Tuesday, April 8, 2008

Chapter 7 – Constructed Response Tests

Educational Assessment - Review By Brenda Roof
Classroom Assessment – What Teachers Need to Know - W. James Popham

Constructed-response tests are the focus of chapter seven. Constructed-response tests are great for assessing if a student has achieved mastery. The creation and scoring of constructed response and effort involved is greater than selected response. They should be used when teachers want to know what is truly mastered. Scoring should be established ahead of time to preserve valid inferences. There are two kinds of constructed-response assessments discussed, short answer and essay.
The first type of constructed response discussed is short-answer item. Short-answer items require a word, phrase or sentence in response to a direct question or incomplete statement. Typically, short-answer response helps answer learning outcomes such as those focused on. The advantage of short-answer is students must produce a correct response, not pick-out a familiar choice. The major disadvantage of short answer is, responses are difficult to score. In accurate scoring leads to reduced reliability and then reduces the validity of the assessment based inferences made about students.
There are five item-writing guidelines for short-answer items. The first guideline is to employ direct questions rather than incomplete statements, particularly for younger students. Younger students especially will be less confused by direct questions. The use of direct question format also helps ensure the item writer phrases the item so less ambiguity is present. The second short-answer guideline is to structure the item so that a response should be concise. The test items should be created to elicit brief clear responses. The third guideline is to place blanks in the margin for direct questions or near the end of incomplete statements. This will also aid in scoring so items are aligned. By placing fill-in-the-blanks toward the end students will read the whole statement to ensure more accurate responses. The fourth guideline is for incomplete statements; use only one, or at most, two blanks. The use of more than two blanks leaves the item with holes galore causing meaning to be lost. The fifth guideline for short-answer response items is to make sure blanks for all items are equal in length. This practice will eliminate unintended clues to the responses.
The second kind of constructed response assessment item is essay items. Essay items are the most common form used of constructed response. Essay items are used to gage a student’s ability to synthesize evaluate and compose. A form of essay item is a writing sample. These are used heavily in performance assessments. A strength of essay item is assessment of complex learning outcomes. Some weaknesses of essay items are, they are difficult to write properly, and scoring responses reliably can also be a challenge.
There are five item-writing guidelines for essay items. The first guideline for essay items is convey to students a clear idea regarding the extensiveness of the response desired. There are two forms used for this, a restricted-response, which limits the form and content of the response. The second is an extended-response item, which gives more latitude in the response. When using these forms you can provide certain amount of space or number of word limits. The second guideline is to construct items so the student’s task is explicitly described. The nature of the assessment task has to be set forth clearly, so the student knows exactly how to respond. The third guideline is to provide students with the approximate time to be expended on each item, as well as each items value. Directions should state clearly how much time and the point values of your essay item responses. The forth guideline is to not employ optional items. Offering a menu of options in turn represents different exams altogether, the consequence is the impossibility of scoring on a common scale. The fifth guideline for essay-items is recursively judge an items quality by composing, mentally or in writing, the item as well as, a precursor for your expectations for a response.
As mentioned earlier the most difficult problem with constructed response items is scoring these items. There are five guidelines for scoring responses to essay items. The first is score responses holistically and/or analytically. Holistic scoring focuses on the essay response as a whole using evaluative criteria. The second scoring focus is analytic; this is a specific point-allocation approach. The second guideline is to prepare a tentative scoring key in advance of judging responses. To avoid being influenced by the students actions in class or quality of the first few responses decide ahead of time how you will score. The third guideline is to make decisions regarding the importance of mechanics writing prior to scoring. It is important to decide upfront how you will score mechanics. If the material is more important to what inferences you want to make, be sure to establish this before you begin scoring. The fourth guideline is to score all responses to one item before scoring responses to the next item. When scoring that item is complete, go to the next item on the same essay as well. Complete a whole essay at once. When scoring the whole essay at once, you increase the reliability of your scoring. This will also avoid having to constantly shift your focus to what you are scoring. It seems like it would take longer but it really wont. The fifth guideline is as much as possible, evaluate responses anonymously, try to have students sign their papers on the back and while scoring do not look at the names. This will help not making judgments outside your scoring rubric.
Assessing students using the constructed response format allows for more in-depth awareness of learned skills. There are two forms used for this short-answer and essay-item. There are also guidelines to implementing these item responses, as well as, scoring guidelines to ensure greater reliability and validity to score-bases inferences.

No comments: