Unit I
Unit I
and Statistical
Inference
Introduction
P
andurang
Vasudeo Sukhatme
(1911–1997) was an award-winning
Indian statistician. He is known for his pioneering
work of applying random sampling methods in agricultural
statistics and in biometry, in the 1940s. He was also influential in
the establishment of the Indian Agricultural Statistics Research
Pandurang Vasudeo Sukhatme Institute. As a part of his work at the Food and Agriculture
(July 27, 1911 - January 28, 1997) Organization in Rome, he developed statistical models for assessing
the dimensions of hunger and future food supplies for the world.
He also developed methods for measuring the size and nature of the protein gap.
In any statistical investigation, the interest lies in the assessment of one or more characteristics
relating to the individuals belonging to a group. When all the individuals present in the study are
investigated, it is called complete enumeration, but in practice, it is very difficult to investigate all
the individuals present in the study. So the technique of sampling is done which states that a part
of the individuals are selected for the study and the assessment is made from the selected group of
individuals. For example
(i) A housewife tastes a spoonful whatever she cooks to check whether it tastes good or not.
(ii) A few drops of our blood are tested to check about the presence or absence of a disease.
(iii) A grain merchant takes out a handful of grains to get an idea about the quality of the
whole consignment.
These are typical examples where decision making is done on the basis of sample information.
So sampling is the process of choosing a representative sample from a given population.
XII Std - Business Maths & Stat EM Chapter 8.indd 175 2/29/2020 11:40:02 AM
8.1.1 Basic concepts of sampling Types of sampling
Population The technique or method of selecting
The group of individuals considered a sample is of fundamental importance in the
under study is called as population. The word theory of sampling and usually depends upon
population here refers not only to people but the nature of the data and the type of enquiry.
to all items that have been chosen for the study. The procedures of selecting a sample may be
Thus in statistics, population can be number of broadly classified as
bikes manufactured in a day or week or month, 1. Non-Random sampling or Non-
number of cars manufactured in a day or week probability sampling.
or month, number of fans, TVs, chalk pieces, 2. Random Sampling or Probability
people, students, girls, boys, any manufacturing sampling.
products, etc…
Note
Finite and infinite population:
When the number of observations/ Here we confine ourselves to Random
individuals/products is countable in a group, sampling or Probability sampling.
then it is a finite population. Example: weights
of students of class XII in a school. Random sampling or Probability sampling
When the number of observations/ Random sampling refers to selection
individuals/products is uncountable in a group, of samples from the population in a random
then it is an infinite population. Example: manner. A random sample is one where each
number of grains in a sack, number of germs in and every item in the population has an equal
the body of a sick patient. chance of being selected.
XII Std - Business Maths & Stat EM Chapter 8.indd 176 2/29/2020 11:40:02 AM
a possibility of selecting the same sample any sample can be done through “Random Number
number of times. So, simple random sampling Table”. The random number table has been so
without replacement is followed. constructed that each of the digits 0,1,2,…,9 will
Thus in simple random sampling from a appear approximately with the same frequency
population of N units, the probability of drawing and independently of each other.
1
any unit at the first draw is , the probability The various random number tables available are
N
a. L.H.C. Tippett random number series
of drawing any unit in the second draw from
1 b. Fisher and Yates random number series
among the available ( N − 1) units is ,
(N − 1) c. Kendall and Smith random number series
and so on. . Several methods have been adopted d.
Rand Corporation random number
for random selection of the samples from the series.
population. Of those, the following two methods
Tippett’s table of random numbers is most
are generally used and which are described
popularly used in practice. Given below the first
below.
forty sets from Tippett’s table as an illustration
(A) Lottery method of the general appearance of random numbers:
This is the most popular and simplest
method when the population is finite. In this 2952 6641 3992 9792 7969 5911 3170 5624
method, all the items of the population are
4167 9524 1545 1396 7203 5356 1300 2693
numbered on separate slips of paper of same
size, shape and colour. They are folded and 2670 7483 3408 2762 3563 1089 6913 7991
placed in a container and shuffled thoroughly.
