Lesson 3.02: Evaluating Evidence

In this module we’re continuing our journey into the world of evaluating evidence—but we’re going deeper. You’ve already learned how to assess the credibility of a source. Now, we’re going to sharpen your skills even further by asking:

Does this evidence truly represent the whole picture?
Is the way this data is presented helping us understand it—or misleading us?

Representativeness

When we look at evidence, especially in research or statistics, we need to ask:

Does this evidence fairly represent the whole group it’s talking about?

Sometimes, researchers can study every single member of a group. This is possible when:

The group is small and clearly defined, or
The data is already available (like birth records or school exam results).

Example:

If a school wants to see if a new A Level teaching policy worked, it could compare the exam results of all students from the year before and after the change.

But—this only works if the two groups of students are similar in every other way (same ability levels, same teachers, etc.).

When We Use Samples

Most of the time, researchers use samples—smaller groups that are supposed to represent the larger population.

But samples can be unrepresentative if:

Everyone in the sample shares a trait that others don’t (like all being the same gender or age).
That trait could affect the results.

Example:
If a study on job satisfaction only includes people under 25, can we assume the results apply to older workers too? Probably not.

Sample Size

A sample is always smaller than the full group—but it still needs to be big enough to draw valid conclusions.

If the sample is too small, the results might not be reliable.

Selectivity (a.k.a. Cherry-Picking)

Sometimes, people only choose evidence that supports their argument and ignore the rest.

This is called cherry-picking—like picking only the best-looking cherries from a tree and pretending they represent the whole crop.

Watch out for this in news articles, advertisements, or biased reports.

Proxy Measurement

Sometimes, researchers can’t measure exactly what they want—so they measure something related instead. This is called a proxy.

Example:

We can’t count every species that goes extinct (some disappear before we even discover them).

So scientists use habitat loss as a proxy to estimate extinction rates.

But this method relies on models and assumptions—and we can’t always check how accurate they are.

⚠️ Be cautious when someone uses a proxy to make a big claim—especially if they have something to gain from it.

Extrapolation

Extrapolation means taking a trend and extending it into the future—even if we don’t have data for that future yet.

Example:

“If our sales keep growing like this, we’ll be billionaires in 3 years!”

Sounds exciting—but if the prediction is based on very little data, or if the current data is unusual, the claim is probably not reliable.

Evaluating the Presentation of Data

When data is shown in a graph, chart, or table, it’s meant to help us understand the information more clearly. But sometimes, the way the data is presented can be misleading—intentionally or not. Let’s look at some common ways this can happen.

Starting the Y-Axis at a Value Other Than Zero (Point Zero)

Graphs should usually start the y-axis at zero to give a fair picture of the data.

If the graph starts at a higher number, it can exaggerate small differences and make them look more dramatic than they really are.

Example:

A bar graph showing sales increasing from $98 to $100 might look like a huge jump—if the y-axis starts at $95 instead of $0.

Irregular Intervals on Axes

Graphs should have even spacing between values on both the x- and y-axes.

If the intervals are uneven, it can distort the trend and make it look like something is increasing or decreasing faster than it really is.

Watch out for:

Uneven time gaps on the x-axis
Skipped or compressed values on the y-axis

Misleading Use of Symbols or Images

Using pictures or symbols in graphs can make data more engaging—but also more confusing.

If the size of the symbol (like a circle or 3D shape) grows in height, width, and depth, it can make the change in data look much bigger than it actually is.

Example:

If one image is twice as tall and twice as wide, it’s actually four times bigger in area, not just twice. That can visually exaggerate the difference.

Graph 1

sets out all the figures from a neutral perspective

Graph 2

makes the growth in income appear greater than it is, because the y-axis does not begin at zero. Yet a graph which did begin at zero would show barely any variation at all over that period, which would not be of much use. By ignoring the period before the present government came to power, Graph 2 distracts attention from the fact that the rise in income has been roughly consistent under successive governments.

Graph 3

appears to overstate even further the growth in income between 2011 and 2015, because it uses a two-dimensional image to express a one-dimensional increase; furthermore, the image may be mentally interpreted as three-dimensional. It is also not very easy to see exactly what values are being expressed by the symbols.

Graph 4

appears to show that taxation is rising at a lower rate under the present government, by using irregular points on the x-axis (years).

Page updated

Report abuse