Toward a framework for evaluating blended learning