(pressing HOME will start a new search)
|
|
CONSTRUCTION
OF TESTS USING COMPUTER SPREADSHEETS
Hendrick
D. Mol |
The making of
new tests for each new class of students is a tedious and repetitive
task, but a necessary part of teaching. This paper presents a model
using a computer spreadsheet to make multiple-choice tests for course
subject material that can be tested by questions that have numeric
solutions. New tests are generated by simply changing the given numeric
data in the test model, which then automatically yields new correct
responses and new plausible incorrect responses to the questions. Use of
this technique saves considerable time and effort in the construction of
computer graded, multiple-choice tests. KEY WORDS:
Computer spreadsheet Multiple-choice tests |
INTRODUCTION
When course subject matter can be tested by problems and questions with numeric solutions, spreadsheet software can be used to construct and print new multiple-choice tests relatively quickly once the basic test is constructed using the spreadsheet. All questions need not have numeric answers, but this type question provides the most savings in time and effort when making new tests from the basic test model. The construction of a computer generated test is divided into three parts: the statement of the test problem and/or question, the test problem computer solution, and the multiple-choice computer generated responses. All three parts become a spreadsheet template for making new7 tests.
STATEMENT OF THE TEST PROBLEM/QUESTION
All
the figures accompanying this paper provide examples of test problem statements
and questions generated with spreadsheet software. The row numbers and column
letters are provided for reference to location of information cells and are not
printed on the student's test question sheet.
Numeric data manipulated by formula
The
spreadsheet allows problems and questions to be stated in sentence format, but
numeric data that are manipulated by formulas must be in unique cells.In Figure
4, the data in cells E137 through E140 are manipulated by formula to create
correct and incorrect responses to the multiplechoice questions and are
therefore in unique cells. If the problem solution is lengthy, the data can be
entered and manipulated at a separate location on the spreadsheet. This will be
discussed later under problem solution.
Data changes for new tests
Data
in the problem statement or in the question which will change on future tests
should preferably be in unique cells so that the data may be changed
efficiently. Otherwise, the changed data will require that the sentences be
edited in addition to changing the data in the unique cells to be manipulated by
formula.
In
Figure 1, the data in cells G9,G11, and G15 are in unique cells because they
will be changed when constructing a new test. The new data are entered at cells
N4,N6, and N11. Cell 123 in question 3 is unique and is updated automatically
from cell Q23 in Figure 1, so that the question's answer is not dependent on the
student's solution of book value. In Figure 3, the numeric data in question 17
are not in unique cells and must be edited in the sentences if they are to be
changed for a new test. The new data for question 17 are entered in the problem
solution cells N88, 89, and 90. A note is provided near the newly entered data
to remind the author to edit the question statement at cell 8116.
In
Figure 7, any of the financial accounts may be varied in dollar amount for a new
test, and the totals, financial ratios, and any question responses dependent on
the changed variables will automatically change.
Questions without numeric answers
The
spreadsheet may contain questions which do not have numeric answers. Figure 4,
question 27, and Figure 6, questions 1 and 2 are examples. Use of a spreadsheet
to generate non-numeric questions has no advantage over wordprocessor
software. Therefore, the nonnumeric questions should only be included if
necessary to compliment testing of numeric solutions to the subject matter.
Variation between tests
New
tests are generated primarily by changing the numeric value of given data. Cells
which contain data that will be updated with each new test are
"un-protected" and all other cells are "protected". This
aids in visually determining where new data is to be entered because the
"un-protected" cells are highlighted on the monitor screen. A box has
been placed around "unprotected" cells in each figure to aid the
reader in determining where new data will be entered.
Alternate
questions may be placed on the spreadsheet as shown in Figure 6 and may be
exchanged with previously used questions to provide variation between tests.
Additional variation between tests is achieved by editing and reordering
questions and responses. Some reordering is necessary to change the key for
correct responses to the questions.
Question dependency
It
is desirable on a computer-graded, multiple-choice test to make all the
questions independent of each other, but this is difficult if not impossible to
do on lengthy problems. This situation may be seen in Figures 1,2 and 3. In
Figure 1, the correct response to question 2 is dependent on having the correct
response to question 1. The remaining four questions in Figure 1 have solutions
that are independent of the other questions. In Figure 2, the correct response
to question 9 is dependent on knowing the correct answers to questions 7 and 8.
In Figure 3, the correct response to question 15 depends on knowing the correct
response to question 14.
A
solution to the question dependency problem is to create many short independent
problems. This was done for the equipment problems in Figure 5. Yet, there is
merit in testing the "bottom line" answer as well as the subpart
answers even though the "bottom line" answer is dependent on the
subparts.
PROBLEM/QUESTION SOLUTION
All
numeric data must be available in unique cells for manipulation by formulas
designed to solve the problem or subparts of the problem. As an example, in
Figure 1, cell N7 contains the formula 0.2*N4 or 0.2 times the content of cell
N4. Cell N4 contains the numeric value 500,000., and the value of the formula,
100,000., is produced in cell N7. The formula in cell N12 could be either +N4-N7
or O.8*N4. In either case the value in cell N12 is 400,000. Additional
representative cell formulas are provided in Figures 1 through 5.
Short test problems
If
the problem solution is not lengthy, the formulas for correct and incorrect
responses may be in the question response cells. This is the case in Figure 2,
questions 8, and in Figure 4, question 21. Each question response cell contains
a formula to generate the response based on data in the problem statement and is
independent of the problem solution.
Lengthy test problems
When
the problem solution is lengthy and can generate many test questions on the
subparts, it is desirable to solve the total problem at a location on the
spreadsheet separate from the test problem statement and questions. The test
problem is designed so that new values for the variable parameters are entered
at the problem solution cell block, and the variables in the problem statements
and questions are automatically changed if they are in unique cells.
Figures
1, 2, and 3 contain problems of this type. In each case the problem is solved in
its entirety, even though the student may not have to solve the complete problem
to answer the test question. Having the entire problem solved with all the
intermediate steps of the solution in one area of the spreadsheet aids in
determining appropriate subpart questions and responses. Another advantage to
this technique is that the entire problem solution can be printed separately
from the test and used to review the problem with the students after the test.
Homework exercises
Homework
exercises may also be generated easily from the problem solution format on the
spreadsheet. Figure 6 provides an example, although the lengthy problem
solutions in all the examples provided are used by the author to produce
homework exercises.
MULTIPLE-CHOICE
RESPONSES:
All
the multiple-choice responses are in unique cells which contain solution
formulas to either provide a numeric answer or accept a numeric answer from a
unique cell in the solution if it is at another area of the spreadsheet. In
Figure 4, the correct response to question 21 is C for response cell G143 which
contains the formula +E138/(lE139-E140). Incorrect responses preferably
contain typical student errors, but in this case the incorrect answers, A and B,
are 60 and 80 percent respectively of the correct answer. Incorrect response D
is generated by the formula (E139+E40)*El37+E138 which contains a typical
student error.
CONCLUSION
Computer
spreadsheet software is a very useful tool for generating multiple-choice test
questions, correct and incorrect responses, and solutions for problems which
have numeric answers. Tests, problem solutions, and homework exercises are
generated rapidly and easily, providing considerable time savings for teachers.
REFERENCES
|
|
|
|
|
|
|
|