Numerical Systems

Introduction Positional Notation Converting Between Bases Binary Addition Binary Multiplication Challenges (Python) Challenges (Java)

Modular Arithmatic

Introduction The Modulus Operator Cyphers Modular Arithmetic Modular Inverse Challenges (Java)

Algorithms

Introduction Algorithms Time Functions Asymptotic Complexity Summation Challenges

Propositional Logic

Introduction Propositional Logic Conjunction and Disjunction Implication and Equivalence Bitwise Operators Challenges

Set Theory

Introduction Sets Subsets and Supersets Set Operations Challenges

Graph Theory

Introduction Graphs Directed Graphs Structure of Language Search Trees Neural networks Game Object Trees Challenges Graph Question Generator

Proof

Introduction Proof Proof by Cases Proof by Contradiction Proof by Induction Challenges

Maths to Code

Maths to Code Challenges Question Generator

Descriptive Statistics

Introduction Statistics Measures of Central Tendency Measures of Spread Data Visualisation Correlation Challenges

Probability

Introduction Probability Basics Combinatorics Probability Trees Law of Total Probability Bayes Rule Challenges

Inferential Statistics

Introduction Probability Distributions Samples and Populations Hypothesis Testing Z Scores Student's T-Test Challenges

Student's T-Test

We are going to learn about a common inferential statistic called the Student’s T-Test (it was published under the psedonym ‘Student’). It is a so-called parametric test, because it assumes your data follows a normal distribution.

Watch the video and then answer the questions below.

Thirty four-minute video

You can also view this video on YouTube

Key Points

The t-test calculates the T statistic
The T statistic can be converted to a p-value by comparing it to the t-distribution
The t-test assumes our data is normally distributed
T-tests can compare only up to two groups

One-sample and paired t-test

If we want to compare a group mean against a known value (e.g the population mean $μ$ ), or a mean of a group of differences with a known value (e.g. (\ \mu = 0\)), we use the following formula:

$t = \frac{\overset{―}{x} - μ}{\frac{s}{\sqrt{n}}}$

We use the t-test (as opposed to the z-test) when we don’t know the population standard deviation $σ$ , so we use the sample standard deviation $s$ . Because of the uncertainty in calculating $s$ , we get a t-statistic instead of a z-statistic, and have to compare it on a t-distribution.

Two-sample t-test

If we want to compare a differences between group means ( ${\overset{―}{x}}_{1} - {\overset{―}{x}}_{2}$ ) against an expected difference (e.g. 0 as in the formula below), we use the following formula:

$t = \frac{({\overset{―}{x}}_{1} - {\overset{―}{x}}_{2}) - 0}{\sqrt{\frac{s_{1}^{2}}{n_{1}} + \frac{s_{2}^{2}}{n_{2}}}}$

Where ${\overset{―}{x}}_{1}$ and ${\overset{―}{x}}_{2}$ are the means of the two groups, $n_{1}$ and $n_{2}$ are the number of observations in the two groups, and $s_{1}$ and $s_{2}$ are the standard deviations of the two groups.

Questions

1. Check your understanding

1. Pick the appropriate statistical test formula

	Expression	$z = \frac{x - μ}{σ}$	$z = \frac{\overset{―}{X} - μ}{\frac{σ}{\sqrt{n}}}$	$t = \frac{\overset{―}{x} - μ}{\frac{s}{\sqrt{n}}}$	$t = \frac{({\overset{―}{x}}_{1} - {\overset{―}{x}}_{2}) - 0}{\sqrt{\frac{s_{1}^{2}}{n_{1}} + \frac{s_{2}^{2}}{n_{2}}}}$
1.	I compare the average height of two groups
2.	I compare a group’s performance in a puzzle against a theoretical mean that assumes completely random behaviour
3.	I assess the IQ of a group assuming $μ = 100, σ = 15$
4.	I investigate if drinking coffee increases a participant’s heart rate compared to a resting value
5.	I run a counterbalanced game enjoyment study. Each participant plays two games and rates each of them. I want to see if one game is more enjoyable than the other

Check Answers

2. Calculate the t statistic

I collect 10 sensor readings each from 2 sensors. I want to see if there is a difference between the means of their readings.

Group 1	Group 2
0.4	6.3
3.6	-1.2
3.3	-11.3
1.5	-6.3
-1.7	-5
0.1	-3.4
4.2	2.4
-1.8	14.7
1.9	-2.9
-3.6	9.9

We should use a:

One-Sample T-test Paired T-test Two-Sample T-test

We get a t statistic of:

t = (2 decimal places)

Here there are 20 data points. Because we “spend” 2 of them to calculate the mean for each group we are left with 18 degrees of freedom (df). We do a 2 tailed test against an $α = 0.05$ . Look up our t and df in a table of t-statistics for different alpha values. If our t value is larger than the one listed for our t and df we have significance.

Is our result significant?

Yes No