.
Metadata: - Site: Monitoring site ID - PM25_Before: PM2.5 concentration before policy - PM25_After: PM2.5 concentration after policy
Research Goal: Evaluate whether PM2.5 concentrations changed following the policy.
Variables: - Independent: Time (before vs after, paired by site) - Dependent: PM2.5 concentration
Assumptions to Consider: - Are observations paired? - Are differences normally distributed? - Are changes consistent across sites?
Common Pitfalls: - Treating paired data as independent - Ignoring direction and magnitude of change - Focusing solely on p‑values instead of effect sizes
FINAL NOTE: There is more than one defensible analytical path. Critical thinking = justification, assumptions, and interpretation — not on naming a specific test.