:e!

.. Distinguish between a population, sample, parameter, and statistic .

Define an "estimator" and "estimate" and list the characteristics of a good estimator.

Q Compare and contrast point estimation and interval estimation.

00 Describe the primary methods of point estimation (MoM, MLE, Least Squares).

Understand and interpret confidence limits and confidence coefficients.

Calculate confidence intervals for means and variances of normal populations.

 •
•
•

Population, Sample, Parameter, and Statistic

A population is the entire group that you want to draw
conclusions about. It includes every member of the group of
interest. Target Population

f
It can be large or small, as long as it includes *all* the items e
I I I e
f
you are studying. I
f

Examples: I I •
I
f Sample
• All voting-age citizens in a country. I

• All lightbulbs produced by a factory last month.

• All grains of sand on a beach.

A sample is a subset of individuals from a larger population.
It is the group you will actually collect data from.
Target Population
Why do we sample?
f
e
It is often too costly, time-consuming, or impossible to I I I e
f
I
collect data from an entire population. We use a sample to f

make inferences about the whole population. I I •

I
f Sample
Example: I

• Population: All 10 million citizens.

• Sample: 1,000 citizens selected for a poll.

Population
• The entire, complete group.

• Size: N

• Characteristics are called Parameters. ••••• •••••••••••••••

'''''''''''''''
''''' •••••••••••••••
'''''''''''''''
• Data is fixed and unknown.
Sample Population
Sample Mea1n Mean
• A subset of the population. WW\[Link] corn

• Size: n

• Characteristics are called Statistics.

• Data is known and random (depends on which sample

you get).
 at is a Parameter?

Parameter Common Parameters

A parameter is a number that describes a µ (Mu) = Population Mean
characteristic of the population.
a2 (Sigma-squared) = Population Variance
Parameters are usually unknown (because we can't
p = Population Proportion
measure the whole population) and are represented
by Greek letters.
 at is a Statistic?

Statistic Common Statistics

A statistic is a number that describes a x (x-bar) = Sample Mean
characteristic of a sample.
s2 Cs-squared) = Sample Variance
We calculate statistics directly from our sample data.
p (p-hat) = Sample Proportion
We use statistics to *estimate* unknown parameters.
Statistics are represented by Latin letters.
 eCoreo

Fixed and
This is the fundamental idea that connects everything: Population Para,eaeter Unknown

1. We want to know somethingabout a Population (the

Representative Estimate
Parameter,µ).

2. We can't measure the whole population,so we take a

Random
Sample
Sample. Variable

3. We calculate a Statistic from our sample (the Statistic, x The above is a diagram of the process of the process of using data to obtain information about a
population. Please answer the following:
). a) What is the difference between populations and samples?
b) What is the difference between parameters and statistics? Explain using an example.
c) How can we obtain representative samples from a population?
4. We use the Statistic to make an educated guess, or d) Describe why parameters are "fixed and unknown" and why statistics are "random
variables"
*inference*, about the unknown Parameter. e) How do we estimate parameters using statistics? What are the things that we need to
take into consideration when estimating parameters using statistics?
Estimator vs. Estimate
Estimator vs. Estimate: Visua

Think of the Estimator as the recipe for baking a cake. It's 12

the set of instructions (the formula). 10 Actual value of y

v" you are
8
atrempting to
predict
The Estimate is the actual cake you baked. It's the single,
6
specific outcome you got when you followed the recipe with
4
your specific ingredients (your sample data).
P?-dia~dVaill-=
2 Error of ofy
estimating µ \'
If you get a new sample (new ingredients), the same •

0 2 4 8 10 12
Estimator (recipe) will produce a new Estimate (a new cake).
 • •
•
•
•

How do we know if our rule (estimator) is a good one?

@ Unbiasedness: On average, does it hit the true target?

Consistency:Does it get better as we get more data?

Efficiency:Does it have the smallest possible variance?

