Marking reliability can be improved in a number of ways:

allocate all assignments to one marker where feasible; alternatively, allocate one particular section of an assignment for marking by the same marker. This will require a system of rotating submitted assessment work across the marking team;
encourage markers to review work marked earlier (e.g. from marking undertaken the day before) to ensure that their marking standard has not changed;
check very high and very low scoring assessments;
use anonymous marking (de-identify the student) to avoid bias towards or against any student in the marking process; and
consider undertaking a second marking of some assessments. In larger units this might involve a random sample being subjected to second (double) marking or blind marking, where markers do not see each other’s comments or grades until after marking is completed. Discrepancy in the grading needs to be considered carefully and resolved. Where markers closely agree, small differences can be resolved by averaging. Larger differences should be handled through discussion involving the Unit Coordinator, and a third marker (ideally the Unit Coordinator) should be used to help resolve any contentious cases.