Probability distribution of marks should not be normal.

What type of variable is the mark, discrete or continuous?
Marks is a discrete random variable that has a finite number of values or a countable number of values.

A continuous random variable has infinitely many values, and those values can be associated with measurements on a continuous scale in such a way that there are no gaps or interruptions.

Requirements for a Probability Distribution
1. ΣP(x) = 1 where x assumes all possible values of marks
2. 0 ≤ P(x) ≤ 1 for every individual value of x

For example, 2000 students gave exams with full marks of 10, the probability distribution of marks to have a normal like curve will have following frequency distribution given in the table.


Marks x Frequency f Probability P(X=x)
0 4 0.002
1 23 0.0115
2 99 0.0495
3 227 0.1135
4 399 0.1995
5 497 0.2485
6 390 0.195
7 251 0.1255
8 84 0.042
9 22 0.011
10 4 0.002




import matplotlib.pyplot as plt
import random
import numpy as np
from collections import Counter, OrderedDict
fig, ax = plt.subplots(1, 1)

od = OrderedDict([(0.0, 4), (1.0, 23), (2.0, 99), (3.0, 227), (4.0, 399),
 (5.0, 497), (6.0, 390), (7.0, 251), (8.0, 84), (9.0, 22), (10.0, 4)])
print(od)
val = sum(od.values())
probability = []
for prob in od.values():
    probability.append(prob/val)
print(probability)
print(sum(probability))
ax.plot(list(od.keys()), probability, 'bo', ms=8, 
label='probability distribution')
ax.vlines(list(od.keys()), 0, probability , colors='b', lw=5, alpha=0.5)

plt.show()

But frequency distribution like this is very difficult to achieve and based on many different factors, such as question difficulty, learning levels of students.

Do we even require such a curve?
About 37% of students will score below marks 5. So the performance of many students is too low.
Only about 5% will score 8 and above. The goal for score 8 and above becomes too unrealistic.
So we don't want a bell curve in education. If we are getting a bell curve, then our education system is very wrongly designed and implemented.


What type of graph should our education system have?

This is perhaps a better distribution, which our education system should have, and should achieve where about 92% of students score 7 and above. The goal should be to make every student score 7 or more, and the difficulty of questions should be realistic to achieve.



Marks x Frequency f Probability P(X=x)
0 0 0
1 15 0.007503751875937969
2 23 0.01150575287643822
3 18 0.009004502251125562
4 16 0.0080040020010005
5 19 0.009504752376188095
6 27 0.013506753376688344
7 575 0.2876438219109555
8 621 0.3106553276638319
9 559 0.27963981990995496
10 126 0.06303151575787894


import matplotlib.pyplot as plt
import random
import numpy as np
from collections import Counter, OrderedDict
fig, ax = plt.subplots(1, 1)
od = OrderedDict([(1, 15), (2, 23), (3, 18), (4, 16), (5, 19),
 (6, 27), (7, 575), (8, 621), (9, 559), (10, 126)])
print(od)
val = sum(od.values())
probability = []
for prob in od.values():
    probability.append(prob/val)
print(probability)
print(sum(probability))
ax.plot(list(od.keys()), probability, 'bo', ms=8, 
label='probability distribution')
ax.vlines(list(od.keys()), 0, probability , colors='b', lw=5, alpha=0.5)

plt.show()


An ideal marks distribution with total of 100 marks.
Post a Comment
My photo
Introvert, open minded, always learning, scientific, critical thinker, logical

Interests are in programming, chemistry, statistics, molecular biology, evolution, psychology, ecology and politics

Youtube Channel:

Popular Posts

Personalized learning goals: New information is built over reusing of previous information

All books that don't meet the learning criteria must be taken off

CBSE and other boards, All about rote memorization of abstract information

Producing a better secure electoral voting system

Why we face difficulty to remember non-concrete information?

What are the problems with grades?

Country needs an independent news analysis institution to provide feedback to media

Going away with money and making a currency that counts to unity always.

Internet: Centralization vs Decentralization, but we need both and going open source