II
....- Sufficiency:Does it use *all* the information available in the sample?
 •
1. Un iase ness

An estimator 8 is unbiased if its expected value (the

average of all possible sample estimates) is equal to the
.
farameter: Data taken from a [Link]
"

4
I "
f
true population parameter 8.
BIAS - when our statistic
does not accurately estimate
"
our parameter
E(8) =8

In English: The estimator is "correct on average." It doesn't (•) H~h blD&; lowvarta~llliy (b} Low bla&; high vanabl'fty

systematically overestimate or underestimate the •

•
parameter.
-
Example: The sample mean X is an unbiased estimator for
the population mean µ.
 •
Visua izin iase ness

If we could take thousands of samples and plot the

estimate from each one (this is called a sampling
distribution), an unbiased estimator's distribution will be
centered perfectly on the true parameter.

A biased estimator's distribution will be centered on the

wrong value.

85 87 89 91 93 95 97 99 101 103 105 107 109 111 113 115

88 88 90 92 94 98 98 100 102 104 108 108 110 112 114
AppleMeans
 •
Exam e: Biase vs. Un iase

Unbiased: Sample Mean Biased: Sample Variance?

-
The sample mean Xis unbiased. If you took 1000 If you calculate sample variance by dividing by n, it is
different samples and calculated 1000 sample a *biased* estimator (it tends to be too low).
means, the average of those 1000 means would be
We use n - 1 (Bessel's correction) to make it an
very close to the true population mean µ.
*unbiased* estimator:

s2 =
2. Consistenc

An estimator is consistent if it converges in probability to 0.45

the true parameter as the sample size n increases.

0.3~
A
- sar,pllng
lim P ( 8 n -8 <e) =1 a lstrlubutlon of
n- oo 0.25 Mean, n = 100

In English: As your sample size gets larger and larger, your - sa r1 piing
0.15 a istrlbutlon of
estimate gets closer and closer to the true value. mean uno er v-alt,
n = 100
0.05
More data = More accuracy. This is a very desirable
property. -0 .. 0~ - -6-5-4-3-2-1 1 2 3 4 S 6
 • •
3.E icrenc

60
•
An estimator is efficient if it has the smallest possible 50
•
variance among all unbiased estimators. This is called the • •
Minimum Variance Unbiased Estimator (MVUE).
40
• •• •
30
"
Var ( 81
"
< Var ( 82 •
) ) ••
20
• • •
In English: If we have two unbiased estimators (both •
10

"correct on average"), we prefer the one that is more •

0
*precise*. The efficient one gives answers that are more
tightly clustered around the true value. •
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
x
Exam e: ean vs.

Efficiency Example But ...

If the population is Normally distributed, both the If the population is *not* normal (e.g., has outliers),
sample mean ( X) and the sample median are the sample median is often more robust and can be
*unbiased* estimators for the center. preferred, even if it's less efficient in the normal case.

However, the variance of the sample mean is smaller This is a classic trade-off: Efficiencyvs. Robustness.
than the variance of the sample median.

Therefore, for normal data, the sample mean is more

efficient.
 • •
[Link] icrenc

9 A statistic is sufficient if it uses *all* the information in the sample about the parameter .

•
f@ In English:Once you have calculated the sufficient statistic, the original sample data provides no
*additional* information about the parameter.

Example: For a Bernoulli population (coin flips), the number of successes ( k) is a sufficient statistic for
the probability of success ( p). You don't need to know the *order* of the flips (e.g., H, T, H vs. H, H, T) to
estimate p. The total number of heads is "sufficient".
 • •
•
•

Point vs. Interval Estimation

1. Point Estimation

~ number line with a single red dot labeled 'Point Estimate', showing its precision
but isolation

A point estimate is a single number that serves as our "best

guess" for the unknown population parameter.

It's what we've been discussing so far: xis a point estimate

forµ.

Example:

"Based on our sample, the estimated average height is 175

II
cm.
Pros an Cons o Point Estimation

O Pros:
Simple, precise, and easy to report.

Gives a single, direct answer.

Cons:
It is virtually guaranteed to be wrong.

The chance that our x is *exactly* equal to µ is zero.

It gives no information about our uncertainty

. How confident are we in that "175 cm"? Is it ±1 cm or
±50 cm?
2. Interva Estimation

~ number line showing a 'Point Estimate' in the center and a 'Margin of Error' bar
extending on both sides, labeled 'Interval Estimate'

An interval estimate is a range (or interval) of values that is

likely to contain the unknown population parameter.