0560 5246 1112 6107 6008 8125 4233 8776
Then the required numbers of slips are selected
for the desired sample size. The selection of 2754 9143 1405 9025 7002 6111 8816 6446
items thus depends on chance.
Suppose, if we want to select the required
For example, if we want to select 10 number of samples from a population of size
students, out of 100 students, then we must write
N(<99) then any two digit random number
the names/roll number of all the 100 students
can be selected from ( 00 to 99) from the above
on slips of the same size and mix them, then we
random number table. Similarly if N(< 999) or
make a blindfold selection of 10 students. This
(< 9999), then any three digit random number
method is called unrestricted random sampling,
because units are selected from the population or four digit random number can be selected
without any restriction. This method is mostly from (000 to 999) or (0000 to 9999).
used in lottery draws. If the population or The procedure of selecting the random samples
universe is infinite, this method is inapplicable. consists of following steps:
(B) Table of Random number 1. Identify the N units in the population
When the population size is large, it is with the numbers from 1 to N.
difficult to number all the items on separate
2. Select at random, any page from the
slips of paper of same size, shape and colour.
‘Random Number Table’.
The alternative method is that of using the table
of random numbers. The most practical, easy 3. Select the required number of samples
and inexpensive method of selecting a random from any row or column or diagonal.
XII Std - Business Maths & Stat EM Chapter 8.indd 177 2/29/2020 11:40:02 AM
One may question, as to how it is ensured
2952 6641 3992 9792 7969 5911 3170 5624
that these digits are random. It may be pointed
4167 9524 1545 1396 7203 5356 1300 2693
out that the digits in the table were chosen
horizontally but the real guarantee of their 2670 7483 3408 2762 3563 1089 6913 7991
randomness lies in practical tests. Tippett’s 0560 5246 1112 6107 6008 8125 4233 8776
numbers have been subjected to numerous 2754 9143 1405 9025 7002 6111 8816 6446
tests and used in many investigations and their
randomness has been very well established for If the population size is 100 and suppose
all practical purposes. we want to select 10 items out of 100. All
itemsfrom 1 to 100 should be numbered as 001
An example to illustrate how Tippett’s
to 100. Now, we may now select 10 numbers
table of random numbers may be used is given
from the random number table. The procedure
below. Suppose we have to select 20 items
will be different, as Tippett’s random numbers
out of 6,000. The procedure is to number all
are available only in four digits. Thus, we can
the 6,000 items from 1 to 6,000. A page from
select the first two digits from the four digit
Tippett’s table may be selected and the first
random sample number. Making use of the
twenty numbers ranging from 1 to 6,000 are
portion of the random number table given, the
noted down. If the numbers are above 6000,
required random samples are shaded in RED
choose the next number ranging from 1 to 6000.
colour.Here, we consider row wise selection of
Items bearing those numbers will be selected
random numbers.
as samples from the population. Making use
of the portion of the random number table 2952 6641 3992 9792 7969 5911 3170 5624
given,the requiredrandom samples are shaded.
4167 9524 1545 1396 7203 5356 1300 2693
Here, we consider row wise selection of random
numbers. 2670 7483 3408 2762 3563 1089 6913 7991
0560 5246 1112 6107 6008 8125 4233 8776
2952 6641 3992 9792 7969 5911 3170 5624
2754 9143 1405 9025 7002 6111 8816 6446
4167 9524 1545 1396 7203 5356 1300 2693
2670 7483 3408 2762 3563 1089 6913 7991 Merits and demerits of Simple Random
0560 5246 1112 6107 6008 8125 4233 8776 Sampling:
2754 9143 1405 9025 7002 6111 8816 6446 Merits
1. Personal bias is completely eliminated.
If the population size is 1,000 and
2. This method is economical as it saves
suppose we want to select 15 items out of 1,000.
time, money and labour.
All itemsfrom 1 to 1000 should be numbered
3. The method requires minimum
as 0001 to 1000.Now, we may now select 15
knowledge about the population in
numbers from the random number table.The
advance.
procedure will be different, as Tippett’s random
numbers are available only in four digits. Thus, Demerits
we can select the first three digits from the four 1. This requires a complete list of the
digit random sample number.Making use of the population but such up-to-date lists are
portion of the random number table given, the not available in many enquiries.
required random samples are shaded in RED 2. If the size of the sample is small, then it will
colour. Here, we consider row wise selection of not be a representative of the population.
random numbers.
XII Std - Business Maths & Stat EM Chapter 8.indd 178 2/29/2020 11:40:03 AM
Example 8.1
Using the Kendall-Babington Smith - Random number table,Draw5 random samples.
23 15 75 48 59 01 83 72 59 93 76 24 97 08 86 95 23 03 67 44
05 54 55 50 43 10 53 74 35 08 90 61 18 37 44 10 96 22 13 43
14 87 16 03 50 32 40 43 62 23 50 05 10 03 22 11 54 36 08 34
38 97 67 49 51 94 05 17 58 53 78 80 59 01 94 32 42 87 16 95
97 31 26 17 18 99 75 53 08 70 94 25 12 58 41 54 88 21 05 13
Solution:
23 15 75 48 59 01 83 72 59 93 76 24 97 08 86 95 23 03 67 44
05 54 55 50 43 10 53 74 35 08 90 61 18 37 44 10 96 22 13 43
14 87 16 03 50 32 40 43 62 23 50 05 10 03 22 11 54 36 08 34
38 97 67 49 51 94 05 17 58 53 78 80 59 01 94 32 42 87 16 95
97 31 26 17 18 99 75 53 08 70 94 25 12 58 41 54 88 21 05 13
There many ways to select 5 random samples from the given Kendall-Babington
Smith - Random number table. Assume that at random we select 3rd column 1st value. This
location gives the digit to be 75. So the first sample will be 75, and then the next choices
can be in the same 3rd column which follows as 55, 16, 67 and 26. Therefore 75,55,16,67
and 26 will be used as random samples. The various shaded numbers can be taken as
5 random sample numbers. Apart from this, one can select any 5 random sample numbers as they like.
Example 8.2 2952 6641 3992 9792 7969 5911 3170 5624
Using the following Tippett’s random 4167 9524 1545 1396 7203 5356 1300 2693
number table, 2670 7483 3408 2762 3563 1089 6913 7991
0560 5246 1112 6107 6008 8125 4233 8776
2952 6641 3992 9792 7969 5911 3170 5624
2754 9143 1405 9025 7002 6111 8816 6446
4167 9524 1545 1396 7203 5356 1300 2693
2670 7483 3408 2762 3563 1089 6913 7991 Example 8.3
0560 5246 1112 6107 6008 8125 4233 8776 Using the following random number table,
2754 9143 1405 9025 7002 6111 8816 6446 Tippet’s random number table
Draw a sample of 15 houses from Cauvery 2952 6641 3992 9792 7969 5911 3170 5624
Street which has 83 houses in total. 4167 9524 1545 1396 7203 5356 1300 2693
Solution: 2670 7483 3408 2762 3563 1089 6913 7991
There many ways to select 15 random 0560 5246 1112 6107 6008 8125 4233 8776
2754 9143 1405 9025 7002 6111 8816 6446
samples from the given Tippet’s random
number table. Since the population size is 83 Draw a sample of 10 children with
(two-digit number). Here the door numbers theirheight from the population of 8,585
are assigned from 1 to 83. Assume that at children as classified hereunder.
