BBA – 2 (Spring – 2022)
Statistical Inference
Estimation
Ahmad Jalil Ansari
Contact:
[email protected] Statistical Inference
A process of drawing
conclusions about an
underlying population based on
a sample
Learning Objectives
To learn how to estimate certain characteristics of a
population from sample
Point Estimator
Interval Estimate or Confidence Interval
To calculate how accurate our estimates are
t-Distribution
Concept of t-Distribution
Z-statistics vs t-Statistic
Interval Estimate for
single population mean (µ) for large sample
single population mean (µ) for small sample
µ1 - µ2 when population variances known (Large sample
case)
The difference between two populations proportions (p1 - p2)
Part I
Estimates and Estimators
Learning Objectives
Concept of Estimates and Estimators
Types of Estimates
• Point Estimate
• Interval Estimate
Estimates
As a human being we have to estimate every forthcoming
event
Should we wait, walk or run to cross a road
Monthly earning and expenditure
Cost of certain events etc.
Business Manager needs to estimate
Budget
Performance of various activities
Risk etc.
The estimates are based on past experience whether it is
Intutive or Scientific
To calculate the future event with 100% accuracy is impossible
The scientific estimates are based on past data and provides:
Reasonably good estimates
A statement which provides the chances of error in estimation
Types of Estimates
A sample statistic can provide two types of estimates about the
unknown parameter of the population from where it is taken
Point Estimates
Interval Estimates
Point Estimate: It is a single number that is used to estimate a
population parameter:
To estimate the average height of students in a class is 68 inches
To estimate the average weight of cereal canned is 500 grams
Interval Estimates: It is a range of values that is used to estimate a
population parameter:
To estimate the average height of students in a class is between
65 and 70 inches
To estimate the average weight of cereal canned is between 480
to 520 grams
Estimator
Any sample statistic that is used to estimates population
parameter is called estimator
A sample mean 𝑥 can be estimator to estimate population mean
µ.
A sample proportion 𝑝 can be estimator to estimate population
proportion p.
A sample range can be estimator to estimate population range.
Estimate: It is a specific numerical value of the estimator.
If we calculate the average mileage of a sample of used cars for
sale as 98,000 mile, then we can use this value to estimate the
average mileage of all used cars for sale in a fleet.
Point Estimates – Sample Mean
If we know the sampling distribution of 𝑥 we can make statements
about any estimate we may make from sampling information.
Consider following sample data about the count of syringes in a
carton. Each syringe is jumble packed in a corrugated carton by
Producer. Supplier of these syringes want to estimate the number of
syringes in a carton (expected to be 100). 101, 103, 112, 102, 98, 97, 93
105, 100, 97, 107, 93, 94, 97
𝑥 97, 100, 110, 106, 110, 103, 99
𝑥= = 3570 / 35 = 102 syringes 93, 98, 106, 100, 112, 105, 100
𝑛
114, 97, 110, 102, 98, 112, 99
Using the sample mean 𝑥 as our estimator, the point estimate of
the population mean µ is 102 syringes per carton
As price is very nominal for each syringe (25 cent) it is acceptable.
Point Estimates – Sample Variance
If we know the sampling distribution of s2 we can make statements
about an estimate of population µ2
Consider same sample data about the count of syringes in a carton.
(𝑥 − 𝑥)2
s2= = 1228 / (35 – 1) = 36.12
𝑛 −1
101, 103, 112, 102, 98, 97, 93 105, 100, 97, 107, 93, 94, 97
97, 100, 110, 106, 110, 103, 99 93, 98, 106, 100, 112, 105, 100
114, 97, 110, 102, 98, 112, 99
Above formula for variance provides us unbiased estimator of
population variance.
Using the sample variance s2 as our estimator, the point estimate of
the population variance σ2 is 36.12 syringes and standard deviation
is estimated as σ is 6.01 syringes
Another formula for finding variance of sample is
1
𝑠2 = ( 𝑥2 − 𝑛 𝑥2)
𝑛−1
Point Estimates – Sample Proportion
If we know the sampling distribution of sample proportion 𝑝 we
can make estimate about population proportion p
Suppose management wants to estimate the proportion of the
damaged syringe carton received by customer to find the handling of
these cartons.
A random sample of 50 cartons showed that 4 of these cartons were
damaged.