This range is calculated from the sample data.

Example:

"We are 95% confident that the true average height is

between 172 cm and 178 cm."
O Pros:
It explicitly states our level of uncertainty
.

A wide interval tells us we are not very certain; a narrow interval tells us we are very precise.

It is much more informative and honest than a point estimate.

Cons:
More complex to calculate.

Can be more difficult for a non-technical audience to understand.

 •
•
•
• •

How do we find the "estimators" (the formulas)?

Common

Method of Moments (MoM): A simple, old method that equates sample moments to population
moments.

J!::
••••
Maximum Likelihood Method (MLE): The most popular and powerful method .

VX Method of Least Squares (OLS): The method used to find the best-fit line in regression.

O Method of Minimum Variance: A principle (MVUE) rather than a direct method.

Method of Minimum Chi-Square: Used for categorical data.

1: omen ts
 ean

1. Population 2. Sample
Moment Moment 3. Equate and Solve
The first population The first sample Set them equal to find the estimator for µ:
moment is the moment is the -

expected value: sample mean:

µ =X
n - The MoM estimator for the population mean is the sample
E[X] =µ 1n . 1: X·1 = X
1= 1 mean. Simple!
1. Population 2. Sample
Moment Moment 3. Equate and Solve
For an Exponential( A The first sample -
Set them equal: ! = X
) distribution, the moment is just the 'A

first population sample mean: A

Solve for the estimator "A:

moment is: -
x
E[X] ={
O Pros:
Very simple and intuitive.

Easy to calculate.

Often provides a good starting point.

Cons:
Often *not* the most efficient estimator (can be high variance).

Can produce estimates that are nonsensical (e.g., a negative estimate for a variance).

May not be unbiased.

 2: aximumLi

This is the most common and important method, Likelihood and Log-Likelihood Function of Poisson Function
0.12 0

---- ----- ----- ------------------ ------- --------

developed by R.A. Fisher. ---- -------
- Likelihood
-- Log-Likelihood ...... ... ------
0.1 .,,,/
.,,.,, 10
,,,,",.,,
The Core Question: 0.08
/
/
,, /,,

I
I 20
I
I
I
I
0.06 I
"Given the data we *observed*, what value of the I
I
I

I
I 30

parameter e 0.04
I
I
I
I
I

I
I
40
0.02;
It finds the parameter value that *maximizes* the
o-l----~---===:::::::;:.._---~--~---~---~----=i.50
LikelihoodFunction. 0 0.5 1 1.5 2 2.5 3 3.5
1. Write the LikelihoodFunction L ( 8 ) : This is the joint probability of observing your data, given a
parameter 8. (Often a large product).

2. Write the Log-LikelihoodFunction 1 ( 8): Take the natural log ( ln) of L. This turns products into
sums, which are *much* easier to work with.

3. Differentiate and Solve: Take the derivative of 1 ( e) with respect to 8, set the derivative to zero,
and solve for 8. This finds the maximum (the peak) of the function.
LE: Exam

The Setup The Result

You flip a coin 10 times and get 7 Heads (Successes). If you take the log, differentiate with respect top, and
set to 0, you get:
Data: n = 10, k = 7
Parameter: p = Probability of Heads p = {o = 0.7
Goal: Find the MLE for p.
In English: The value of p that makes getting 7 heads
The likelihood function is the Binomial probability: in 10 flips most likely is ... 0.7. It's perfectly intuitive,
and MLE gives us the formal math to prove it.
2nd Sessional start

LE: Pros an Cons

O Pros:
Generally the "best" estimator.

Consistent and Efficient (achieves minimum variance as n gets large).

Invariance Property: The MLE of a function of e is just the function of the MLE of 0. (e.g., MLE of µ2
.IS µ"'2) .

Cons:
Can be mathematically *very* difficult to compute.

May be biased for small sample sizes.