random we first choose 2nd column. So the first Height (cm) 105 107 109 111 113 115 117 119 121 123 125
sample is 66 and other 14 samples are 74, 52, 39, Number of 2 4 14 41 83 169 394 669 990 1223 1329
children
15, 34, 11, 14, 13, 27, 61, 79, 72, 35, and 60. If the
Height(cm) 127 129 131 133 135 137 139 141 143 145
numbers are above 83, choose the next number
Number of 1230 1063 646 392 202
ranging from 1 to 83. 79 32 16 5 2
children
XII Std - Business Maths & Stat EM Chapter 8.indd 179 2/29/2020 11:40:03 AM
Solution: 141 16 8578
The first thing is to number the 143 5 8583
population(8585 children). The numbering 145 2 8585
has already been provided by the frequency
Total 8585
table. There are 2 children with height of 105
cm, therefore we assign number 1 and 2 to the Now we take 10 samples from the tables,
children those in the group 105 cm, number 3 since the population size is in 4 digits we can
to 6 is assigned to those in the group 107 cm use the given random number table. Select
and similarly all other children are assigned the the10 random numbers from 1 to 8585 in the
numbers. In the last group 145 cms there are two table, Here, we consider column wise selection
children with assigned number 8584 and 8585. of random numbers, starting from first column.
Example 8.4
Using the following random number table (Kendall-Babington Smith)
23 15 75 48 59 01 83 72 59 93 76 24 97 08 86 95 23 03 67 44
05 54 55 50 43 10 53 74 35 08 90 61 18 37 44 10 96 22 13 43
14 87 16 03 50 32 40 43 62 23 50 05 10 03 22 11 54 36 08 34
38 97 67 49 51 94 05 17 58 53 78 80 59 01 94 32 42 87 16 95
97 31 26 17 18 99 75 53 08 70 94 25 12 58 41 54 88 21 05 13
Draw a random sample of 10 four- figure numbers starting from 1550 to 8000.
XII Std - Business Maths & Stat EM Chapter 8.indd 180 2/29/2020 11:40:03 AM
Solution:
Here, we have to select 10 random numbers ranging from 1550 to 8000 but the given random number
table has only 2 digit numbers. To solve this, two - 2 digit numbers can be combined together to make a
four-figure number. Let us select the 5th and 6th column and combine them to form a random number, then
select the random number with given range. This gives 5 random numbers, similarly 8th and 9th is selected
and combined to form a random numbers, then select the random number with given range. This gives
5 random numbers, totally 10 four-figure numbers have been selected. The following table shows the
10 random numbers which are combined and selected.
23 15 75 48 59 01 83 72 59 93 76 24 97 08 86 95 23 03 67 44
05 54 55 50 43 10 53 74 35 08 90 61 18 37 44 10 96 22 13 43
14 87 16 03 50 32 40 43 62 23 50 05 10 03 22 11 54 36 08 34
38 97 67 49 51 94 05 17 58 53 78 80 59 01 94 32 42 87 16 95
97 31 26 17 18 99 75 53 08 70 94 25 12 58 41 54 88 21 05 13
XII Std - Business Maths & Stat EM Chapter 8.indd 181 2/29/2020 11:40:03 AM
Solution
Homogenous Percentage from Number of people
Stratum Random Samples
group population in each strata
Lower income 39 68
Category1 39 × 500 = 195 195 × = 26.5 26
class 100 500
Middle income 38 68
Category2 38 × 500 = 190 190 × = 26.5 26
class 100 500
Upper income 23 68
Category3
class
23 × 500 = 115 115 × = 15.6 16
100 500
Total 100 500 68
XII Std - Business Maths & Stat EM Chapter 8.indd 182 2/29/2020 11:40:04 AM
(iii) If the selected first random sample is 5, then (i) Sampling Errors
the rest of the samples are automatically selected (ii) Non-Sampling Errors
by incrementing the value of the sampling interval
(i) Sampling Errors
(k = 10) i.e., 5, 15 , 25, 35, 45, 55, 65, 75, 85, 95.
Errors, which arise in the normal course
Example: of investigation or enumeration on account of
Suppose we have to select 20 items out of chance, are called sampling errors. Sampling
6,000. The procedure is to number all the 6,000 errors are inherent in the method of sampling.
items from 1 to 6,000. The sampling interval is They may arise accidentally without any bias or
N 6000 prejudice. Sampling Errors arise primarily due
calculated as k= = = 300 . Thus
n 20 to the following reasons:
sampling interval = 300 denotes that for every (a) Faulty selection of the sample instead
300 samples one sample has to be selected. of correct sample by defective sampling
The first sample is selected from the first 300 technique.
