Izvorni znanstveni članak
Mean values in grading: assessment and evaluation by statistical analysis
Mirna Valdevit
; Zagreb, Hrvatska
Zrinka Jelaska
; Filozofski fakultet Sveučilišta u Zagrebu, Zagreb, Hrvatska
Sažetak
This article discusses grading using a
single number as part of assessing students' knowledge in education, especially
in Croatian language. The first part of the paper introduces the mean and
median which should replace the arithmetic mean which is broadly used,
regardless of the fact that Croatian grades at all levels of education (from
primary school to university) are ordinal variables, not numerical.\\
In the main part of the article mode and median are used, as well as
percentage, to evaluate the grading system at the University School of
Croatian Language and Culture (UNICLAC). The intensive language
program, running at the University of Zagreb since 1990 and having more
than 700 students so far, developed it's program on the basis of
research conducted by some of its teachers. Since the summer of 1994 the
School reported grades within Croatian traditional 4--grade scale (5, 4,
3, 2) and 12--letter grade scale (A+ to D-). Data from 513 grades
was collected (1995--2008). Earlier grades were excluded as winter,
summer and autumn schools had different language level division
(particularly because former advanced level mostly covered students that
would later belong to intermediate level). Also, since 1995 each of the
three levels was divided into three sub-levels (beginning P1, P2, P3,
intermediate S1, S2, S3, advanced N1, N2, N3).\\
Three of the hypotheses were confirmed by statistical analysis, two were
partially confirmed. Croatian traditional grades were not easily
transferable to \textsc{ects} grading scale (out of 92\% who passed
Croatian language exam, 49.67\% got the highest grade, and only 2.38 the
lowest passing grade). When compared to the ECTS grading scale,
the letter grade showed different distribution: ECTS A was
comparable to UNICLAC A+ (10% to 15.8%), B to A (25% vs.
22.5%), C to A- & B+ (30% vs. 25%), D to B, B- & C+ (25%
vs. 26.8%), whilst E was comparable to C, C-, C+ & D (10% vs.
10.2%). Hence, there was some inflation of A+ grades (due to the
teachers' fascination with oral exams of some P1a students in years 2000
to 2008) as letter grades show intercultural system (teachers had often
been translating numbers into letters and modifying them with + or -
when appropriate). The grades were not quite evenly distributed through
different levels; there was a slight tendency to grade P3 and S2
students higher.
Ključne riječi
statistics; mean variables; grading; language assessment
Hrčak ID:
34722
URI
Datum izdavanja:
30.12.2008.
Posjeta: 19.828 *