Well, if there were one thing to be agreed over by all COEPians,

it will be : The GPA system is both interesting and intriguing, leaving everyone sans any clues even after they finish off the 8 sems.

To understand the system and strategise in accordance with the GPA system is something everyone wants to do but miserably fails at.

This is an account from a US insti educationist, analysing how effective the GPA system is?

Distorting the Record

Consider students X, Y, and Z. Suppose X gets an “A-” in

course #1 with 91.0% and 3 x 4 = 12 grade points. Student Y, taking

the same course, scores 89.0% for a B+, with grade points 3 x 3 = 9,

and Z scores 81.0% for a B, also 9 grade points. Comparing X and Y, a

2% difference in original course performance has produced a 25%

difference in grade points! In other words, the difference between

students has been grossly magnified, with a difference magnification

factor greater than 10X! Yet, comparing students Y and Z, an 8%

difference in score has produced absolutely no difference in grade

points. Is this logical or fair? Absolutely not, because the ability

demonstrated by student Y is much closer to X than to Z.

Given the fact that my college grading system includes one decimal

place in the percentage scores (table, above), how close could

students X and Y be, and still receive different letter grades? My

biology course has 1000 total possible points, 600 from lecture

exams, and 400 from laboratory quizzes and reports. Suppose X gets

900 points out of 1000, and Y gets 899. According to my current

college grading system, student X receives A-, and student Y a B+.

The difference in their performances, 1 point out of 1000 or 0.1%, is

utterly trivial, the equivalent of one incorrect guess on a single

true-or-false question. Yet, a grading system using quality points

records the difference as 25%, with a difference magnification factor

of 250X. How absurd!

Consequences

What effect does such a system have on people? First, students worry

excessively about every point lost for fear of a borderline near-miss

at semester end. Second, professors are disturbed by excessive

student worry, challenged by complaints about points lost, and nagged

by conscience when assigning borderline final letter grades. The

syndrome can affect final letter grade distribution in a very

negative way, because many professors plot final numerical

percentages of all students graphically, then look for natural

“breaks” in the curve so that borderline grades are minimized. Since

a natural break might occur at 86% rather than 90%, the result is

grade inflation. Third, recognizing that the system leaves something

to be desired, institutions periodically spend inordinate amounts of

time attempting to modify the system to make it more fair. Hence, the

most recent vote at my college is to count the + and – in the GPA

numerical calculation in the future. What is the significance of

adding + and – to letter grades and counting them numerically? This

is akin to improving N, S, E, and W compass points: NE is more

accurate than just E, and NNE more accurate than NE. So B+ is more

accurate than B, and, presumably, B++ even more accurate. By doing

this, what we really hope to achieve is letter grades that reflect

the original numerical percentage evaluations more accurately.

Likewise, NNE really represents only a 1/16th wedge in a 360 degree

directional circle, whereas N represents a far less precise 1/4 wedge.

Is There a Better Way?

Given students X, Y, and Z with 91%, 89%, and 81% respectively, in

the same course, what letter grades are really appropriate? None. No

letter grade will be as fair and accurate as the original percentage

grade. By definition, the percentage grade has at least 100 units of

assessment, and it typically has 1000 units by including one decimal

place, as in my current system. Thus, the ultimate absurdity: the

deserved performance assessments are, of course, simply the original

91%, 89%, and 81% scores.

What happens, then, if the professor never uses numerical

percentages, but rather uses letter grades directly on exams and

papers? These letter grades should be converted into percentages,

using current percentage-letter grade conversions (table, above) in

reverse. Should students receive QPs for these percentages, and

would a GPA be calculated? Absolutely not! Fairness demands that

there be no QPs and no GPA to magnify student differences. Since we

must still take into account the number of credits per course and all

courses taken, performance overview should be determined as

percentage per credit, or PPC. For example, suppose students X and Y

take the same five courses: three 3 credits, one 2 credits, and one 4

credits. Suppose also that student X gets 91% in every course for a

91 PPC, and Y gets 91% in the three 3 credit courses and 89% in the

other two for a 90.2 PPC ((9 x 91%) + (6 x 89%)) ÷ 15 = 90.2%. Both

students would thus be in the “A” category using a 90 PPC standard,

whereas the traditional GPAs of 4.0 for X and 3.6 for Y are

misleading. The situation can be even more ridiculous: if Y received

98% in each 3 credit course, and 89% in the other two, Y’s PPC would

be 94.4, clearly showing superiority to X. Yet the GPAs remain as

before: 4.0 for X and 3.6 for Y.

Now, let’s reconsider the F grade, with zero QPs for anything below

60%. Suppose student X takes five 3 credit courses, and gets 79%,

79%, 69%, 69%, and 59%. The PPC will be 71 (classical C category),

but the GPA is 1.2 (much closer to D). If Y gets the same grades in

the first four, but 30% in the last, the GPA remains 1.2 but the PPC

is 65.2, correctly reflecting the difference in level of F

performance. One more extreme example to make the point: suppose

student Z takes five 3 credit courses, receives 99% in four, but has

one very bad day and gets 59% in the fifth. Clearly, this is a

superior student (PPC = 91), but the GPA is a so-so 3.2.

Conclusion

Some may argue that the GPA is nevertheless acceptable because

borderline grades may not make much difference in the long run. But

why not just get it right? Why begin the grading process with an

accurate numerical evaluation, convert it to a less accurate letter

grade, and back again to a still less accurate number? With its

potential for producing distortion and unnecessary agonizing, the GPA

should be discarded and the PPC, or something better, should take its

place.

Professor William D. Cohen

Department of Biological Sciences