(sampling interval) samples through random (b) The investigator substitutes a convenient
selection procedures. If the selected first sample if the original sample is not available
random sample is 50, then the rest of the samples while investigation.
are automatically selected by incrementing the (c) In area surveys, while dealing with border
value of the sampling interval (k=300) ie,50, lines it depends upon the investigator
350, 650, 950, 1250, 1550, 1850, 2150, 2450, whether to include them in the sample or
2750, 3050, 3350, 3650, 3950, 4250, 4550, 4850, not. This is known as Faulty demarcation of
5150, 5450, 5750. Items bearing those numbers sampling units.
will be selected as samples from the population.
(ii)Non-Sampling Errors
Merits The errors that arise due to human
1. This is simple and convenient method. factors which always vary from one investigator
2. Th
is method distributes the sample more to another in selecting, estimating or using
evenly over the entire listed population. measuring instruments( tape, scale)are called
Non-Sampling errors.It may arise in the
3. The time and work is reduced much.
following ways:
Demerits (a) Due to negligence and carelessness of the
1. Systematic samples are not random samples. part of either investigator or respondents.
2. If N is not a multiple of n, then the sampling (b) Due to lack of trained and qualified
interval (k) cannot be an integer, thus investigators.
sample selection becomes difficult. (c) Due to framing of a wrong questionnaire.
(d) Due to apply wrong statistical measure
8.1.2 Sampling and Non-Sampling Errors: (e) Due to incomplete investigation and sample
A sample is a part of the whole population. survey.
A sample drawn from the population depends
8.1.3 Sampling distribution
upon chance and as such all the characteristics
Definition 8.3
of the population may not be present in the
sample drawn from the same population. The Sampling distribution of a statistic is the
errors involved in the collection, processing frequency distribution which is formed with
various values of a statistic computed from
and analysis of the data may be broadly
different samples of the same size drawn from
classified into two categories namely,
the same population.
XII Std - Business Maths & Stat EM Chapter 8.indd 183 2/29/2020 11:40:04 AM
For instance if we draw a sample of size n 8.1.4 C
omputing standard error in simple
from a given finite population of size N, then cases
the total number of possible samples is Example 8.6
N N!
Cn = = k (say). For each of these A server channel monitored for an hour
n !(N − n)! was found to have an estimated mean of 20
k samples we can compute some statistic, transactions transmitted per minute. The
t = t (x1 , x2 , x3 ,...xn ), in particular the mean variance is known to be 4. Find the standard
x , the variance S2, etc., is given below: error.
Statistic
Sample Number Solution:
t x s2 Given s 2 = 4 which implies s = 2 ,
1 t1 x1 s12 n = 1 hour = 60 min, X = 20 /min
s 2
2 t2 Standard Error = = = 0.2582
x2 s2
2 n 60
3 t3 x3 s32 Example 8.7
Find the sample size for the given standard
. . . . deviation 10 and the standard error with respect
. . . . of sample mean is 3.
. . . .
Solution:
k tk xk sk2 Given s = 10 , S.E. X =3 We know that
The set of the values of the statistic so S.E = s
obtained, one for each sample, constitutes the n 10 10
sampling distribution of the statistic.
Therefore, 3= ⇒ n=
n 3
Taking Squaring on both sides we get
Standard Error
The standard deviation of the sampling
2
10 100
distribution of a statistic is known as its Standard n= = = 11.11 ≅ 11 ,
3 9
Error abbreviated as S.E. The Standard Errors
(S.E.) of some of the well-known statistics, for The required sample size is 11.
large samples, are given below, where n is the Example 8.8
sample size, s 2 is the population variance. A die is thrown 9000 times and a throw of
Sl. 3 or 4 is observed 3240 times. Find the standard
Statistic Standard Error
No. error of the proportion for an unbiased die .
