8 The Normal Curve and its

Applications

WHAT IS A NORMAL CURVE?

The literal meaning of the term normal is average. We make use of this
term while computing the average in data related to education,
psychology or sociology. In these areas, those who are able to reach a
particular fixed level in qualification or characteristics are termed as
normal, while there who are above or below this point are abnormal.
Nature has been kind enough to distribute quite equally most of the
things and attributes like wealth, beauty, intelligence, height, weigat
and the like. As a result, a majority among us possess average beauty,
wealth, intelligence, height and weight. There are quite a few persons
who deviate noticeably from average, either be above or below it.This
is equally true for data in terms of achievement scores, intelligence
sCores, rating scores, etc. collected through tests, surveys and
experiments performed in education, psychology and sociology on a
randomly selected sample or population.
If we plot such a distribution of data on graph paper (Figure 8.1),

Frequencies

X
’ Scores

Figure 8.1 Normal curve.

111
 ducation
112 Statistics in Bychology and
we would get an interesting typical curve often resembling a verical
CrOSs-section of a bell. This bell-shaped curve is called a normal curve.
certain coin or a dice-throwing
The data fromn a
involving a chance success or probability, it plotted on
the normal
experiment
a graph nen
trequency curve which closely resembles curve.
because
give a of this reason and because of its origin from a gaine of chance
probability
called the normalcurve
that the normal curve is often
derived this
Curve.
Laplace and Gauss
They worked on
(1777-1855)
experimental errors in Physics and
distributed
iAstronomy
ndependently.
normally. This is the reason why the
and
found the errors to be characteristics is also namedas
Curve of error or
curve representing the
simplynormal
the curve of error where 'error' is usednormal
in the
Because of its
sense of a deviation from the true value.
curve is also named as Gaussian curve in
discovery
by
Laplace and Gauss, the
honour of Gauss.
the law which states that th.
The normal curve takes into account
average value in a series, the lese
greater a deviation from the mean or describing manu
used for
frequently it occurs. This is satisfactorily education, psychology and
distributions which arise in the fields of
sociology. distribution to be
However; it is not at all essential for a normal
described by an exactly perfect bell shaped curve as shown in
Figure 8.1. Such a perfect symmetrical curve rarely exists in our actual
as we usually cannot measure an entire population. Instead.
dealings
we work on representative samples of the population. Therefore,in
actual practice, the slightly deviated or distorted bell-shaped curve is
also accepted as the normal curve on the assumption of normal
distribution of the characteristics measured in the entire population.
From the above account, it should not be assumed that the
distributions of the data in all cases will always lead to normal or
approx1mately normal curves. In cases where the scores of individuals
In the group seriously deviate from the average, the curves repre
senting these distributions also deviate from the shape of a normal
curve. This deviation or divergence from normality tends to valy
two ways:

In Terms of Skewness
Skewness refers to the lack of symmetry. Anormal curve is a perfectly
symmetrical curve. In case the curve is folded along the vertical middle
line (perpendicular drawn from the highest point of the curve to its
base line) the two sides of the base line will overlap. Also, for this
curve, mean, median and mode are the same. In many distributions
which deviate from the normal, the values of mean, median and mode
ale ditleet and thee is nw mmetrv bemcen the ught and the let
hales ot the une Surh distTtbutons r sd to be skewed. being
ita lned moe towards the let or the nght to the centre of the cuve
as shown tn Figues NN and s 3 tor onmparison, a nomal (snmetr
cal) cue is also pvded in igue St

Mean Median Mode

Figure 8.2 Negative skewness (the uve inclines more to the left).

Mode Median Mean

the right).
Figure 8.3 Positive skewness (the uve inclines more to
Highest point of the cuve

Vertical middle tine

Base line

Mean Median Mode

Figure 8.4 Normal (svmmerical) curve.

given types of skewed

figures, there are two negatively
As illustrated in the wvhen
distributions. The distributions are said to be skewed the
individuals in a group with their scores higher than
there are many Similarly, the distributions are saYd to be
average score of the group.
positively when there are nore individuals in a group who scoe
skewed
their group.
less than the average score for distribution mav be computed by the
Skewness in a given
following formula:
114 Statistics in Psychology and Education

