Descriptive statistics: Statistics that help you describe or summarise each
variable in your dataset - i.e., that describe your sample.
Categorical Numerical
What is it?
Data that reflect categories of things. For
example, male/female, left/right, etc. You
can assign a number to reflect each
category.
Data that are naturally numbersbased.
For example, age, height,
salary, etc.
Common
descriptive
statistics used
Count/Frequency
Proportion/Percentage
Mean (average)
Median
Standard deviation
Common graphs Bar chart, pie chart Box plot, stem and leaf, scatter plot
KEY TERMINOLOGY
STATS 101 “Cheat Sheet”
A quick reference guide for key terminology & concepts
Dataset: Your collection of data (e.g., survey responses, lab test results, etc.)
Population: The entire group of people (or animals or whatever unit) that you’re
interested in researching. E.g. the population of a city or country.
Sample: The subset of the population that you can collect data from/about.
Variable: Something that changes in value and can be directly measured.
Hypothesis: An assumption that can be supported or rejected by your data.
Inferential statistics: Statistics that help you test hypotheses, assumptions or
predictions about an entire population using your dataset.
KEY CONCEPT
How closely your sample reflects
your population of interest is
referred to “representativeness”.
Generally speaking, you’ll want
your sample to be as representative
as possible, as this allows you to
draw conclusions about the
population more confidently.
TYPES OF DATA
Data can come in all shapes and sizes, and it’s important to understand what type of data you’re working with, as this will
influence what types of statistical tests (descriptive and inferential) you can apply to your dataset. At the broadest level, the
two types of data are categorical (category-based) and numerical (numbers-based).
CONNECTED CONCEPT: NORMAL DISTRIBUTION
A normal distribution, or bell curve, is a graph showing how variables like
heights or scores are spread out. When a distribution is “normal”, most values
are near the average, creating a high peak in the middle, and fewer values are far
from the average, making the graph taper off like bell ends on both sides.
You’ll likely encounter many bell curves within your research, but it’s useful to
note that data can take many other shapes as well.
WHY DESCRIPTIVES MATTER
Descriptive Statistics 101
Descriptive statistics allow you to provide a simple summary of large amounts of data, making it easier for your reader
(and yourself!) to understand patterns, trends, and key points at a glance. They also help you identify potential issues or
errors within your dataset, so that you can address those before doing further analysis or drawing conclusions.
WHY INFERENTIALS MATTER
Inferential Statistics 101
Inferential statistics help you to draw conclusions about a larger population of interest, based on a sample of data (your
dataset). This approach is not only more practical and cost-effective than studying the entire population, but it also allows
for testing hypotheses. Three common inferential tests include the t-test, chi-squared and correlation (see below).
You’ll need to report on PROBABILITY and EFFECT SIZE when reporting the outcomes of inferential tests.
Probability: Often reported as the “p-value”, is the likelihood of your test being supported by other studies, or that the
outcomes are simply a product of chance.
Effect size: How “big” the effect is (e.g., how strongly variables are correlated or how different the groups actually are
from each other).
You’ll also need to choose a SIGNIFICANCE LEVEL for your tests. This is typically a “cut off” where p-values below this level
are seen as “significant” (e.g., significantly related or significantly different) p < 0.05 (5%) is the more typical significance
level for most fields, but can be lower to provide more cautious interpretations or if you are running multiple tests.
OTHER TESTS
There are many different
types of inferential tests,
beyond the three common
ones we’ve covered here.
Some other tests you may
encounter or use include:
ANOVA: Compares a
numeric variable to two (or
more) groups (e.g., testing
mathematics scores across
three schools). Similar to a
t-test, but it allows for
more than two groups.
Multiple regression: Allows
you to predict the value of a
dependent variable (e.g.,
market price for a house)
using a combination of
other independent variables
(e.g., the number of
bedrooms, bathrooms, floor
space, etc.).
Learn more on our
YouTube channel
@gradcoach