1. –)
Sample mean (x s n
Solution :
Observed sample
2. PQ n If the occurrence of 3 or 4 on the die is
proportion (p)
called a success, then
Sample standard
3. s 2 2n Sample size = 9000; Number of Success =
deviation (s)
3240
4. Sample variance (s2) s2 2 n 3240
Sample proportion = p = = 0.36
5. Sample quartiles 1.36263 s n 9000
Population proportion (P) = Prob (getting
6. Sample median
1.25331 s n 3 or 4 when a die is thrown)
7. Sample correlation 1 1 2 1
coefficient (r) (1 − r 2 ) n = + = = = 0.3333
6 6 6 3
184 12th Std. Business Mathematics and Statistics
XII Std - Business Maths & Stat EM Chapter 8.indd 184 2/29/2020 11:40:08 AM
Thus P = 0.3333 and Exercise 8.1
Q = 1–P=1– 0.3333 = 0.6667.
1. What is population?
The S.E for sample proportion is given by
2. What is sample?
PQ (0.3333)(0.6667) 3. What is statistic?
S. E . = n = = 0.00496
9000 4. Define parameter.
Hence the standard error for sample 5. What is sampling distribution of a statistic?
proportion is S.E=0.00496.
6. What is standard error?
Example 8.9 7. Explain in detail about simple random
The standard deviation of a sample of size sampling with a suitable example.
50 is 6.3. Determine the standard error whose 8. Explain the stratified random sampling
population standard deviation is 6? with a suitable example.
Solution: 9. Explain in detail about systematic random
Sample size n = 50 sampling with example.
Sample S.D s = 6.3 10. Explain in detail about sampling error.
Population S.D s = 6 11. Explain in detail about non-sampling error.
12. State any two merits of simple random
The standard error for sample S.D is given
sampling.
by
σ2 6 6 13. State any three merits of stratified random
S. E . = = = = 0. 6 sampling.
2n 2(50) 100
14. State any two demerits of systematic random
Thus standard error for sampling.
sample S.D = 0.6. 15. State any two merits for systematic random
sampling.
Example 8.10
A sample of 100 students is chosen from a 16. Using the following Tippet’s random
large group of students. The average height of these number table
students is 162 cm and standard deviation (S.D) is 2952 6641 3992 9792 7969 5911 3170 5624
8 cm. Obtain the standard error for the average
4167 9524 1545 1396 7203 5356 1300 2693
height of large group of students of 160 cm?
2670 7483 3408 2762 3563 1089 6913 7991
Solution: 0560 5246 1112 6107 6008 8125 4233 8776
Given n = 100, x =162 cm, s = 8 cm is
2754 9143 1405 9025 7002 6111 8816 6446
known in this problem
˘ Draw a sample of 10 three digit numbers
since σ is unknown, so we consider σ = s which are even numbers.
and m= 160 cm
˘ 17. A wholesaler in apples claims that only 4%
σ s 8 of the apples supplied by him are defective.
S. E . = = = = 0. 8
n n 100 A random sample of 600 apples contained
Therefore the standard error for the 36 defective apples. Calculate the standard
average height of large group of students of 160 error concerning of good apples.
cm is 0.8. 18. A sample of 1000 students whose mean
weight is 119 lbs(pounds) from a school
Sampling Techniques and Statistical Inference 185
XII Std - Business Maths & Stat EM Chapter 8.indd 185 2/29/2020 11:40:08 AM
in Tamil Nadu State was taken and their Estimator:
average weight was found to be 120 lbs with
Definition 8.5
a standard deviation of 30 lbs. Calculate
Any sample statistic which is used to
standard error of mean.
estimate an unknown population parameter
19. A random sample of 60 observations was is called an estimator ie., an estimator is a
drawn from a large population and its sample statistic used to estimate a population
standard deviation was found to be 2.5. parameter.
Calculate the suitable standard error that Estimate:
this sample is taken from a population with
standard deviation 3? Definition 8.6
When we observe a specific numerical value
20. In a sample of 400 population from a village of our estimator, we call that value is an
230 are found to be eaters of vegetarian estimate. In other words, an estimate is a
items and the rest non-vegetarian items. specific observed value of a statistic.