 3: Least S uares

This method is used when we are trying to fit a model to

Predicted vs Actual Residuals
data, most famously in linear regression.
55
• •
50
.
.••••. ...~· •
.• .. 2
• • •
The Concept: • •
cu
45
.•••••• .
.
•
••
.
•
• • • •
Find the parameter estimates (e.g., the slope and intercept
:,
c
cu
>
cu 40
ex: . -
••••••
•••
..
. ·
•
•
• •
Ill
n,
:,
-0
'ex:la
1
• ••
.. .
•
••
•• •• •
•••
• •
' .
of a line) that minimize the sum of the squared errors (or --
....
0
ui
cu
35
•
..
c
".·
,.,
••..
~
. ••
. -0
s-0
0 . . - .. !
• ~. ! .. - - -~~ ·• .. - ~
• .
:,
n,
>
-n, 30
. .
.... ~- . ....
. ••
.:.·#
• . ·-
....
n,
-0 •• •
• • ••
• •• •
• •
•
#.
•
• • • • • •

"residuals").
,... .
c:
...u
:,
.... 2 -1 •• • •
.·~· . .... • •• • ••
VI
< 25 ••

An error is the vertical distance between the actual data 20

•'
.. ••
.•
•• ~

-2 • •
•

• •
point and the fitted line. 1s=•~~~~~~~~~~~~~~~
20 30 35 40 45 50 20 25 30 40 45
• 50
25 35

Predicted values for Revenue Predicted values for Revenue

n Note that we've colored in a few dots in orange so you can get the sense of how this transformation works.
" 2
Minimize: I: ( Y.1 - Y.1 )
i = 1
O Pros:
The foundation of regression analysis.

Very intuitive and visually understandable.

Under certain assumptions (Gauss-Markov), OLS estimators are the "BLUE" (Best Linear Unbiased
Estimators).

Cons:
Can be highly sensitive to outliers (since errors are *squared*, big errors have a huge effect).

Relies on several assumptions (like linearity, no multicollinearity).

Brie

•••
•••
•••
Min. Chi-Square

This method is for categorical data (e.g., in contingency

tables). It finds the parameter estimates that *minimize*
the Chi-Square statistic between the observed frequencies
in your data and the expected frequencies predicted by
your model.
 •
•
•
• •

Confidence Limits and Confidence Coefficients

Reca • eNee Interva s
•

~ number line showing a 'Point Estimate' in the center and a 'Margin of Error' bar
extending on both sides, labeled 'Interval Estimate'

A point estimate (e.g., x = 175) is precise, but wrong.

It doesn't tell us about [Link] we took

another sample, we'd get x = 176. And another, x = 174.5.
An interval estimate uses this variability (measured by the
standard error) to build a "margin of error" around our point
estimate.
Confidence Interval(CI) Confidence Coefficient
The full range of values, e.g., [172, 178]. The "success rate" of the method, expressed as a
proportion.
Confidence Limits e.g., 0.95

The two endpoints of the interval.

Confidence Level
Lower Confidence Limit (LCL): 172
Upper Confidence Limit (UCL): 178 The same thing, expressed as a percentage.
e.g., 95%

This is the 1 - a, where a is the significance level.

 O/o

This is the most misunderstood concept in introductory

95% Confidence Level
statistics.

CORRECT:
95%
•
"We are 95% confident that this *method* works. If we were I

to take 100 different samples and construct 100 different ~ ~

1 Mean 1
I I

95% Cls, we would expect about 95 of those intervals to Confidence Interval

Lower limit Upper limit

capture the *true, fixed* population parameter."

~ maze
The confidence is in the process, not in any single interval.
 RONGInter retation

O WRONG: "There is a 95% probability that the true population meanµ is in our interval [172, 178]."

to
• Why is this wrong? Because the true parameterµ is a *fixed, unknown constant*. It is not a random
variable. It doesn't have a probability of being anywhere. It just "is",

O Either our specific interval [172, 178] *did* captureµ (100% chance) or it *did not* (0% chance). We just
don't know which. The 95% refers to the "success rate" of the method in the long run.
 • •
•
•
•

Calculating Intervals for Normal Populations

 GeneralFormula

Almost all confidence intervals have this same basic structure:

Point Estimate: Our best guess (e.g., x)

Critical Value: A number from a distribution (z, t) that defines our confidence level (e.g., 1.96 for 95%).
Standard Error: The standard deviation of the estimator.
 ean Po u ation Variance a2 Known

0 Scenario: The "textbook" case. We somehow know the true population variance o2. (Very rare in real
life).

Estimator: Sample Mean ( X).

