In scientific investigations there are three types of variable - key independent variable (what you change), dependent variable (what you measure) and the controlled variables (factors you control to ensure they are the same throughout your experiment to make it fair). This GCSE Biology quiz is all about the third type.
Controlled variables make sure that your experiment is a fair test, but sometimes you may miss one of the control variables. When this happens the results and conclusion may not be fully reliable. At other times, some important variables may be outside of your control or be extremely difficult to control. In some cases, changes in a variable that have not been measured could actually be the cause of changes in the dependent variable. This can lead to a false correlation between the variables you studied but this is rare. Where you do identify a control variable outside of your control, you should at least try to monitor it. That way, you can probably spot if it has had an effect on the results even if you can't measure it.
Experiments in biology are actually very difficult to control completely. Take the study of "Does adding fertiliser make a plant grow bigger?". For a start, there are many different fertilisers so the results could differ from one to another. You would need to do many different experiments. The plants you use is a variable that needs to be controlled so you use the same type of plant in all of your tests. But genetics of each plant should also be controlled by using clones. Even if you did use cloned plants, there are other factors that should be controlled to ensure a fair test, like using the same soil. same size pot, same amount of light and temperature and some other factors too.
The conclusions that you draw from any biological experiment should show that you understand this and suggest some of the places that inaccuracy could have come into your experiment through no fault of your own. To overcome this, professional scientists repeat experiments many, many times so that they can 'average' the results. This helps to reduce any inaccuracies caused by the controlled variables.