A dataset can establish a pattern within the conditions that were tested. It cannot establish what happens outside those conditions, why the pattern occurs, or that it holds for other materials. Examiners reward a conclusion that says exactly that much and no more.
This lesson follows comparing a repeat with an independent method, which showed how far a single measurement can be trusted. Here the question is how far a whole table can be trusted.
What does a good conclusion contain?
It has four parts: the pattern, the range, the support and the limit. The pattern says how the dependent variable changes.
The range names the tested values. The support quotes figures from the data. The limit says what the data do not show.
Step by step
- Describe the pattern in words, with direction.
- Quote the data, including a calculated value such as a rate or a ratio.
- State the tested range and keep the claim inside it.
- Name what was not tested: other values, other materials, the mechanism.
- Separate the observation from the explanation. An explanation is a further idea that can be tested.
Worked example
The data are invented. A student times how long one antacid-type tablet takes to disperse completely in 100 cm³ of water at four temperatures. One run was done at each temperature.
| Temperature (°C) | Time (s) | Rate = 1 ÷ time (s⁻¹) |
|---|---|---|
| 20 | 95 | 0.0105 |
| 30 | 62 | 0.0161 |
| 40 | 41 | 0.0244 |
| 50 | 28 | 0.0357 |
Check the rates: 1 ÷ 95 = 0.0105, 1 ÷ 62 = 0.0161, 1 ÷ 41 = 0.0244 and 1 ÷ 28 = 0.0357, all to 3 significant figures.
Pattern and support: As temperature rises from 20 °C to 50 °C, the time falls from 95 s to 28 s. The rate at 50 °C is about 3.4 times the rate at 20 °C, because 95 ÷ 28 = 3.39.
What the data establish: within 20 to 50 °C, this tablet in this volume of water dispersed faster at higher temperature.
What they cannot establish:
- The behaviour at 60 °C or 10 °C, because those values were not tested.
- The reason, such as particle collisions, because no particles were measured.
- That another tablet type behaves the same way.
- How reliable each time is, because there was only one run per temperature.
A strong answer writes: “The data support the hypothesis between 20 and 50 °C for this tablet. A repeat at each temperature would show the reliability.”
The mistake to watch for
Students often write a conclusion that explains more than the table shows.
Mistaken answer: “This proves that higher temperature always makes reactions faster because particles move faster.”
“Proves” and “always” claim every temperature and every reaction. “Because particles move faster” offers a mechanism the experiment did not measure. A better wording is: “In this investigation, rate increased from 0.0105 s⁻¹ to 0.0357 s⁻¹ between 20 and 50 °C.”
Check yourself
1. Using the table, a student predicts that the time at 60 °C would be about 19 s. Is this established by the data?
Show answer
No. 60 °C is outside the tested range, so it is only an extrapolation. It could be tested, but the data alone do not establish it.
2. Calculate the rate at 40 °C from the time of 41 s, to 3 significant figures.
Show answer
Rate = 1 ÷ 41 = 0.0244 s⁻¹.
3. Pupils find that students who sleep longer scored higher in a test (a survey of ten pupils, invented). Why can they not conclude that sleep causes the higher score?
Show answer
Many other things differ between pupils, such as revision time and prior knowledge, and none was controlled. The survey shows the two variables go together, not that one causes the other.
Where this leads next
When your conclusion stays inside what the data support, you are ready to suggest changes. Continue to proposing a safe improvement connected to a real limitation. The scientific investigation critic prompts you to write the support and the limit as separate lines.
Careful conclusions are a habit that grows with feedback. In online one-to-one Co-ordinated Sciences tuition, a teacher can respond to your own written conclusions.