7. Hypothesis Testing
Up to this point, we have focused on understanding the data we have: what variables were measured, how those variables are set up, whether they need to be cleaned, how to describe them, and how to visualize them.
Those steps are always the starting point. Now we are ready to ask a different kind of question:
What can these data tell us beyond this sample?
That is where come in. In this chapter, we will introduce the logic of , or NHST. You do not need to master every statistical test yet. Instead, this chapter gives you the decision-making framework that will repeat throughout the rest of the book.
Hypothesis testing is not about proving that a hypothesis is true. It is about evaluating whether the data are surprising enough, assuming the is true, that we are willing to reject the null hypothesis.
Learning Objectives
By the end of this chapter, you should be able to:
- describe the purpose of null hypothesis significance testing;
- explain the four-step hypothesis-testing process used throughout this book;
- identify the independent variable, dependent variable, and design in a study;
- write plain-language null and alternative hypotheses;
- distinguish between directional and non-directional hypotheses;
- explain the basic role of assumptions, alpha, p-values, effect sizes, and power;
- interpret hypothesis-testing results using reject or fail-to-reject language; and
- describe what NHST can and cannot tell us.
The Four-Step Process
Each inferential statistic we learn in this book follows the same general process:
- Look at the data.
- Check assumptions.
- Perform the test.
- Interpret the results.
The details change depending on the test. A chi-square test, t-test, ANOVA, correlation, and regression all have different procedures and different output. But the overall process is the same.
I created a four-page handout that summarizes this process across the inferential statistics covered in this book: Statistics with jamovi - Overarching handout.
I recommend saving it and returning to it as you move through the inferential statistics chapters. The handout is the quick reference. This chapter explains the logic behind it.
Our Running Example
To make the logic concrete, we will use a simplified example inspired by Bandura’s classic Bobo doll studies.
Imagine a researcher randomly assigns children to one of two conditions:
- In one condition, children watch a video of an adult behaving aggressively toward a Bobo doll.
- In the other condition, children watch a video of an adult playing calmly with a Bobo doll.
Afterward, the children are observed while playing in a room with a Bobo doll. The researcher records the number of aggressive behaviors each child shows.
This is a useful example because it has the pieces we need for hypothesis testing:
- an independent variable: the type of video the child watched;
- a dependent variable: the number of aggressive behaviors observed;
- two independent groups; and
- a research question about whether the groups differ.
You do not need to understand the specific statistical test yet. Later, we will learn that this kind of design can be analyzed with an independent-samples t-test. For now, we are using the example to understand the logic of hypothesis testing.
Looking Ahead
The next sections walk through the four steps one at a time. As you read, focus less on memorizing vocabulary and more on understanding the process. Every inferential test you learn later will use this same basic structure.