1 Problem Statement

ODU offers a number of courses via the internet. A common requirement among these courses is for a system of online assessment. An assessment is any form of graded question-and-answer activity. Examples include exams, quizzes, exercises, and self-assessments. In preparation for automating such a system, our group has undertaken a study of assessment techniques in traditional classrooms.

An assessment can contain a number of questions. Questions come in many forms, including true/false, single-choice from among multiple alternatives, multiple choices, fill-in-the-blank, and essay. There may be other forms as well.

Students take assessments that are administered by instructors. The students’ responses to each question are collected by the instructor, who grades them by comparison to a rubric for each question. The instructor may also elect to provide feedback (written comments), particularly about incorrect responses.

A total score for the assessment is computed by the instructor. If this is a self-assessment, the score is for informational purposes only. For other kinds of assessments, the instructor records the score in his/her grade book.

Information is returned to the student about their performance. At a minimum, the student would learn of their score and any instructor-provided feedback. Depending upon the instructor, students may also receive the questions, a copy of their own responses, and the instructor’s correct answer.

2 Identifying Candidate Classes and Operations

For the initial list, mark up the description, looking for noun phrases and verb phrases.

ODU offers a number of courses via the Internet. A common requirement among these courses is for a system of on-line assessment. An assessment is any form of graded question-and-answer activity. Examples include exams, quizzes, exercises, and self-assessments. In preparation for automating such a system, our group has undertaken a study of assessment techniques in traditional classrooms.

An assessment can contain a number of questions. Questions come in many forms, including true/false, single-choice from among multiple alternatives, multiple choices, fill-in-the-blank, and essay. There may be other forms as well.

Students take assessments that are administered by instructors. The students’ responses to each question are collected by the instructor, who grades them by comparison to a rubric for each question. The instructor may also elect to provide feedback (written comments), particularly about incorrect responses.

A total score for the assessment is computed by the instructor. If this is a self-assessment, the score is for informational purposes only. For other kinds of assessments, the instructor records the score in his/her grade book.

Information is returned to the student about their performance. At a minimum, the student would learn of their score and any instructor-provided feedback. Depending upon the instructor, students may also receive the questions, a copy of their own responses, and the instructor’s correct answer.

2.1 Candidate Classes

• assessment,
• exams,
• quizzes,
• exercises,
• self-assessments,
• questions,
• true/false question,
• single-choice question,
• multiple choices question,
• fill-in-the-blank question,
• essay question,
• students,
• instructors,
• responses,
• rubric,
• feedback,
• score,
• grade book,
• information,
• performance,
• instructor’s answer

2.2 Candidate Operations

• contain (questions),
• take (assessment),
• administer,
• collect (responses),
• grade,
• provide (feedback),
• compute (score),
• record (score),
• return (information)

3 Assign Operations to Classes

Start by setting up some class diagrams.

3.1 Probable Inheritance Hierarchies

Since all of the various kinds of assessments are likely to have similar attributes and operations, I’m going to set them aside for now.

I’ll do the same with the various kinds of questions.

3.2 Fill in the Candidate Operations

Now fill in the operations known so far:

• contain (questions)

• take (assessment),

• administer,

• collect (responses),

• grade,

• provide (feedback),

• compute (score),

• record (score),

• return (information),

3.2.1 contain questions

An assessment can contain a number of questions.

This is really a statement about attributes of an assessment

• I use “sequence of” or, abbreviated, “seq” for a generic collection of things that have some order, but for which I really don’t even want to hint at whether we would use an array, vector, lists, etc.
• I would call a collection of things that have no special order a “collection of” or “coll”, for short.

3.2.2 taking and administering assessments

Students take assessments that are administered by instructors.

• Is this really two separate operations?

  The language (plurals) is a bit tricky.

  • Instructors administer an assessment to an entire class.

  • Each student individually takes the assessment.

• Surprised that I put that in Instructor?

• Surprised that I put that in Instructor?

• Remember the basic rule: if A does B to C, then “do B” is usually a responsibility of C

• Surprised that I put that in Instructor?

• Remember the basic rule: if A does B to C, then “do B” is usually a responsibility of C

• It would not be a responsibility of the assessment

• Surprised that I put that in Instructor?

• Remember the basic rule: if A does B to C, then “do B” is usually a responsibility of C

• It would not be a responsibility of the assessment

  • Tests don’t administer themselves ITRW

• Surprised that I put that in Instructor?

• Remember the basic rule: if A does B to C, then “do B” is usually a responsibility of C

• It would not be a responsibility of the assessment

  • Tests don’t administer themselves ITRW

• Could it be a responsibility of the Student?

• Surprised that I put that in Instructor?

• Remember the basic rule: if A does B to C, then “do B” is usually a responsibility of C

• It would not be a responsibility of the assessment

  • Tests don’t administer themselves ITRW

• Could it be a responsibility of the Student?

  • No, the statement says that students “take”assessments,

• Surprised that I put that in Instructor?

• Remember the basic rule: if A does B to C, then “do B” is usually a responsibility of C

• It would not be a responsibility of the assessment

  • Tests don’t administer themselves ITRW

• Could it be a responsibility of the Student?

  • No, the statement says that students “take”assessments,
  • But is “take assessment” simply a synonym for “accept administration of an assessment”?

