Drawing conclusions and evaluations

In a nutshell

Evaluating your data is important to ensure your results are valid and your conclusions can be supported. Looking at correlation and percentage error are ways to evaluate your data.

Drawing conclusions

When analysing your data, you need to make sure your conclusions are valid. They can only be considered valid if they answer the original question that you set out to investigate. One way to draw conclusions is to look at the correlation between two variables.

Relationship	Description	Example
Positive correlation	When 'A' increases, 'B' also increases.
Negative correlation	When 'A' increases, 'B' decreases.
No correlation	There is no relationship between 'A' and 'B'.

Note: Correlation does not equal causation. Just because it appears that there is a relationship between two variables, it does not mean that one is causing a change in the other.

Uncertainty

Definition

Uncertainty is the amount of error your measurements might have. There is always a degree of uncertainty in your measurements because the equipment and apparatus have a limited sensitivity.

Example

A measuring cylinder may measure to the nearest $1\ cm^3$ but the real volume may be $0.5\ cm^3$ higher or lower. The measuring cylinder has an uncertainty value of  $\pm 0.5\ cm^3$ , this is called the margin of error.

Calculating percentage error

Percentage error can be calculated using this formula:

$percentage\ error = {uncertainty\over reading} \times 100$

Example

Calculate the percentage error of $15\ cm^3$ of water, with an uncertainty value of $0.01\ cm^3$ .

Input the values into the formula:

$percentage \ error = {0.01\over 15} \times 100$

Simplify:

$percentage\ error = 0.00067\times 100= \underline{0.067\%}$

$\underline{ Therefore,\ the \ percentage\ error\ is\ 0.067\%.}$

It is possible to minimise errors in your measurements by using the most sensitive equipment you have available.

Evaluation

In your evaluation, you need to assess the repeatability, reproducibility and validity of your experiment and data. You also need to evaluate your method and explain whether your results confidently back up your conclusion.

There are some important questions you should answer when evaluating.

Did you do enough repeats?
Do you think you'd get similar results if you were to repeat this experiment? Why?
How does your data compare to existing data?
Do other people make the same conclusions from your results?
Does your data answer the question you set out to answer?
How could you make your results more reliable and accurate?
What were the limitations in your method?
What were the sources of error in your experiment?
How could you improve the experiment if you were to do it again?