Introduction to Statistical Testing

In Geography, we often collect a lot of data, whether it’s measuring the size of pebbles on a beach or looking at the relationship between a country’s wealth and its level of globalisation. But how do we know if the patterns we see are "real" or just down to "random luck"?

That is where statistical tests come in. These tests allow us to prove (with a certain level of confidence) that there is a significant relationship or difference in our data. Think of it like a court of law: we assume there is no pattern until we have enough evidence to prove otherwise.

The Starting Point: The Null Hypothesis \( (H_0) \)

Before running any test, we start with a Null Hypothesis. This is a statement that assumes there is no relationship or no difference between the things we are measuring.

Example: "There is no significant relationship between the distance from the coastline and the size of beach sediment."

The goal of our statistical tests is to see if we can "reject" this null hypothesis. If the test shows a strong pattern, we reject the \(H_0\) and accept the Alternative Hypothesis (\(H_1\)), which says there is a relationship.

1. Spearman’s Rank Correlation Coefficient \( (r_s) \)

What is it for?

Spearman’s Rank is used to test the strength and direction of a relationship (correlation) between two sets of data. It tells us if, as one variable increases, the other one also increases (positive) or decreases (negative).

Geographical Context

You might use this to see if there is a correlation between:

  • The height of a glacial cirque and its orientation.
  • The Human Development Index (HDI) of a country and its level of globalisation.
  • Distance from a central business district (CBD) and land values.

How it works (The Logic)

  1. Rank both sets of data from highest to lowest.
  2. Find the difference (\(d\)) between the ranks for each pair.
  3. Square those differences (\(d^2\)) to get rid of negative numbers.
  4. Use the Spearman’s Rank formula to find the value of \(r_s\).

Understanding the Result

The answer (\(r_s\)) will always be between \(+1.0\) and \(-1.0\).

  • \(+1.0\): A perfect positive correlation (as one goes up, the other goes up).
  • \(0\): No correlation at all (the data is totally random).
  • \(-1.0\): A perfect negative correlation (as one goes up, the other goes down).

Quick Tip: If your calculation gives you a number like \(1.2\) or \(-5.4\), stop! You’ve made a mistake. \(r_s\) can never be greater than \(1\) or less than \(-1\).

2. Student’s t-test

What is it for?

The t-test is used to see if there is a significant difference between the means (averages) of two sets of data. It helps us decide if two groups are truly different or if they just look different because of a few odd samples.

Geographical Context

This is very common in fieldwork. For example:

  • Is the average sediment size on a glaciated outwash plain significantly different from the sediment size at the glacier snout?
  • Is there a significant difference in perception of safety between two different urban neighbourhoods?

How it works (The Logic)

The test looks at the average (mean) of Group A and Group B, but it also looks at the spread of the data (standard deviation). If the data in both groups is very spread out and overlaps a lot, the t-test will likely show the difference is not significant.

Key Takeaway

A high t-value usually means there is a significant difference between your two study areas. A low t-value suggests the two areas are actually quite similar.

3. Chi-squared \( (\chi^2) \)

What is it for?

Chi-squared is used for categorical data. It compares the Observed frequencies (what you actually counted) with the Expected frequencies (what you would expect to see if everything was distributed evenly or randomly).

Geographical Context

You use this when you are counting things in categories rather than measuring them on a scale. For example:

  • The number of different coastal management types (sea walls, groynes, rip-rap) used in two different counties.
  • The pebble shapes (angular, rounded, etc.) found at different points along a river.

The Formula Logic

The test calculates: \( \chi^2 = \sum \frac{(O - E)^2}{E} \)

Where:

  • \(O\) = Observed value (your field data).
  • \(E\) = Expected value (the average/random distribution).

If the difference between what you saw (\(O\)) and what was expected (\(E\)) is very large, your \(\chi^2\) value will be high, suggesting a significant pattern exists.

Interpreting the Results: Significance Levels

Once you have calculated your "test statistic" (\(r_s\), \(t\), or \(\chi^2\)), you aren't finished! You need to check if your result is statistically significant using a Critical Values Table.

Step 1: Degrees of Freedom (\(df\))

Every test has a slightly different way of calculating \(df\), but it is usually based on your sample size (\(n\)). You need this number to find the right row in the statistical table.

Step 2: The \(0.05\) (5%) Significance Level

In Geography, we usually use the \(0.05\) significance level. This means we are 95% certain that the results are not due to chance.

  • If your calculated value is greater than the critical value from the table, you reject the Null Hypothesis. There is a significant relationship/difference!
  • If your calculated value is smaller than the critical value, you accept the Null Hypothesis. Any pattern you saw was likely just down to luck.

Did you know? Sometimes scientists use the \(0.01\) (1%) level. This means they are 99% certain! This is usually reserved for very important things like testing new medicines or high-stakes climate data.

Quick Review Box

Use Spearman’s Rank when: You want to see if two variables are linked (e.g., wealth and health).
Use t-test when: You want to compare the averages of two groups (e.g., pebble size at Site A vs. Site B).
Use Chi-squared when: You are counting frequencies in categories (e.g., the number of people who shop at different types of stores).
Key Rule: Calculated Value > Critical Value = Reject the Null Hypothesis (\(H_0\)).

Common Mistakes to Avoid

  • Mixing up \(n\) and \(df\): Always check the specific rule for the test you are using to find the degrees of freedom.
  • Ranking incorrectly: In Spearman's Rank, if you have two identical values (a tie), you must give them the average of the ranks they would have taken.
  • Ignoring the sign: In Spearman's Rank, the \(+\) or \(-\) tells you the direction of the relationship. Don't ignore it in your conclusion!