Introduction: The Foundation of Statistics

Welcome to the first step of your Statistics journey! Before we can calculate averages or draw fancy graphs, we need to gather information. Think of data collection like cooking: if you start with poor-quality ingredients, it doesn’t matter how good the recipe is—the meal won't be great. In this chapter, you’ll learn how to plan an investigation, identify different types of "ingredients" (data), and choose the best ways to collect them fairly. Don't worry if this seems like a lot of terminology at first; once you see how it applies to real life, it becomes much easier!


1. Planning an Investigation

Every statistical enquiry starts with an idea called a hypothesis. This is a statement that you can test to see if it’s true or false.

Example: "Students who eat breakfast have higher test scores than those who don't."

Constraints and Risks

In the real world, you can’t always do a perfect study. You will face constraints (limitations):

  • Time: You might only have a week to finish.
  • Cost: Printing 1,000 surveys costs money.
  • Ethics: You must keep data confidential and ensure participants aren't harmed.
  • Convenience: It’s easier to ask your friends, but is it fair?

Mitigation Strategies

To "mitigate" means to make a problem less serious. If you think people won't respond to your survey (non-response), you might offer a small incentive or keep the survey short. If you can't find a list of everyone in the town, you might change your target population to just one school.

Quick Review: A good plan considers what could go wrong before you start!


2. Types of Data

Data comes in many "flavours." Using the right name for data helps you decide which graph or calculation to use later.

Qualitative vs. Quantitative

  • Qualitative (Categorical): Described by words. Example: Eye colour, car make.
  • Quantitative (Numerical): Described by numbers. Example: Height, temperature.

Discrete vs. Continuous

This is a common area for mistakes! Here is a simple trick:

  • Discrete: Things you count. It can only be specific values (like shoe size 5 or 5.5, but never 5.23).
  • Continuous: Things you measure. It can be any value on a scale. Example: Time (10.2 seconds), Weight (65.43kg).

Higher Tier Focus: Bivariate and Multivariate Data

  • Bivariate Data: Involves two variables to see if there is a link. Example: Comparing hours studied vs. exam marks.
  • Multivariate Data: Involves three or more variables. Example: Comparing house prices based on size, location, AND age.

Explanatory and Response Variables

When looking for a link between two things:

  • The Explanatory (Independent) variable is what you think causes the change. This goes on the x-axis.
  • The Response (Dependent) variable is what you are measuring. This goes on the y-axis.

Key Takeaway: Always check if your data is measured (continuous) or counted (discrete) before you start your analysis!


3. Population and Sampling

A population is the entire group you are interested in (e.g., every teenager in the UK). A sample is a small group you actually talk to.

The Sampling Frame

A sampling frame is the actual list of people or items from which you choose your sample. Example: If your population is "students in a school," your sampling frame is the school register.

Higher Tier: Stratified Sampling

Sometimes a population has distinct groups (strata), like different year groups or genders. To be fair, your sample should represent these groups in the same proportions as the population.

The Formula:
\( \text{Number to sample from group} = \frac{\text{Number in group}}{\text{Total population}} \times \text{Total sample size} \)

Example: If there are 200 boys and 300 girls in a school (Total 500), and you want a sample of 50 students:
\( \text{Boys to sample} = \frac{200}{500} \times 50 = 20 \)

Other Sampling Methods

  • Simple Random: Like picking names out of a hat. Everyone has an equal chance.
  • Systematic: Picking every \( n^{th} \) person (e.g., every 10th person on a list).
  • Quota: You keep asking people until you have filled your "quota" (e.g., "I need 10 men and 10 women").
  • Opportunity/Convenience: Asking whoever is available at the time. This is often biased.

Did you know? A census is when you collect data from the entire population. It's accurate but very expensive and time-consuming!


4. Collecting Data and Avoiding Bias

Bias is the enemy of Statistics. It happens when your data doesn't represent the truth because of how it was collected.

Reliability and Validity

  • Reliability: If you did the test again, would you get the same result? (Consistency).
  • Validity: Does the test actually measure what it's supposed to? (Accuracy).

Higher Tier Focus: Sensitive Data and Control Groups

When asking sensitive questions (e.g., "Have you ever broken the law?"), people often lie. To fix this, we use the Random Response Technique. This involves the participant flipping a coin or rolling a die in secret to decide whether to answer truthfully or just say "Yes." This protects their confidentiality while allowing the researcher to calculate the overall percentage using probability.

Control Groups and Matched Pairs

In experiments, we use a control group. This group does not receive the "treatment" (e.g., a new medicine) so we can compare their results to the experimental group.

To make this even better, we use matched pairs. We find two people who are very similar (same age, fitness, health) and put one in each group. This helps control extraneous variables (outside factors that might mess up the results).

Data Cleaning

Before using data, you must "clean" it. This involves:

  • Removing outliers (values that are obviously wrong or extremely different).
  • Fixing formatting (e.g., changing "10 mins" and "10m" to just "10").
  • Dealing with missing data or incomplete responses.

Quick Review: Always use pilot surveys (small trial runs) to find and fix confusing questions before you do the real thing!


Summary Checklist

- Can you define a hypothesis?
- Do you know the difference between discrete and continuous data?
- Can you calculate a stratified sample size?
- Do you understand why control groups and matched pairs make experiments fairer?
- Can you explain how the Random Response technique protects privacy?