Introduction: Telling the Right Story

In Statistics, collecting data is only the first step. To make that data useful, we have to show it to others. Think of a graph like a photo: it should show the truth clearly. However, just like a photo can be filtered or cropped to change how things look, graphs can be chosen poorly or designed to mislead. In this chapter, you will learn how to pick the best "filter" for your data and how to spot when someone is trying to "Photoshop" the truth!

Part 1: Choosing the Right Representation

The first rule of Statistics is that the type of data you have tells you which graph to use. If you pick the wrong one, your data might not make sense to your audience.

1. Categorical and Qualitative Data

When your data is in categories (like "Favourite Colour" or "Type of Pet"), you want to show how many items are in each group.

  • Bar Charts: Best for comparing the size of different categories.
  • Pie Charts: Best for showing "parts of a whole" (e.g., what percentage of the class likes dogs).
  • Pictograms: Great for making data look engaging, but they must have a clear key.

2. Discrete Quantitative Data

When you have numbers that you count (like "Number of Siblings"), you usually use:

  • Bar Line Charts (Vertical Line Charts): These use thin lines instead of wide bars to show that the data is discrete (you can't have \(2.5\) siblings).
  • Stem and Leaf Diagrams: Good for showing the actual raw data values while still seeing the shape of the distribution.

3. Continuous Quantitative Data

When you have measurements (like "Height" or "Time"), the data can be any value.

  • Histograms: Best for grouped continuous data. Note for Higher Tier: If the class widths are different, you must use frequency density on the vertical axis.
  • Frequency Polygons: Useful for showing the "shape" of the data or comparing two different sets of data on the same axes.

4. Relationships and Trends

Sometimes you want to see how two things change together.

  • Scatter Diagrams: Used for bivariate data (two variables) to see if there is a correlation between them (e.g., "Ice cream sales" vs. "Temperature").
  • Line Graphs / Time Series: Used when the horizontal axis (\(x\)-axis) represents time. They help us see trends (is it going up or down over the months?).

Quick Review: Before you draw, ask yourself: "Is my data words or numbers?" and "Am I comparing groups or looking for a trend?"

Part 2: Spotting Misleading Graphs

Sometimes, graphs are drawn incorrectly—either by mistake or on purpose to make a point seem stronger than it really is. As a statistician, you need to be a "Data Detective."

1. The Truncated (Broken) Axis

This is the most common trick. The vertical axis (\(y\)-axis) should usually start at \(0\). If it starts at a higher number, it can make small differences look huge.

Example: If one bar is at \(100\) and another is at \(105\), but the graph starts at \(95\), the second bar will look twice as tall as the first!

2. Unequal or Missing Scales

Always check the numbers on the axes. Are the gaps between the numbers equal? If the \(y\)-axis goes \(0, 10, 50, 100\), the scale is inconsistent and the graph is misleading.

3. Distorted Sizing (The Area Trick)

In pictograms or 3D charts, if you double the height of a picture, you often double the width too. This makes the area look four times bigger, even if the value only doubled.
Remember: In a fair diagram, the area should be proportional to the frequency.

4. 3D Effects

Using 3D pie charts or bars can be confusing. The pieces at the "front" of a 3D pie chart often look much larger than the pieces at the "back," even if they represent the same percentage. Always stick to 2D for clarity.

5. Higher Tier: Incorrect Frequency Density

In a histogram with unequal class widths, the height of the bar must represent frequency density, not just frequency. If a graph uses frequency on the \(y\)-axis for bars of different widths, it is misleading because the widest bar will look like it has the most data just because it's "fat."

The correct formula is: \(Frequency\ Density = \frac{Frequency}{Class\ Width}\)

Summary Checklist for the Exam

If you are asked to evaluate or criticise a graph, look for these "Red Flags":

  • Labels: Are the axes labelled? Is there a title?
  • The Origin: Does the \(y\)-axis start at \(0\)? If not, is there a "break" symbol?
  • The Scale: Are the increments (the jumps between numbers) equal?
  • The Key: If it's a pictogram or multiple bar chart, is there a key to explain the symbols/colours?
  • The Data Type: Is it a bar chart for continuous data? (It should probably be a histogram).

Did you know? Misleading graphs are often called "wicked" graphs in the media because they are designed to grab your attention and make you believe something that isn't quite true!

Key Takeaway: A good statistical representation should be honest, clear, and appropriate for the data type. If a graph looks too dramatic, check the axes first!