Jtr Distribution: We use the Standard Normal (z-distribution) because we know o.

4 Formula (z-interval):
-
x +
- Za I 2 ( Jri )
For a 95% Cl, Za I 2 = 1.96.
The "Critical Value" for a z-interval comes from the standard
normal (Z) distribution.

To get a 95% confidence level, we want the "middle 95%" of

the data.

This leaves 5% for the tails, or 2.5% in each tail.

The z-score that cuts off the top 2.5% of the data is 1.96.
This is our critical value, za 12.
 ean Variance a2 Un nown

~Graph comparing the standard normal 'Z-distribution' (blue line) to the 't-
distribution' (red dashed line) which is slightly flatter with fatter tails

Scenario: The *real-world* case. We do NOT know a, so we

must estimate it with the sample standard deviation, s.

This adds *extra uncertainty*. To account for this, we

cannot use the z-distribution,

We use the Student's t-distribution

, which is like a z-
distribution but with "fatter tails" to account for the extra
uncertainty.
 ean t interva Formu a

Estimator: Sample Mean ( X).

-'l.r Distribution:t-distribution with n - 1 "degrees of freedom".

4 Formula Ct-interval):
-
x +
- ta I 2 , n - 1 { Jn )
The critical value t depends on the confidence level AND the sample size ( n). As n gets large (e.g., > 100),
the t-distribution becomes almost identical to the z-distribution.
 ean: Exam

Problem Solution
We sample 10 students and find their average height 1. df = n - 1 = 9
is x = 175 cm, with a sample std. dev. of s = 5 cm. Find 2. Critical t-value (from t-table for 95%, 9 df) =
the 95% Cl for the true mean height. to.02s, 9 = 2.262
3. Standard Error = s I ...;n = 5 I -!ID= 1.581
4. Margin of Error = 2.262 x 1.581 = 3.576
5. Interval: 175 ± 3.576
[171.4, 178.6]
CI or Variance a2

~
II)
c di= 2
Q) ('!
O O

To estimate the variance a2, we use the sample variance s2. ~ L------=~==:::::;:::====::;=J
0 5 10 15

The sampling distribution for variance follows a Chi-Square

( x2) distribution with n - 1 degrees of freedom.
~
--en
c
Q)

0
0
~
0
.
di: 4

0
q
This distribution is not symmetric (it is skewed to the 0
0 5 10 15

right), which means the Cl will *not* be symmetric around (\J

,...
s2. ~
0

(I) di: 6
.iii q
c O
Q)
0 ....
q
0

0
0
0
0 5 10 15
Chi Square
CI or Variance Formu a

Estimator: Sample Variance ( S2 ).

-'1r Distribution:Chi-Square ( x 2) with n - 1 degrees of freedom.

4 Formula:

( n - 1) s2 ( n - 1 ) s2
2
Xa I 2, n - 1
' x21-a/2 n-1
'

Note the "flip": the larger critical value ( x2 ) goes in the denominator for the *lower* bound.
a/2
• I
•

Comparing µ1 and µ2
CI

..
:.:. Goal: We want to estimate the *difference* between the means of two independent groups (e.g.,
"Treatment" vs. "Control").
-
Point Estimate: ( X1 - •

x
2
)

Interpretation: If the 95% Cl for the difference contains zero (e.g., [-5, +10]), then we are NOT 95%
confident that there is any real difference between the means!
 Paired(Dependent) Samples
IndependentSamples
The two groups are related.
The two groups are totally unrelated.
Example: Patient's blood pressure *before* and *after* a
Example: 50 men vs. 50 women.
drug.
I - t wf!h 21 <Mgf"5 0, l'rNQOm

100 *
i

75
'
(I)

"'c0
c. 50
"'
(I)
0::
""9ct lhe ~POhli& of «.AT Retitct the ~[Link] d equal
meeN' ttletest~ h leloe,, "*"" It ths ~ Nllmtic • iarp
25
"""''*'
•
~
0
0 1 3 4
Before After
CI a
0 Scenario: Independent samples. We know both population variances, af and a~. (Again, rare) .