3(Mean- Median)
Skewness = Standard deviation

3(M - M)
S;=
SD
the value of skewness may
In case when the percentiles are known, be
computed from the following formula:
S_ = Pao t1o- Ps0

In Terms of Kurtosis
the
When there are very few individuals whose scores are near to
area of the
average score for their group (too few cases in the central
'flattened'
curve) the curve representing such a distribution becomes
in the middle. On the other hand, when there are too many cases in
the central area, the distribution curve becomes too peaked' in
comparison to normal. Both these characteristics of being flat or
peaked, are used to describe the term kurtosis.
Kurtosis is usually of three types:
Platykurtic. Afrequency distribution is said to be platykurtic, when
it is flatter than the normal.

Figure 8.5 Platykurtic.

Leptokurtic. Afrequency distribution is said to be laptokurtic, when
it is more peaked than the normal.

Figure 8.6 Leptokurtic.

 The Nomal Cure and its
Applications 115
Mesokurtic. A frequency distribution is said to be
it almost mesokurtic, when
resembles the normal curve (neither too flattened
peaked). nor too

Figure 8.7 Mesokurtic.

The value of kurtosis for a given curve may be
the following formula: computed through
Kurtosis Quartile deviation
Or
90th percentile - 10th percentile

K, =
Pgo - Po
In the case of a normal curve, this value is, equal to 0.263. Conse
quently, if the value of kurtosis is greater than 0.263, the distribution
is said to be platykurtic; if less than 0.263, the distribution is
leptokurtic.

CHARACTERISTICS AND PROPERTIES OF A NORMAL CURVE

For this curve, mean, median and mode are the same.
2. The curve is perfectly symmetrical. In other words, it is not
skewed. The value of the measure of skewness computed for
this curve is zero.
S The normal curve serves as a model for describing the
peakedness or flatness of a curve through the measure of
kurtosis. For the normal curve, the value of kurtosis is 0.263.
Iffor adistribution the value of kurtosis is more than 0.263,
the distribution is said to be more flat at the top than the
normal curve. But in case the value of kurtosis is less than
0.263, the distribution is said to be more peaked than the
normal.
4. The curve is asymptotic. It approaches but never touches the
base line at the extremes because of the possibility of locating
in the population, a case which scores still higher than our
highest score or lower than our lowest score. Therefore,
theoretically, it extends from minus infinity to plus infinity.
 Education
116 Statistics in Psychology and
8. As the curve does not touch the base line, the nean 1s used
working with the normal curve.
G The curve has its for
maximum height or ordinate at the starung
as the starting point
point, i.e. the mean of the distribution. In a unit nOrtnal

this ordinate is equal to .3989.

curve, the value of from the point of departure (i.e. mean),
deviations
Z. To find thedeviation of the distribution () is used as a
unit of
standard
measurement. distance on the
A8. The curve
extends on both sides -30 left to
+3o distance on the right.of the curve occur at tl stand..
inflection
9. The points of and below the mean. Thus .the
deviation unit (tl) above concave in relation to
from convex to
curve changes
horizontal axis at these points. extending from -3o to +3ai.
the curve
10. The total area under 10,000 because of the greater ease in
taken arbitrarily to be found
fractional parts of the total area
the computation of the ordinates erected at various distances
for the mean and the
such fractional parts of
from the mean. The computation ofdistances from the mean
the total area for travelling desired o
Table B given in
may be conveniently made with the help of
the appendix of this textbook. 34.13% of the
1. We may find that 3413 cases out of 10,000 or
entire area of the curve lies between the mean and +io on
the base line of the normal curve. Similarly, another 34.139%
cases lie between the mean and -lo on the base line.
Consequently, 68.26% of the area of the curve falls within
the limits t1 Standard deviation (tlo) unit from the mean.
Going further it may be found out that 95.44% cases lie from
-2a to +2o and 99.74% cases lie from -3G to +3o.
Consequently only 26 cases in 10,000 (10,000 - 9974) should
be expected to lie beyond the range t3¡ in a large sample as
shown in Figure 8.8.
y

.3413 .3413
.02145 .02145
.0013 .1359
0 -30 .1359 .0013
-26 -1o M
Figure 8.8 Normal curve showing areas
+1o +26 +3o X

from the mean. at different distances

 The Normal Curve and ts Applications 117
in chs curve. che lmits of the distances +.96g include 95%
and the hmts t238c inchude 99% of the total area of the
curve. 3 and t% of the area, respectively falling beyond
chese lumats as shown in Figure 8.9.

