Lab reports involve a unique combination of objective and subjective elements. This is a brief explanation of how I assign grades.
Remember that your lab report grade is 40 percent of your grade for a particular lab assignment. The other 60 percent comes from your preparedness/participation, timeliness with hitting milestones, and the quality of your final demonstration.
Documentation is part of engineering. This is a design class, and you will be evaluated on both the design/thoughtfulness of your engineered device, and also the design/thoughtfulness of the documentation for that device.
Every lab assignment has a set of required sections that will be specified on the course website. For 4760, these required sections are listed on the policy page. Previous TA's have expanded on these guidelines and the course website includes a couple example reports. My first step in grading a collection of lab reports is to scan through all of them to briefly check the following:
Having all required sections and having non-trivial content within each section will bring a group's grade up to approximately a B. If the content in any of these sections is lacking or trivial, their grade after the objective scan will be lower than a B by one half-letter grade per trivial/missing section. After the objective scan, a group's grade may be improved but will not be worsened by the subjective readthrough.
After scanning through each lab report once, a second pass is performed in which each lab report is read very thoroughly. In this thorough readthrough, I am asking myself the following question:
If I only had this lab report as a resource, could I:
My advice to students is to consider the audience of their lab reports to be themselves in 5 years. Assume that they haven't been working with this particular microcontroller since this course, and must re-create their work in one day. Could they do it (and justify the solution to a boss) using only the report?
When performing a subjective evaluation of a group of lab reports, I am trying to determine how well each student has communicated the technical content in the report. In other words, I'm not only asking "Could I recreate this student's work using this report?" but also "How easily could I recreate this student's work?"
This information may be communicated in a number of different ways, including written text, captioned images, figures, plots, code snippets, pseudocode, figures, flow charts, diagrams, etc. Different groups may use very different strategies for communicating the same information, and that's perfectly ok.
For this reason, however, it is very difficult to judge a lab report in isolation. There is no "perfect" lab report against which to compare a particular lab report. However, it can be quite obvious that one is communicating information more clearly than another. The two contain the same information, but one presents that information in a way that is clearer. This is of course subjective (like an essay), but clarity is an important element of the lab report.
The next step in the grading process is to rank and bin the assignments for content and clarity. I read thoroughly through each report and ask myself "does this report communicate the student's ideas more or less clearly than this other report?" I do that for all combinations of reports, and end up with a set of rank-ordered bins.
I don't enforce any number of bins though. If all lab reports are equally excellent, then they all end up ranked together!
I then consider the top-ranked report (or reports). I ask myself What would I need to change about this report to use it as an example in future semesters? If the answer to that question is "Nothing" then the report gets an amount added to its objective score to bring it to a 100 percent. If anything would need to change, then the top-ranked report is given a score that is less than 100 percent. The total deduction is based on the number of required changes.
There is no "correct" way to present information. There may be two equally excellent reports that report the same information in very different ways.
I next move down the rank-ordered list. For each report in the next-lower bin, I ask myself How does this report compare to the one ranked above it?. If one is only marginally clearer than the other, then it receives only a marginally better score. If the differences in clarity are greater, then the difference in scores are greater. Each is evaluated relative to the one ranked above it.
Once all reports have been assigned their grades, I then start again from the bottom to revise and confirm that the scores are fair. I will consider the worst-ranked report in the class and ask myself How many changes would be required for me to use this as an example in a future class? If the previously assigned score matches (in a subjective sense) with the first-pass grade, then the grade remains the same. If not, then the grade is adjusted and all intermediately-ranked grades are also adjusted relative to the report ranked below it.
After this second pass from bottom-to-top, the grades are finalized.