𝑑𝑎𝑚𝑎𝑔𝑒𝑑 𝑐𝑎𝑟𝑡𝑜𝑛𝑠
𝑝= = 4 / 50 = 0.08
𝑁𝑜 𝑜𝑓 𝑐𝑎𝑟𝑡𝑜𝑛𝑠 𝑖𝑛 𝑡ℎ𝑒 𝑠𝑎𝑚𝑝𝑙𝑒
Sample proportion is convenient estimator of population mean.
Therefore, we can estimate that the proportion of the damaged
cartons in the population is 0.08.
Point Estimates – Class Assignment
An event management company is considering expanding its
seating capacity. It needs to know both the average number of
people who attend the events and variability of this number.
Following are the attendances (in thousands) at randomly
selected events. Find the point estimate of mean and variance of
the population from which the sample was drawn:
8.8 14.0 21.3 7.9 12.5 20.6 16.3 14.1 13.0
Point Estimates – Class Assignment
8.8 14.0 21.3 7.9 12.5 20.6 16.3 14.1 13.0
# x x2
1
2
3
4
5
6
7
8
9
Total
Point Estimates – Class Assignment
8.8 14.0 21.3 7.9 12.5 20.6 16.3 14.1 13.0
# x x2 𝑥
𝑥= = 128.5 / 9 = 14.2778 thousand of
1 8.8 77.44 𝑛
people
2 14.0 196.00
Using the sample mean 𝑥 as our
3 21.3 453.69
estimator, the point estimate of the
4 7.9 62.41
population mean µ is 14,278
5 12.5 156.25
6 20.6 424.36 1
7 16.3 265.69
𝑠2 = ( 𝑥2 − 𝑛 𝑥2)
𝑛−1
8 14.1 198.81 1
𝑠2 = (2003.65 − 9 × (14.2778)2 )
9 13.0 169.00 8
Total 128.5 2003.65 s2 = 21.11873
Using the sample variance s2 as our estimator, the point
estimate of the population variance σ2 is 21.12
Interval Estimates – Sample Mean
It describe a range of values within which a population
parameter is likely to lie
It is an statement about the uncertainty that would be likely to
accompany this statement.
For that we need to find standard error of mean
We know that if we select a large number of means and plot them,
it will approximate a normal curve (Central Limit Theorem) with:
Mean of sample mean = population mean
Measure of dispersion
𝜎
Standard error of mean = 𝜎𝑥 =
𝑛
Now we can calculate the interval of different width based on the
confidence level required by using (for mean)
Interval = µ ± SE (z-value of confidence level required)
Interval Estimates – Example
A company needs an estimate of average life of car batteries it
manufactures. It selected 200 users and found that average life of the
batteries as 36 months. They knew that the standard deviation of the
population of all batteries is 10. Find the confidence intervals?
Mean of sample mean = 36 P(-1 < Z < 1) = 0.84134 – 0.15866 = 0.68268
Standard error of mean = 𝜎𝑥 P(-2 < Z < 2) = 0.97725 – 0.02275 = 0.95450
𝜎 10
= = = 0.707 P(-3 < Z < 3) = 0.99865 – 0.00135 = 0.99730
𝑛 200
Therefore,
If z =1, the actual life of all batteries lies between 36 – 0.707 and
36 + 0.707 or between 35.293 and 36.707
If z = 2, the actual life of all batteries lies between 36 –2 × 0.707 and
36 + 2 × 0.707 or between 34.586 and 37.414
If z = 3, the actual life of all batteries lies between 36 – 3 × 0.707 and
36 + 3 × 0.707 or between 33.879 and 38.121
Interval Estimates – Example
Mean of sample mean = 36
Standard error of mean = 𝜎𝑥 =
𝜎 10
= = 0.707
𝑛 200
Therefore,
If z = 1, the actual life between
35.293 and 36.707 with
probability = 68.27%
If z = 2, the actual life between
34.586 and 37.414 with
probability = 95.45% P(-1 < Z < 1) = 0.84134 – 0.15866 = 0.68268
P(-2 < Z < 2) = 0.97725 – 0.02275 = 0.95450
If z = 3, the actual life between P(-3 < Z < 3) = 0.99865 – 0.00135 = 0.99730
33.879 and 38.121 with
probability = 99.73%
Thank you
Ahmad Jalil Ansari
Contact:
[email protected]
