In research, an independent variable is the variable that is the "input" and the dependent variable is the outcome or perceived outcome. In organizational behavior, one example would be to measure productivity in an entity with a "flat" organizational structure, as opposed to an organization with a layered hierarchy. In such a case, the independent variable would be structure and the dependent variable would be productivity.
The difficulty in research that has qualitative variables is always in isolating the independent variable in order to show some cause and effect, a way of demonstrating that it is only the independent variable that causes a particular dependent variable. Using the above example, one might look at two organizations with different structures and measure productivity, concluding that the flatter structure produces better results. But there are so many other variables that could create the outcome, for example, a different work culture, a different management style, a better-trained workforce, and so on. How do we know that what we have called the independent variable is creating the dependent outcome? Sometimes, we simply do not.
A well-designed study tries to control for stray independent variables to the degree possible, trying to match the two groups as closely as possible with the exception of the independent variable. So, a study looking at structure and productivity would try to "match" for work culture, management style, the training of the workforce, and any other attribute that could influence productivity.
One can find a correlation between an independent and dependent variable, but without a good match of all other variables, it becomes difficult to infer causation. It is hard to know if an independent variable causes a dependent variable if other variables cloud the picture.