In statistics, the term floor effect refers to when data cannot take on a value lower than some particular number, called the floor.
An example of this is when an IQ test is given to young children who have either (a) been given training or (b) have been given no training. If the test is too difficult (so difficult that no amount of training will affect the ability to carry out the test), both group (a) and group (b) will perform particularly badly. This does not necessarily lead to the conclusion that the training has no effect on the ability to complete the IQ test. In fact, this may lead to a Type II error. The IQ test is too difficult, and by making the questions less difficult, training may have an effect on the ability to complete the IQ score.
Here, the floor effect is the data all hitting the bottom end of the distribution due to the extreme difficulty of the task. A ceiling effect is precisely the opposite - all participants reaching the high end of the distribution (e.g. the test was too easy).