Statistics for Economists

09-04-24
Descriptive statistics: methods of organizing, summarizing & presenting data in a useful and informative
manner. Two methods:
1. graphical/tabular techniques such as a histogram
-chapters 2 & 3
2. numerical techniques such as mean, mode, etc.
-chapter 4
Inferential statistics: methods for drawing conclusions/inferences about characteristics of a population
based on a sample.
-population: group of interest of a study
-sample: subset within a group of interest in which the data is actually drawn
-parameter: a descriptive measure of a population
-ex: population mean, population variance
-often uses greek letters
-population mean: mu - μ
-population variance: sigma - σ
-statistic: descriptive measure of a sample
-sample mean - x
-sample variance - s^2
-statistical inference: the process of drawing conclusions/inferences about a population on basis of sample
-because statistical inference is not always accurate, we build into statistics measures of reliability such as
confidence levels and levels of significance
-confidence level + significance level = 1
-confidence level: the proportion of times that an estimating procedure will be correct
-significance level: measures how frequently the conclusion will be wrong - α

09-06-24
-variable: characteristics of a population or sample
-grades, height, age, etc.
-values: all possible values of variables
-data: observed values of variable (from population or sample)
Three types of data:
1. Interval data:
a. Numerical and quantity data
b. Real numbers(negative infinity to positive infinity)
c. Can perform operations - addition, multiplication, etc.
2. Nominal data
a. Qualitative or categorical
b. Values are categories
c. Mostly based on counts of frequencies of the values
3. Ordinal data
a. Also called ranked data
 b. Categorical in nature, but there is an order, ranking
c. Cannot perform operations
d. Ex: consumer ratings like 5 stars, 4 stars
e. If the values are categories and you can objectively rank them, it is ordinal
-can describe nominal data tabularly or graphically
-(relative) frequency distribution table
-bargraph
-pie chart
-relative frequency distribution table: table that lists all categories(values and variables) and proportions
-bar graph: displays frequency distribution.
-x-axis: variable
-y-axis: counts(frequencies)

09-11-24
Steps in constructing a frequency distribution table:
1. Collect data
2. Determine appropriate # of classes
a. Sturges formula: # of classes = 1 + 3.3log(n)
3. Compute range
4. Compute width - width = range/# of classes
5. List classes
6. Determine frequencies of each class

9-13-24
Cumulative frequency table steps:
1. Compute frequency counts
2. Compute cumulative frequencies
3. Graph
Time series v. Cross sectional data:
1. Cross sectional data - data measured at one point in time
a. Relationship between 2 interval variables (observations are cross sectional)
b. Scatter diagram
2. Time series data - observations measured over time (at successive points in time)
a. Line chart
Describing the relationship between 2 interval variables:
1. Use 2 cross sectional data sets (variables)
2. Look at relationships between 2 variables - x and y
a. X - independent variable
b. Y - dependent variable
3. Scatter diagram/plot is graph of relationship between x and y
a. Look for pattern
b. Linear relationship - positive or negative
c. No linear relationship
-ogive: graphical representation of cumulative (relative) frequency
Chapter 4 - Descriptive statistics - Numerical techniques
1. Measures of central location/tendency
a. Mean - average
i. Arithmetic (simple) mean:
1. Population mean
2. Sample mean

3.
ii. Geometric mean
1. Applies to growth rates
b. Median - observation in the middle of data (after we order observations)
2. Measures of dispersion, variation, spread, variability
3. Measures of relative standing
4. Measures of linear association/relationship

09-20
1. Coefficient & Variation
-relative measure of dispersion
-used when comparing dispersion of several data sets with different means
-standardized measure of dispersion

-
-measure standard deviation as a percent of the mean
2. Measures of relative standing
a. Z-scores: measure standing of observation as does not equal of standard deviations away
from the mean
 b.
c. (pop z-score = x - population mean over the population standard deviation)
d. (sample z-score = x - sample mean over the sample standard deviation)
-assuming any distribution(unknown distribution) we can say at least (1-(1/k^2)) of all observations have
z-scores between -k and k
Detecting Outliers:
-bell shaped(empirical rule): any observation whose absolute value of z-score is more than three
is deemed to be an outlier
-unknown distribution(Chebyshev rule): any observation whose absolute value of z-score is
greater than two is a suspected outlier
3. Box plots:
a. Graphical approach
b. Graphs 5 statistics:
i. Minimum
ii. Maximum
iii. 1st quartile
iv. 2nd quartile
v. 3rd quartile
c. Whiskers - lines extended to the left and right of the box
i. Whisker = 1.5 x (Q3-Q1)
d. Box plots can be used to detect outliers–compute the whiskers
4. Fractiles: Divide data set into equal parts
a. Fractile examples:
i. Percentiles - divide data set into 100 equal parts *most relevant
ii. Deciles - divides data set into 10 equal parts
iii. Quartiles - divides data set into 4 equal parts
b. Median is technically the 50th percentile, 5th decile, 2nd quartile
Finding values of Pth percentile:
1. Order data
2. Find location - P = (n+1) x (P/100)
3. Example:
a. Data: 4, 7, 8, 3, 20, 22, 14, 18, 19, 25
b. N = 10
c. 3, 4, 7, 8, 14, 18, 19, 20, 22, 25
d. L25 = (10+1) x (25/100) = 2.75
 e. L75 = (10+1) x (75/100) = 8.25
f. 25th percentile = Q1 = 4 + .75(7-4) = 6.25
g. 75th percentile = Q3 = 20 + .25(22-20) = 20.5
h. Interquartile range!!!(IQR) = Q3 - Q1 = 20.5 - 6.25 = 14.25
i. IQR is a measure of relative standard

