Attributes

Definition:
• An attribute is a quality or a characteristic which cannot be measured
but which can be marked by their presence or absence. For instance,
sex, literacy, honesty, nationality etc. are attributes. Given an attribute
the population can be divided into two classes; one possessing that
attribute and the other not possessing it. Such a classification into two
classes is called dichotomous.
Notations:
• Suppose the population is divided into two classes according to the
presence or absence of an attribute. The class possessing the attribute is
called a positive class and is denoted by capital letters A, B, C etc. The
class not possessing the attribute is called a negative class and is
denoted by small Greek letters α, β, γ etc. Thus, if ‘A' denotes the class of
'males' then ' α ' will denote the class of 'females';
Order of Classes and Class-Frequencies:
• A class representing one attribute is called a class of first order. Thus, A, B, C,
α, β, γ are classes of first order. A class representing two attribute is called a
class of second order. Thus, AB, AC, Aβ etc. are classes of second order.
Similarly, AβC, ABC, αβC are classes of third order. i. e. order denotes the
number of attributes in that class.

Class-Frequencies:
• The number of items belonging to a class is called the frequency of that class.
The class frequency is denoted by putting the letter (or letters) denoting the
class in a bracket. Thus, (A) stands for the number of items possessing the
attribute A ; (αB) stands for the number of items, not possessing A and
possessing B. The frequency of a positive class is called positive class
frequency e.g. (AB) and frequency of a negative class is called negative class
frequency e.g. (αβγ )
Ultimate Class frequencies:
The class-frequencies of highest order are called ultimate class-frequencies.
Thus, in the case of two attributes class-frequencies of order two are ultimate
class-frequencies. If A and B are attributes then (AB), (Aβ), (αB), (αβ) are
ultimate class-frequencies. If we are considering n attributes, the Ultimate class
frequencies will have n symbols.

Thus the total number of ultimate class frequencies in case of two attributes are
2 2 = 4 and for three attributes are 2 3 = 8
The total number of ultimate class frequencies in case of n attributes are 2n
The total number of positive class frequencies are 2n The total number of class
frequencies of all order are 3n.
Consistency of Data:
• If all the class frequencies are positive then the data is said to be
consistent. Or if all the ultimate class frequencies are non-negative then
the data is consistent.
Conditions of Consistency:
• For one attribute
• (i) (A) > 0, (ii) (α) > 0. But (A)+(α)= N,
• (iii) (A) ≤ N, (iv) (α) ≤ N
• For Two attributes
• (i) N = (A) +(α) = (B) +(β) (ii) (A) = (AB) + (Aβ), and
• (B) = (AB) + (αB) (α) = (αB) + (αβ), and (β) = (Aβ) + (αβ)
• Class symbols as an operator:
• If we look at the class symbols A, B as an operator, A stands for the ratio
of items possessing the attribute A. Then AN means multiplying N by this
ratio but this is the class frequency (A) of A. Hence, we have AN = (A).
Similarly, A(B) means multiplying (B) by the ratio A, but this will be the
number of members having both attributes AB i.e. (AB). Thus, we have,
A*(B)=(AB)=AB*N. Using class symbol as an operator we can obtain
various relations as follows :
• (AB) = N-(α)-(β)+(αβ)
• (αβ) = N-(A)-(B)+(AB)
• (AB) ≥ (A) + (B) - N iv) (αβ) ≥ (α) + (β) - N
• Thus the consistency conditions for two attributes are
(i) (AB) ≥ 0. (ii) (αβ) ≥ 0 (Aβ) ≥ 0. (iii) (αB) ≥ 0 (iv) (αβ) ≥ 0 (v) (A) ≥ (AB)
(vi) (B) ≥ (AB) (vii) (AB) ≥ (A) + (B) – N
• The consistency conditions for Three attributes are
(i) (ABC) ≥ 0. (ii) (AB) ≥ (ABC) (iii) (AC) ≥ (ABC) (iv) (BC) ≥ (ABC)
(v) (ABC) ≥ (AB) +(AC) - (B) (vi) (ABC) ≥ (AB) + +(AC) - (A)
(vii) (ABC) ≥ (AC) + (BC) – (C) (viii) (ABC) ≤ (AB) + (BC) + (AC) – (A) – (B) - (C) + N
(ix) (AB) + (AC) + (BC) ≥ (A) + (B) + (C) – N
(x) (AC) + (BC) - (AB) ≤ (C)
(xi) (AB) + (BC) - (AC) ≤ (B)
(xii) (AB) + (AC) - (BC) ≤ (A)
• Example 1: From the following data check whether the data are
consistent or not.
(A) = 120, (B) = 165, (AB) = 160, N = 400.