.025

.005
+1.96

-1 +36 X
+1o +2o
-258 +2.58a

Figure Normal urre shOwing 3% and 1% cases lying beyond the

bmits -l.960 and +2.58o, respectively.

APPLICATIONS OF THE NORMAL CURVE

Normal cune has wide significance and applications in the field of

easurement concerning education, psychology and sociology. Some
of its man applications are discussed in the following sections:

Use as a Model
Normal curve represents a model disribution. It can be used as a
nodel to

cOmpare varnous distributions with it i.e. to say, whether the

distribution is normal or not and if not, in what way it
dIverzes from the normal:
compare two or more distributions in terms of overlapping:
and
3. evaluate students performance from their scores.

Computing Percentiles and Percentile Ranks

ormal probab1lity curve mav be conveniently used for computing
peTtenules and percentule ranks in a given normal distribution.
 118 Slatstics in whology and Educatron

Understanding and Applying the Concept of

Errors of Measurement
Standard
The normal curve as we have pointed out earlier. is also
the nornmal curve of error or simply the curve of errOr on
that it helps in understanding the concept of the nown
measurement. For example, if we compute mean for thestandard grondA
errors
thdiesestributn
of various samples taken from a single population, then.
will be found to be distributed normally around the
population. The sigma distance of a particular mean of the means
us determine the standard error of measurementsample mean mar
mean of heip
for the
sample. tha

Ability Grouping
A group of individuals may be conveniently grouped into certain
categories as A, B, C, D, E(Very good, good, average, poor, very poor
in termns of some trait (assumed to be
normally distributed), with the
help of a normal curve.

Transforming and Combining Qualitative Data

Under the assumption of normality of the distributed variable, the sets
of qualitative data such as ratings, letter grades and categorical ranks
on ascale may be conveniently transformed and combined to provide
an average rating for each individual.

Converting Raw Scores into Comparable Standard

Normalized Scores
Sometimes, we have records of an individual's performance on to
more different kinds of assessment tests and we wish to compare
SCore on one test with the SCore on the other. Unless the scales of these
tWo tests are the same, we cannot make a direct comparison. Withthe
help of a normal curve, we can convert the raw scores
different tests into standard normalized scores like belon5s
z
sCores)
(orSCore,
and T-scores. For sigma we
converting a
subtract the mean of the scores ofgiven raw SCOre into a z
Tespective
the distribution from thedistribution
raw scores and divide it by the standard
X- M deviation of the
indicates hon
i.e.,z= In this way, a standard z score clearly
the meal
many standard deviation units a raw Score is above or below
o fvaluable
and thus provides a standard scale for the purpose
 Determining the Relative Difficulty of Test Items
ormal urve prev des the nplest tatonal method d aling teu
tems ft diffx ulty and therefore. may e comveniently empleyed frr
determ1ing the rrlative diffsty of tet uestons problems and
ot her test ite ms

ILLUSTRATION OF THE APPLICATIONS OF THE

NORMAL CURVE

ow, let us study some of the applcatsons of the normal curve

diussed o far. with the help of a few examples Solutions to thee
ribted probiems require
Mores and
aknowedge of the converson of raw scores into
VICe versa. ad
2 the use of the normal curve Table B (Appendi) showing the
o
ca
fractional parts of the total area of the curve in relation
ssgma distances
variable Det
Converting Raw Scores into zScores and Vice Versa
categoraz
bined 1o p
scores can be expresed by
The relationsh1p between z cores and raw
the formula
X M
ndard
where
X=a gIven raw score
nanceon X scores
M= Mean of the disuribution of cores
a =SD of the distribution of X
tocompu
c scales
a
Gven Mean 495 and SD 143 lor a distribution.
r i s o nh Ezample 8.1:
es
bekng chanze the score of 80 into zor sigma score
Solution
X-M
n the
re
hediscrb
where
XB0, M 49.5, SD I43
indiae

Putting thee values in the fonula, we get

e l ot
n he