09-25-24
4. Measures of linear relationship
a. Covariance: 1st measure of linear relationship
i. For population we use greek letter sigma lol to represent parameter: σ

ii.
iii. N = population size
iv. Formulas are very similar to variance

v.
b. Coefficient of correlation(correlation) - greek letter rho(P)
i. Another measure of linear relationship
ii. An improvement over covariance
 iii.
iv. Coefficient of correlation provides information about the direction and strength of the
linear relationship, but we cannot interpret it precisely
1. Strength is measured by how close we are to 1 or -1
2. −1 ≤ r ≤ 1 (same for population coefficient)
v. EXAMPLE: (homework due on canvas 9-26)
1.
x y

6 2

15 4

24 6

30 8
a. Find sxy
b. Find r - what do you say about the relationship between a and b?
c. Find R² - interpret
d. Find the least squares equation
c. Coefficient of determination - R²
i. Another measure of linear relationship
ii. Improvement over coefficient of correlation
1. Can be interpreted precisely
iii. Example from class:
1. r = 0.9444
2. R² = (0.9444)^2 = 0.8911 or 89.11%
3. Interpretation:
a. R² measures the total variation in the dependent variable (y) that can be
explained by the variation in the independent variable (x)
i. (how much x explains the y)
b. The closer to 100% R² is, the better the fit (the better x does at explaining
y)
d. Least squares equation(approach) -
 i. Least squares method produces a straight line that goes through the points on a graph so
that the squared deviation between the points and the line is minimized
ii. Equation: ŷ = b0 + b1x
1. b0 = vertical intercept(y-intercept)
2. b1 = slope
3. Basically y = mx + b
iii. b1 = (sxy /sx²)
iv. b0 = ȳ - b1x̄
v. Example from class:
1. b1 = (sxy /sx²) = 17.5/7 = 2.5
2. b0 = ȳ - b1x̄ = 20 - (2.5)(5) = 7.5
3. ŷ = b0 + b1x = 7.5 + 2.5x = least squares equation
Chapter 6: Probability
1. What is probability?
a. The chance of likelihood that an event will occur
2. Key concepts/terms
a. Random experiment: an experiment that produces one or more possible outcomes
i. Ex: you roll a die and the possible outcomes are 1, 2, 3, 4, 5, or 6
b. Sample space: set of all possible outcomes of an experiment(denoted as S)
i. Ex: roll a die, there are 6 possible outcomes. S = {1, 2, 3, 4, 5, 6}
c. Simple event or sample point: one outcome of the experiment
i. Ex: in the random experiment that is dice rolling, we have 6 simple events
d. Event: subset of a sample space
i. Ex: when rolling dice, event A = observe a number at most 3 = {1,2,3}
1. What is the chance of event A? 50%
3. Two requirements of probabilities
a. S = {E1, E2,..., En}
i. n simple events
1. 0 ≤ P(Ei) ≤ 1
2. The sum of all probabilities from i to n must equal 1
4. Methods for assigning probabilities
a. 3 methods:
i. Classical method(approach) - based on equally likely events
1. If an experiment has n possible outcomes, this method would assign a
probability of 1/n to each outcome. It is necessary to determine the
number of possible outcomes.
a. Experiment: rolling a die
b. Outcomes: {1, 2, 3, 4, 5, 6}
c. Probabilities: each sample point has a ⅙ chance of occurring
i. P(E1) = ⅙
 d.
ii. Relative frequency method(approach) - method for assigning probabilities
based on experimentation of historical data.
1. Construct a relative frequency table - from there you can read the
probabilities(proportions)