..JL- Distribution: z-distribution.

4 Formula:

+
-
CI a
40

30
Scenario: Independent samples. We don't know the
variances, but we assume they are equal ( of = o~ ). ......
c
Q)

~
Q)
20
We calculate a "pooled" sample variance, s~, which is a Cl..

weighted average of si and sz,

10

Distribution: t-distribution with n, + n2 - 2 degrees of

freedom.
o=---=====---~--====--~---=====--~----=====~----======----=-.,
50 55 60 65 70 75
Height

Sex --F --M

• •

Comparing ai and a~
 30
Scenario: Independent samples. We don't know the
variances, but we assume they are equal { ifi = ~ ).
i,
<.:>
;;; 20
We calculate a "pooled" sample variance, s~, which is a a.

weighted average of sf and sz,

10

Distribution: t-distribution with n1 + n2 - 2 degrees of

freedom.
0
55 60 65 75
Heigh1
__ , --U
CI for Ratio of Two Variances
Comparing a~ and~
CI for of I 02

Goal: We want to compare the *spread* or *consistency*

of two independent groups.

We estimate the ratio of population variances, crf I 02.

Distribution
: The ratio of two sample variances follows an
F-distribution
, which is also skewed and defined by *two*
sets of degrees of freedom. F10,2s

(a) (b)
Interpretation
: If the 95% Cl for the ratio contains 1.0, we
are not confident that the variances are truly different.
 .
Unbiased: Correct on average. E ( 0 ) = 0

~ Consistent
: Gets better as sample size n increases.

fj Efficient: The unbiased estimator with the "smallest" variance (MVUE).

-
... , Sufficient
: Captures all relevant information from the sample.

Review: Part3 & 4 - Methods

 CI for Ratio of Two Variances
Comparing <Ji and a~

CI for of I 02
CI for af I oz

Goal: We want to compare the *spread* or *consistency*

of two independent groups.

We estimate the ratio of population variances, a~ I <12.

Distribution: The ratio of two sample variances follows an

F-distribution, which is also skewed and defined by *two*
sets of degrees of freedom. F10,2s F4040
'
(a) (b)
Interpretation
: If the 95% Cl for the ratio contains 1.0, we
are not confident that the variances are truly different.
 .
@ Unbiased: Correct on average. E ( 0 ) = 0

~ Consistent: Gets better as sample size n increases.

e Efficient: The unbiased estimator with the *smallest* variance (MVUE).

!a! Sufficient: Captures all relevant information from the sample.

• - ... .. . .. ,
Review: Part 3 & 4 - Methods

O Point Estimation: A single-number guess.

fffl Interval Estimation: A range that accounts for uncertainty.

O Method of Moments (MoM): Simple, equate moments.

~:: Maximum Likelihood (MLE): Most common, most likely .

••••
..fx Least Squares (OLS): For regression, minimizes squared error.
Review: Part 5 & 6 - Confidence

X 95o/o Confident: Refers to the long-run success rate of the "method", not the probability for one
interval.

L Margin of Error: (Critical Value) x (Standard Error).

 -----
••••• Each
.
S,QU¥f' • "'
. .. .
of 01l' wmo(~'s favorite fruit choicest.,,;auq
' ' .• (1,;
, (<)

1,,,,,,,,,,,,,,,,,,,1 I ?
' ' A
I
'

•
••
••• I
"
c
D
E
i ./
./
-
-,'.\_
"
c
D
<
••••
••••• l- I w In I ro I I" I"!" I"' I ro J" I"-
0'

1,, ,111111•••••
<Q
.1
1111,,1,,,1 t :,

,/

• •
•
••
• • •
•I •I • •I • I • • I
•••••
••••••••••
•••••••••• ·~·,y --
7 ,/

![Link]..il
D \\'1111.:;·U

-
=i Fl111 otQ -

b11·11(1,,.'\\0,.....·r~1h
I I I I I I LI I I I I I I I I

Unbiased Consistent Efficiem

 ------- . . .. .
1,,,1,,,,,,,1,,,1,,,1•••••
Eaeh sq~ • 11' of UW wmp<e'sfavorite fnJJt choices 4i,;a.l()(Jt (b> M <J'.l
I,
' ' •• 1234!1 1?3~!1

• f-+-+-1--1-1 ••
12

I " !./ e
f-+-+-+--1-1 • !•
••• I
c. ./ ./ c
0
./.

