Book/Testing Costs
< Book
The preparation of top-notch test questions, and the execution and grading of tests are professionally demanding, laborious and expensive items. The cost of a single test can be expected to drop as the number of students being tested increases (economies of scale), because the fixed costs are spread over multiple units.
For small numbers of examinees, it is definitely less laborious to take oral exams than to prepare, administer and grade tests. There are usually good reasons for deciding to test a small number of students using tests. Assessment of students by means of tests may be the method of choice if we need the results to be demonstrably objective and reproducible, e.g., when there is a risk of appeals from the test takers. From an economic point of view, testing pays off with a large number of test takers, as high acquisition costs will be balanced by low operating costs.
- Item cost
The creation of high-quality test questions requires a team of experts in the relevant field, who are also trained in the methodology of test creation. Additional expenses come from reviewing questions and pilot testing with a large enough group of students.
The cost of preparing calibrated items for critical tests is estimated at $1,000 per question and generally speaking, the cost per item does not fall below $300[1].
Rudner[2], for example, calculated the total cost of developing one quality item for an adaptive acceptance test. He showed that a good quality calibrated item costs (in the US) USD 1,500–2000. If we compare this figure with Breithaupt's estimate of the required number of items in an item bank for common adaptive testing (about 2,000 items in the bank), we arrive at an astronomical sum of 3-4 million USD for the content of the item bank[3][4].
Item costs can be reduced if we can reuse already finished items. These may be items that we have prepared and calibrated in previous rounds of testing, as long as we are sure they have not been exposed by previous use.
At the 1st Faculty of Medicine of Charles University, we calculated the costs of a new item in the academic year 2020/21. 230 items were prepared, at a final price of CZK 1,500 per item. The cost of the work of authors and reviewers and, to a lesser extent, the costs of operating and acquiring an item bank are reflected in the price. It is obvious that compared to foreign countries, we have the advantage of qualified labor at lower prices. |
- Cost Per Examinee
It is interesting to look at the cost of testing per student tested. A large dispersion of data is evident here, due to different conditions and requirements.
In a 1996 report, the Center for Research on Educational Standards and Student Testing (CRESST) calculated the cost of test-based assessment with teacher salaries factored in. His estimates ranged from USD 848 to USD 1,792 per student[5].
How significant economies of scale can be is shown by the example of the well-known ACT and SAT exams, which are part of the university admissions process in the United States. In 2020, 1.65 and 1.1 million examinees took these exams. Prices for these tests, which cover all costs of test development, scoring, and administration, range from USD 20 to USD 70.
Although cost-benefit estimates are usually context- and scale-dependent, it is clear that the benefits of computer-assisted testing substantially outweigh its costs[6].
Again, we can compare these costs with the costs of the admissions process for the First Faculty of Economics of the Charles University in the academic year 2020/21. If we limit ourselves to master's fields, for which there were approximately 4,000 applicants, we get an approximate cost of CZK 250 per examinee. Of this, the so-called "test day costs" make up the largest portion of the total costs, representing more than a third of the total amount. Another third or so goes to printing costs, and the rest is divided into item creation and item bank costs. Here, too, it is evident that we are at a fraction (about a third) of the compared costs of SAT and ACT exams, despite the disparity in the number of test takers. We attribute the lower cost ratio than in the case of question creation, where the ratio was more than 1:20, to the difference in the number of test takers and the incompressibility of printing costs, which are not as geographically sensitive as labor costs. |
Odkazy
Reference
- ↑ Chybná citace: Chyba v tagu
<ref>
; citaci označenéDowning
není určen žádný text - ↑ RUDNER, Lawrence M.. Implementing the graduate management admission test computerised adaptive test [online] . In van der Linden, Wim J.; Glas, Cees A.W. Elements of Adaptive Testing. 1. vydání. New York : Springer, 2010. s. 151-165. Dostupné také z <https://link.springer.com/chapter/10.1007/978-0-387-85461-8_8>. ISBN (Print) 978-0-387-85459-5, (Online) 978-0-387-85461-8
- ↑ BREITHAUPT, Krista, Adelaide A ARIEL a Donovan R HARE. Assembling an inventory of multistage adaptive testing systems [online] . In van der Linden, Wim J.; Glas, Cees A.W. Elements of Adaptive Testing. 1. vydání. New York : Springer, 2010. s. 247-266. Dostupné také z <http://link.springer.com/chapter/10.1007%2F978-0-387-85461-8_13>. DOI:10.1007/978-0-387-85461-8_13. ISBN (Print) 978-0-387-85459-5, (Online) 978-0-387-85461-8
- ↑ GIERL, Mark J, Hollis LAI a Simon R TURNER. Using automatic item generation to create multiple-choice test items. Medical Education [online]. 2012, roč. -, vol. 46, no. 8, s. 757-765, dostupné také z <http://onlinelibrary.wiley.com/doi/10.1111/j.1365-2923.2012.04289.x/full>. sv. DOI:. ISSN 1365-2923. PMID: 22803753.DOI: 10.1111/j.1365-2923.2012.04289.x.
- ↑ PICUS, Lawrence O., Alisha TRALLI a Suzanne TACHENY. Estimating the Costs of Student Assessment in North Carolina and Kentucky: A State-Level Analysis. In: CSE Report 408: National Center for Research on Evaluation, Standards, and Student Testing (CRESST). [online]. Los Angeles: University of California, 1996 [cit. 2021-11-17]. Dostupné z: https://cresst.org/publications/cresst-publication-2780/
- ↑ PHELPS, Richard. Estimating the Costs and Benefi ts of Educational Testing Programs. In: The Education Consumers Consultants Network: Issues in Public Education: Research and Analysis from the Education Consumers Foundation [online]. Arlington: Education Consumers Foundation, 2002 [cit. 2021-11-18]. Dostupné z: http://education-consumers.org/research/briefs_0202.htm