2.
iii. Subjective method(approach) - method for assigning probabilities based on the
assignor’s subjective judgment or opinion. *will not use this approach in class*
09-27-24:
5. Combinations and relationships between events
a. Complement events
i. The complement of event A is the set of all sample points(simple events) that are
not in “A”
ii. Can be denoted A’
iii. P(A) + P(not A) = 1

iv.
v. Example: random experiment - test coin twice:
1. S = {HT, HH, TH, TT}
2. A = observe at least one H
 3. P(A) = {HT, HH, TH} = ¾
4. P(not A) = {TT} = ¼
5. P(A) + P(not A) = ¾ + ¼ = 1
b. Intersection of events
i. The intersection of events A and B is the set of all sample points tha are both in
A and B
ii. Denoted by A and B or A∩B

iii.
iv. Joint probability: probability of intersection of two events (A and B) = P(A and
B)
v. Example: random experiment - toss two coins
1. S = {HT, HH, TH, TT}
2. A = observe at least 1 H = {HT, HH, TH}
3. B = observe exactly 1 T = {HT, TH}
4. A and B = {HT, TH,}
c. Union of events
i. The union of two events, A and B, is the set of sample points that are in A or B or
both
ii. Denoted by A or B, A∪B

iii.
iv. Example: random experiment - toss two coins
1. A = observe at least 1 H = {HT, HH, TH}
2. B = observe exactly 1 T = {HT, TH}
3. A∪B = {HT, HH, TH}
d. Mutually exclusive events
i. Events A and B are mutually exclusive if they cannot occur together–i.e. they
have NO elements in common.
 ii. P(A and B) = 0

iii.
e. Marginal probabilities
i. Context of cross classification table
ii. Calculated by adding joint probabilities across rows and down columns
1. They appear at the margin of the table

iii.
f. Conditional probabilities - the probability of an event given that a related event
occured(given the occurrence of some related event)
i. Denoted P(A|B)
1. Read as: “probability of event A given that event B has occured”
ii. Computed as: P(A|B) = P(A and B)/P(B)
iii. Example: random experiment - toss 2 coins:
1. S = {HT, HH, TH, TT}
2. A = observe at least one H = {HT, HH, TH,}
3. B = observe exactly one T = {HT, TH,}
4. P(B|A) = P(A∩B)/P(A) = (½)/(¾) = ⅔
 iv.
g. Independent vs. dependent events
i. Events A and B are independent if the occurrence of one event has no impact on
the probability of the other event
1. P(A|B) = P(A) or P(B|A) = P(B)
ii. Events A and B are dependent if the occurrence of one event has an impact on
the probability of the other event

iii.
h. Union of events(revisited) in the context of the table - Ex 6.1
i. Compute P(A1 or B1)
1. P(A1∪B1) = .11 + .29 +.06 = .46
 2.
3. Rule for finding P(Union of events):
a. P(A1∪B1) = P(A1) + P(B1) - P(A1∩ B1)
6. Rules of Probabilities
a. Complement rule - P(A) + P(A’) = 1
b. Multiplication rule - P(A∩B) = P(B) x P(A|B) *just the conditional probability
formula rewritten* *can be rewritten to solve for any variable*
i. Rule for joint probabilities
ii. Based on the conditional probability formula
iii. If A and B are independent events, then P(A∩B) = P(A) x P(B)
c. Addition rule - P(A∪B) = P(A) + P(B) - P(A∩B) *10-02-24 technically starts
here*
i. Applies to probability of unions of events
ii. If A and B are mutually exclusive, then P(A∪B) = P(A) + P(B)
1. If A and B are mutually exclusive & exhaustive, P(A∪B) =
1

10-02-24

7. Probability trees
a. A tree is an effective and simple method of applying the probability rules
b. The events of an experiment are shown by a line

