Before any test, model or p-value, there is a simpler and more important step: looking at your data as if you were seeing it for the first time. That is what descriptive and exploratory statistics is for.
Why start here?
Descriptive and exploratory statistics is the first step of any research project that works with numbers. It lets you summarize, organize and present the data so you can interpret it and spot patterns before moving on to more complex analyses.
Skipping it is risky. Mistakes show up later, or valuable information goes unnoticed.
Instead of a long list of separate values, you get measures and charts that give you the general picture:
- how each variable behaves;
- how the values are distributed;
- which values are the most common;
- how much the values differ from one another.
Tables for discrete variables
A table is one of the simplest ways to synthesize data. For discrete variables, those with countable values, frequency tables are essential:
- the absolute frequency is how many times each value occurs;
- the relative frequency is the proportion each value represents in the whole dataset.
A complete table can also include cumulative frequencies, which are running totals, and relative and cumulative percentages. Together they describe the empirical distribution of the variable.
Choosing the right chart
The right chart depends on the measurement scale of the variable:
- for discrete variables, a bar chart shows how frequent each category is;
- for continuous variables, a histogram or a box plot reveals the shape of the distribution and possible outliers.
A good choice helps you see patterns, catch possible data errors and build an intuitive feel for the distribution before any advanced analysis.
In short
Informative tables and well-chosen charts are an essential stage of any study. They prepare the ground for the later analyses that answer your research question rigorously.