The scatter plot below shows annual rainfall (in mm) on the x axis and the number of plant species recorded at a site on the y axis. Each point is one location, and the points trace out a clear increasing pattern that looks close to a straight line as rainfall rises.
Based on the pattern in the graph, which one of the following statistical methods would be most appropriate to model the relationship between annual rainfall and the number of plant species?
Show Hint
Both rainfall and species number are continuous variables with a straight-line trend, so use linear regression, not a categorical-group test.
Step 1: Ask what each candidate test actually needs.
Before looking at the picture, it helps to remember what each of the four listed methods requires. The Kruskal-Wallis test and Student's t-test both need a categorical grouping variable (two groups for the t-test, three or more for Kruskal-Wallis) plus a measured outcome for each group. The chi-squared test needs two categorical variables summarized as counts in a table. Linear regression needs two continuous, numeric variables where one is thought to depend on or track the other.
Step 2: Check what variables we actually have.
Here both variables, annual rainfall and number of plant species, are continuous measurements, not category labels. There is no grouping factor anywhere in the description, and nothing is being counted into categories.
Step 3: Eliminate the mismatched tests.
Since there is no grouping variable, the Kruskal-Wallis test and the t-test cannot apply. Since neither variable is categorical, the chi-squared test cannot apply either. Only linear regression is built to handle two continuous variables at once.
Step 4: Confirm using the scatter plot's shape.
The points climb steadily from the bottom left to the top right of the plot without bending back on themselves or breaking into separate clusters, the classic sign of a straight-line, linear trend, exactly the pattern linear regression is meant to summarize.