Solution :
(αB) = (B) – (AB) = 165-160 = 5,
(α) = N - (A) = 400-120 = 280
(αβ) = (α) – (αB) = 280 – 5 = 275
(Aβ) = (A) – (AB) = 120 – 160 = -40
since one of the ultimate class frequency is negative the given data is not
consistent.
• Association of Attributes:
• To study relationship if the characteristics cannot be measured i.e. to
study relationship between two attributes we use the technique called
association of attributes. If the attributes are not independent and they
are related with each other in some way then they are said to be
associated to one another.

• Types
• Positive
• Negative
• completely associated
• completely disassociated
• i) Positive Association: If ‘A’ occurs large number of times with ‘B’ than β
then A & B are said to be Positively Associated. i.e. (AB) > (A)(B)/ (N)
then A & B are Positively Associated.
• ii) Negative Association: If ‘A’ occurs small number of times with ‘B’ or α
occurs large number of times with ‘B’ then they are said to be negatively
Associated. i.e. (AB) < (A)(B)/ (N) then A & B are Negatively Associated.
i.e. if δ > 0 then A & B are Positively Associated and if δ < 0 then A & B
are Negatively Associated.
• iii) If A cannot occur without B or all A’s are B’s then (AB) = (A) then A &
B are completely associated.
• iv) If all A’s are β’s then (αB) = (α) then A & B are completely
disassociated
• Measures of Association There are two methods of measuring association:
• a) Yule's coefficient of Association,
• b) Coefficient of Colligation.
Yule's Coefficient of association :
This is the most commonly used method of studying association and it is
denoted by Q and is defined by
• It A and B are independent then (AB)(αβ) - (Aβ)(αB) = 0, hence Q = 0. ii)
If A and B are completely associated, then (AB) = (A) and (AB) = (B) i.e.
(Aβ) = 0 and (αB) = 0. Thus Q =1 iii) If A and B are completely dissociated,
then (AB) = 0 & (αβ) = 0 Thus Q = -1 i.e. Q lies between -1 to 1.

• Coefficient of Colligation : Another measure of association coefficient

i)(AB)(αβ) = (Aβ)(αB) Thus Y = 0

2𝑌 1+𝑄− 1−𝑄
Q= Y=
(1+𝑌2) 1+𝑄+ 1−𝑄
• Independence of Attributes: The two attributes are said to be
independent if one is not affected by the presence or absence of other.
If two attributes A and B are independent, we expect the proportion of
A’s amongst B’s is same the proportion of A’s amongst β’s.
• i. e. (AB)/(B) = (Aβ)/ (β) Similarly, (Aβ)/(A) = (αB)/ (α)
• Criterion of independence : If A and B are independent then by the
above definition of independence we get,
• (AB)/(B) = (Aβ)/ (β) and (Aβ)/(A) = (αB)/ (α) which gives
• (i) (Aβ)/(A) = (αβ)/ (α)
• (ii) (AB) = (A) *(B)/ (N)
• (iii) (AB) * (αβ) = (Aβ)*(αB)
• These are the three criterions of independence of attributes.
Symbols (AB)0 and δ :

• If A & B are independent then we have (AB) = (A) *(B)/ (N), we denote
(AB)0 = (A)(B)/ (N) and the difference between (AB) &(AB)0 is denoted
by δ,

• i.e. δ = (AB) - (AB)0 = (AB) - (A)(B)/ (N) If A & B are independent, then δ =
0, δ can also be represented as δ = (1/N) [(AB) (αβ) - (Aβ) – (αB)]