explain why you cannot draw a conclusion about the effect of one variable in an investigation when the other key variables are not controlled