•••
•••• •
••••• ro] ll~lo I
••••• 1, ,,1,,,1 (d ,
1,,, 1,,,,,,, 3 t :, 2 3 4
' •

• •
• ••••• ./
./ ./ -
• • • •
• •••• •• •
l111l111l111l111l111I
••••••••••
••••••••••
~ ~
D \\'11!1(0:U 0•1111<'\<U -0.1!lt.'ccdl

- [x1m.°tCdf1,.y..·rMh - -
=i~~~ . :;:
S:r~httlO'""l':,:h

Unbiased Consistent Efficient

Ima e Sources

t. . [Link]
I •';tr
1i
ti. I t'
i t t Sample Source: [Link]

iilllillillllli [Link]
VS iiiiitiiitiHH
Population
Mean Source: [Link]

- ·- [Link]
- ...y--E::::-:::J=
_____ _ __ --,-· -.
___ .
.......... ...,....
. ..... . -----·---
'! :"''"' !"":--:-~
Source: [Link]

[Link] [Link]

Source: [Link]

[Link]
On4CLCOmJmjzt18VcRzV9PMyvtgxo7xlQ
lol .... _ .. ........, (» ... - .... -

Source: [Link]

[Link]

Source: [Link]
Ima e Sources

[Link]

Source: [Link]

\
[Link] files/figure-html/[Link]
-:• • • ,• .. ,'I·.· ~.·
.: ·~
. ...---·-__p- ••
. /.\. . . . . Source: [Link]
I . .

,1 -..1 ... _ .. _ ...... _ ... _. [Link] [Link] m/vi/h5yrl Dk nCdA/hqdefault.j pg?sqp=-oaymwEm COA DEOgC8quK qQMa8A EB-AH U BoAC4AOKAgwlA BA BGG U gZS hi MA 8 =& rs =AOn4CLCMzYe TOrW 013 Rl b TQx
'
' lgYXdOZ5GA
••
-, .
.. ~ • (l,,I

Source: [Link]

[Link] mg. co m/vi/J QuP alYs ivYI [Link]

[ 0, 3)
Source: [Link]

[Link]

Source: [Link]

..,.. ..... . .
···~·
..
.c":"~·
,
•

.. .,
. . . . . .. . ,,.,·.. .. . . .
•• •!
.
.
I . . • -· •' ..·~
..
. . .
• '.... ,
.. . .
[Link] ml m/assets Is uppo rt/wp-co ntent/ uploads /2017 /07 /Screen-Shot- 2017-07-19-at-9.46.11-A M. png
I'".·.:.. . . ...
\.
;
. .. . -· ..
•I • • Source: [Link]
Ima e Sources

[Link]
...
~
-- - ....... -. Source: [Link]

[Link]
FCritical_ Values_Ol.png&w=3840&q=75

Source: [Link]

;,:
•
§
,,,

N
l ·-·-
o..
,_....,_
......

ca-w ..
-
-nii-
[Link]
ww,....... '

Source: [Link]

[Link]
J•
a • •••

Source: [Link]

[Link]

Source: [Link]

•! " [Link]
8. 50

i
" Source: [Link]
Ima e Sources

' [Link]
; .
•
Source: [Link]

\ [Link]

Source: [Link] i [Link]

•

•••••
1, 11111,•••••
111 ,1, 111111 I [Link]
•
• •• ••• ••• •• • ••
1 , II I II I I II " I II 111 I I Source: [Link]

[Link]

.....
..••••••.• ·~~-
••••••••••
[Link]?width=600&height=600&name=[Link]

Source: [Link]

https:// pub. m dpi-res. com/water /water-17-01273/article _ deploy/htm 1/i m ages /water-17-01273-gOOl. png?1745671093

Source: [Link]
----·

2nd Sessional End