Compute the standard error assuming that
both vegetarian and non-vegetarian foods Characteristic of a good estimator
are equally popular in that village? A good estimator must possess the
following characteristic:
Statistical Inference
(i) Unbiasedness (ii) Consistency (iii)
One of the main objectives of any statistical
Efficiency (iv) Sufficiency.
investigation is to draw inferences about a
population from the analysis of samples drawn (i)
Unbiasedness: An estimator
from that population. Statistical Inference Tn = T (x1 , x2 ,...., xn ) is said to be an
provides us how to estimate a value from the unbiased estimator of
sample and test that value for the population. γ (θ ) if E(Tn ) = γ (θ ) , for all θ ε Θ
This is done by the two important classifications (parameter space), (i.e) An estimator is
in statistical inference, said to be unbiased if its expected value is
equal to the population parameter.
(i) Estimation;
Example: E( x ) = m
(ii) Testing of Hypothesis
(ii)
Consistency: An estimator
8.2 Estimation: Tn = T (x1 , x2 ,...., xn ) is said to be consistent
It is possible to draw valid conclusion estimator of γ (θ ) , if Tn converges to
about the population parameters from sampling γ (θ ) in Probability, i.e.,
P
distribution. Estimation helps in estimating Tn → γ (θ ) as n → ∞ , for all θ εQ .
an unknown population parameter such as
fficiency: If T1 is the most efficient
(iii) E
population mean, standard deviation, etc., on
the basis of suitable statistic computed from the estimator with variance V1 and T2 is any
samples drawn from population. other estimator with variance V2 , then the
V1
efficiency E of T2 is defined as E =
Estimation: V2
Obviously, E cannot exceed unity.
Definition 8.4
The method of obtaining the most (iv) S ufficiency: If T = t ( x1 , x2 ,...., xn ) is an
likely value of the population parameter using estimator of a parameter q , based on a
statistic is called estimation. sample x1 , x2 ,...., xn of size n from the
XII Std - Business Maths & Stat EM Chapter 8.indd 186 2/29/2020 11:40:11 AM
population with density f ( x , q ) such that would be expected to lie is called an interval
the conditional distribution of x1 , x2 ,...., xn estimation.
given T , is independent of q , then T is For example,
sufficient estimator for q . If T is a good estimator of θ with standard
error s then, making use of general property of
8.2.1 Point and Interval Estimation:
the standard deviations, the uncertainty in T, as
To estimate an unknown parameter of the
an estimator of q, can be expressed by statements
population, concept of theory of estimation is
like “ We are about 95% certain that the unknown
used.There are two types of estimation namely,
q, will lie somewhere between T-2s and T+2s”,
1. Point estimation “we are almost sure that q will in the interval
2. Interval estimation ( T-3s and T+3s)” such intervals are called
1. Point Estimation confidence intervals and is explained below.
When a single value is used as an
Confidence interval
estimate, the estimate is called a point estimate
After obtaining the value of the statistic ‘t’
of the population parameter. In other words, an
(sample) from a given sample, Can we make some
estimate of a population parameter given by a
reasonable probability statements about the
single number is called as point estimation.
unknown population parameter ‘ q ’?. This
For example question is very well answered by the technique of
(i) 55 is the mean mark obtained by a sample Confidence Interval. Let us choose a small value of
of 5 students randomly drawn from a class a which is known as level of significance(1% or
of 100 students is considered to be the mean 5%) and determine two constants say, c1 and c2
marks of the entire class. This single value such that P (c1 < θ < c2 |t ) = 1 − α .
55is a point estimate.
The quantities c1 and c2, so determined
(ii) 5 0 kg is the average weight of a sample of
are known as the Confidence Limits and the
10 students randomly drawn from a class of
interval c1 , c2 within which the unknown
100 students is considered to be the average
value of the population parameter is expected
weight of the entire class. This single value
to lie is known as Confidence Interval. (1− a) is
50 is a point estimate.
called as confidence coefficient.
