The foremost consideration when grading students is the type of evaluation. If you're grading the assignment as a series of right or wrong answers, there is no room for subjectivity. If you're grading on a rubric, there's some room for subjective interpretation. This is not to say that you can evaluate rubric-graded assignments such as essays or art projects based on your personal tastes; however, with these types of assignments, you have the ability to make aesthetic judgment calls when grading.
Sometimes, students with perfectly solid understanding of course material can suffer because of confusing test construction. Elements that can make tests difficult to understand include strangely worded questions and unclear instructions. Having a substitute teacher present during the administration of a test can also skew the results, especially if the students report that the substitute was unable to answer questions about the test. If you notice that several students make complaints of this nature and that test or assignment marks were unusually low for the class, consider striking the results from the offending section and computing grades based on the remaining sections.
When grading students, consider the purpose of assessment in general and with respect to the assessment. According to educational researcher and public speaker Alfie Kohn, knowing why we grade students can help us assess students more effectively. He says educators must think about the purpose of assessment at all times, because every assessment is set relative to an educational objective. The logical corollary of this is that every teacher should be willing to make adjustments if a student hands in an assignment that, while failing to meet the specific standards set out in the assignments, nevertheless demonstrates proficiency in the relevant subject matter.
Although the modern American education system generally approves the use of letter grades based on student performance, letter grades are derived by two means. The first and most common method is to give a grade based on the percentage of questions answered correctly. The second method is to give a grade based on each student's standing relative to other students or percentile ranking. This is sometimes called grading on a "curve." If you're grading based on a percentage system, marks are directly convertible to percentages. If grading based on a percentile system, you need to mathematically determine the average grade in the class, the standard deviation from the average, and where each student fits in the picture. This may require you to draw a graph.