The problem statement tells us:

Students take assessments that are administered by instructors. The students’ responses to each question are collected by the instructor, who grades them … The instructor may also elect to provide feedback (written comments), particularly about incorrect responses.

A total score for the assessment is computed by the instructor. … Information is returned to the student about their performance.

We’re looking at the instructor’s method for administering an assessment.

• a strong suggestion that administration of an assessment is far more involved than simply having a student “take” it.
  • and that they are, therefore, separate responsibilities.

So we add the student’s role into our model:

3.2.3 collecting responses

The students’ responses to each question are collected by the instructor

This is really just describing the output from the request sent to students asking them to take the assessment.

3.2.4 grading responses

the instructor, who grades them by comparison to a rubric for each question.

• not a responsibility of the instructor, because we are still tracing out the steps that constitute the instructor’s method for administering an assignment

We are told there is a separate rubric for each question. So the “comparison” is between a response to a single question and a rubric.

• This highlights the distinction between the response to an assessment and the responses to individual questions.

  • We don’t know what the proper terminology here would be, so we use a placeholder and make a note to consult the domain experts.

At this point, sanity reasserts itself

• ITRW, when a student returns an exam sheet or a bluebook (the Response), those things don’t grade themselves.
  • They’re just paper
  • And while some anthropomorphism is common in OO modeling, that may be going a little too far.

• First thought: the instructor does the grading

  • but it’s not really an operation of the instructor because no one in this model tells the instructor to take this step
  • it’s part of the Instructor’s “administer an assessment” method

• But not mentioning it all in the model seems wrong

  • So let’s think about what we want to eventually capture

One reason that we really want to model the grading process is that we know that we have many different kinds of questions:

and we suspect that the grading method varies from one type of question to another.

• So perhaps we should say

… and here’s why:

… and here’s why:

• Essay questions can cover all kinds of things.

… and here’s why:

• Essay questions can cover all kinds of things.

  • “English” essays where grammar and sentence construction are graded

… and here’s why:

• Essay questions can cover all kinds of things.

  • “English” essays where grammar and sentence construction are graded
  • General essays where content is more important than form

… and here’s why:

• Essay questions can cover all kinds of things.

  • “English” essays where grammar and sentence construction are graded
  • General essays where content is more important than form
  • Mathematical proofs

… and here’s why:

• Essay questions can cover all kinds of things.

  • “English” essays where grammar and sentence construction are graded
  • General essays where content is more important than form
  • Mathematical proofs
  • Code an algorithm, and so on

… and here’s why:

• Essay questions can cover all kinds of things.

  • “English” essays where grammar and sentence construction are graded
  • General essays where content is more important than form
  • Mathematical proofs
  • Code an algorithm, and so on

• Are these really the same kind of question?

… and here’s why:

• Essay questions can cover all kinds of things.

  • “English” essays where grammar and sentence construction are graded
  • General essays where content is more important than form
  • Mathematical proofs
  • Code an algorithm, and so on

• Are these really the same kind of question?

  • They share lots of behaviors (e.g., they are represented the same way on the printed page, students use the same mechanism for answering them)

… and here’s why:

• Essay questions can cover all kinds of things.

  • “English” essays where grammar and sentence construction are graded
  • General essays where content is more important than form
  • Mathematical proofs
  • Code an algorithm, and so on

• Are these really the same kind of question?

  • They share lots of behaviors (e.g., they are represented the same way on the printed page, students use the same mechanism for answering them)
  • But the rubrics for grading them are very different

… and here’s why:

• Essay questions can cover all kinds of things.

  • “English” essays where grammar and sentence construction are graded
  • General essays where content is more important than form
  • Mathematical proofs
  • Code an algorithm, and so on

• Are these really the same kind of question?

  • They share lots of behaviors (e.g., they are represented the same way on the printed page, students use the same mechanism for answering them)
  • But the rubrics for grading them are very different

• So maybe it’s the rubrics that capture this behavior

• I’m much happier with that

• Can rubrics be “intelligent” or is this unacceptable anthropomorphism again?

  • ITRW, rubrics for essay question, in particular, are often expressed in ways that assume a human intelligence

• We’re probably going to wind up with a hierarchy of Rubric classes for different variants.

3.2.5 provide feedback

The instructor may also elect to provide feedback (written comments), particularly about incorrect responses.

• Another case of needing to consult the domain experts to find the proper name for a graded question response.

This suggests that the various grading functions we have written that have been returning scores should really be returning a GradedQuestionResponse.

3.2.6 computing scores

A total score for the assessment is computed by the instructor.

the instructor records the score in his/her grade book.

Information is returned to the student about their performance.

It’s a pretty good bet that we don’t want a class with as vague a name as “Information”.

• But we’ve already encountered the concept under a better name

The clue is the description: “At a minimum, the student would learn of their score and any instructor-provided feedback.”

4 The Story So Far

We might (cautiously) question whether some of the empty class boxes represent classes that we need to retain in the model.

• But we’re still very early in the discovery process

4.1 That’s Far Enough for Now

Right now, we have as many questions as answers.

• but finding useful questions is part of the process

• We can’t go much further without more info

  • … and it’s very dangerous to start making stuff up based on intuition about how we think the program could work.