Note Confidence Interval for the population mean
The sample mean ( x ) is the sample statistic for Large Samples (when s is known)
used as an estimate of population mean (m) If we take repeated independent random
samples of size n from a population with an
Instead of considering, the estimated value unknown mean but known standard deviation,
of the population parameter to be a single value, then the probability that the true population
we might consider an interval for estimating the mean m will fall in the following interval is
value of the population parameter. This concept (1− a) i.e
is known as interval estimation and is explained σ σ
below. P = x − Zα ≤ µ ≤ x + Zα = (1− a)
n n
2 2
2. Interval Estimation So, the confidence interval for population
Generally, there are situations where point mean (m), when standard deviation (s ) is known
estimation is not desirable and we are interested σ
and is given by x ± Z α .
in finding limits within which the parameter 2 n
XII Std - Business Maths & Stat EM Chapter 8.indd 187 2/29/2020 11:40:13 AM
For the computation of confidence intervals and for testing of significance, the critical values
XII Std - Business Maths & Stat EM Chapter 8.indd 188 2/29/2020 11:40:17 AM
be 1350 hours with a standard deviation of 8.3.1 M
eaning : Null Hypothesis and
100 hours. Establish 90% confidence limits Alternative Hypothesis - Level of
within which the mean life time of light bulbs is Significants and Type of Errors
expected to lie. Statistical Hypothesis
Solution: Statistical hypothesis is some assumption
Given: n = 169, x = 1350 hours, s = 100 or statement, which may or may not be true,
hours, since the level of significance is (100- about a population.
90)% =10% thus a is 0.1, hence the significant There are two types of statistical hypothesis
value at 10% is Za = 1.645 (i) Null hypothesis (ii) Alternative hypothesis
2
s 100 Null Hypothesis
S.E. = = = 7.69
n 169 Definition 8.7
Hence 90% confidence limits for the According to Prof. R.A.Fisher, “Null
population mean are hypothesis is the hypothesis which is tested
x − Zα SE < µ < x + Zα SE for possible rejection under the assumption
2 2 that it is true”, and it is denoted by H0 .
1350 − (1.645 × 7.69) ≤ m ≤ 1350 + (1.645 x 7.69)
For example: If we want to find the population
1337.35 ≤ m ≤ 1362.65 mean has a specified value m0 , then the null
Hence the mean life time of light bulbs is hypothesis H 0 is set as follows H 0 : m = m0
expected to lie between the interval (1337.35,
1362.65) Alternative Hypothesis
Any hypothesis which is complementary
8.3 Hypothesis Testing
to the null hypothesis is called as the alternative
One of the important areas of statistical
hypothesis and is usually denoted by H1 .
analysis is testing of hypothesis. Often, in
real life situations we require to take decisions For example: If we want to test the null
about the population on the basis of sample hypothesis that the population has specified
information. Hypothesis testing is also referred mean m i.e., H0 : m = m0 then the alternative
to as “Statistical Decision Making”. It employs hypothesis could be any one among the
statistical techniques to arrive at decisions in following:
certain situations where there is an element of
(i) H1 : m ≠ m0 (m > or m < m0)
uncertainty on the basis of sample, whose size is
fixed in advance. So statistics helps us in arriving (ii) H1 : m > m0
at the criterion for such decision is known as
Testing of hypothesis which was initiated by J. (iii) H1 : m < m0
Neyman and E.S. Pearson. The alternative hypothesis in H1 : m ≠ m0
For Example: We may like to decide on the is known as two tailed alternative test. Two
basis of sample data whether a new vaccine is tailed test is one where the hypothesis about the
effective in curing cold, whether a new training population parameter is rejected for the value of
methodology is better than the existing one, sample statistic falling into either tails of the
whether the new fertilizer is more productive sampling distribution. When the hypothesis
than the earlier one and so on. about the population parameter is rejected only
for the value of sample statistic falling into one
XII Std - Business Maths & Stat EM Chapter 8.indd 189 2/29/2020 11:40:18 AM