c.
d. Dots = nodes, lines = branches
 e. There is no requirement that the branches splits be binary–there can be as many branches
as there are events that come from a node
i. There can be as many levels as possible
ii. There does not need to be the same amount of splits at each node/subnode
f. The probabilities associated with any set of branches (from a node) must add up to one
g. At the end of the branches, we can get the joint probabilities *using multiplication rule*
h. P(B) = P(A∩B) + P(not A∩B)
8. Bayes Theorem THERE WILL BE A BAYES PROBLEM ON TEST
a. P(A|B) = P(A∩B)/P(B)
b. Law used to revise prior probabilities, given that new information has arrived
i. A: main event
ii. P(A): prior probability (what we are trying to revise)
iii. B: new information
iv. P(B|A): conditional probability
v. If we know P(B|A), we can use Bayes law to find P(A|B)
c. 5 steps for using Bayes Theorem:
1. Identify the main events → Ai
2. Write the prior probability → P(Ai)
3. Identify the new events → Bj
4. Write the conditional probability → P(Bj|Ai)
5. Use Bayes Law to find posterior probabilities
ii. P(Ai|Bj) = P(Ai ∩ BJ)/P(Bj)
10-04-24
CHAPTER 7
1. Random Variable
a. Rule or function that assigns a number to each outcome of a random experiment
b. Value of a random variable is a numerical event
c. Example: random experiment: toss a coin
i. {H, T}
ii. X → random variable
1. note that X → variable, x → one value of variable
iii. {0, 1}
d. Example: random experiment: roll a die
i. X → random experiment
ii. X = # observed
iii. {1, 2, 3, 4, 5, 6}
2. Type of random variables
a. Discrete random variable - takes on a countable number of values
i. 1, 2, 3, 4… integers
ii. Examples: number of students in a class, number of cars in a lot, number of shirts
one owns; countable values
b. Continuous random variable - does not take on a countable number of values
i. Values belong to an interval
 ii. Infinite number of possible values within an interval
iii. Examples: weight, height, time
3. Probability distributions
a. Distributions of a random variable
b. Probability distribution could be a table, a graph, mathematical formula that describes
values of a random variable and probability associated with each value
c. There are 2 types of probability distributions:
i. Discrete probability distribution
ii. Continuous probability distribution
d. Notations:
i. X → variable
ii. x → a value of variable
e. Examples:
i. What is the probability that a random variable will be 7?
1. We write P(X = 7)
ii. What is the probability that a random variable will be x value?
1. We write P(X = x)
iii. Random experiment - toss a coin
1. X = number of heads
2. What is the probability of observing one head?
3. P(X = 1) or P(1)
4. Discrete Probability Distributions
a. Probability distribution of a discrete random variable
b. Conditions/requirements
i. The probabilities of values of a discrete random variable can be derived by
different statistical tools
ii. Regardless of tools used, 2 requirements:
1. 0 ≤ P(x) ≤ 1 for all x
2. Σall x P(x) = 1
c. Example: a census asks households to report the number of people living in the
household. Develop the probability distribution of the random variable defined as the
number of people per household.
i. X - number of people in the household

ii.
iii. 1/31.1 = .268 and so on, totals to 1
d. Example: a salesman calls three people knowing there is a 20% chance of closing a sale
on each call. Determine the probability distribution of the number of sales the salesperson
will make
i. S = sale → P(S) = 0.2
ii. not S = not sale → P(not S) = 0.8

iii.
iv. Events are independent because the outcome of one call does not affect the
outcome of another start 10-09-24 notes below
e. Describing discrete probability distributions:
i. Probability distributions represent populations
ii. We can compute various parameters about distributions
1. Mean

a.
2. Variance
a.

3. Standard deviation

a.
iii. Laws of expected values
1. X → random variable
 2. E(C) = C, c = constant
3. E(ax + b) = aE(x) + b
iv. Laws of variances
1. V(c) = 0
2. V(ax + b) = a^2 V(x)

5. Bivariate probability distributions

a. In previous sections, we looked at univariate probability distributions, probability
distributions of ONE variable
b. Now we look at bivariate probability distributions = probability distributions of
combinations of two variables
c. They are so called joint probability distributions

d. Notation:
e. The requirements for a bivariate probability distribution are:

i.
f. Marginal probabilities: obtained by adding across rows and down columns to determine
P(X) and P(Y)
g. Describing bivariate probability distributions → describe each random variable
i. Covariance
ii. Coefficient of correlation

6. Binomial Distribution
a. The binomial distribution is the probability distribution that results from doing a
“binomial experiment”; binomial experiments have the following properties:
i. Fixed number of trials, represented as n
ii. Each trial has two possible outcomes, a “success” and a “failure”
iii. P(success) = p and thus P(failure) = 1 - p, for all trials
iv. The trials are independent
b. The random variable of a binomial experiment is defined as the number of successes in
the n trials, and is called the binomial random variable:
i. Discrete random variable

ii.
 c. Cumulative binomial probability - the probability of observing less than or equal to a
given number: P(X ≤ x)
i. Table 1 of appendix B gives the cumulative binomial
probabilities for P(X≤ k) *same thing as x
ii. Need n and P for table(number of trials & probability of success)
iii. NEED APPENDIX B TABLE FOR FUTURE CLASSES
iv. The binomial table gives cumulative probabilities for P(X≤ k),
but we can find X values more than or equal to k by subtracting
our answer from 1:
1. P(X≥k) = 1 - P(X≤(k-1))
d. Describing binomial distribution - mean, variance & standard deviation:

i.

7. Poisson Distribution
a. Discrete probability distribution that results from a Poisson experiment
b. Poisson experiment has four characteristic properties:
i. The number of successes that occur in any interval is independent
ii. The probability of a success in an interval is the same for all equal-size intervals
iii. The probability of a success is proportional to the size of the interval
iv. The probability of more than one success in an interval approaches 0 as the
interval becomes smaller
c. Poisson random variable - random variable of poisson probability distribution
i. Denoted by x
1. # of success(events) in an interval of time, space, area, or volume
d. Poisson probabilities - the probability that a poisson random variable, x, assumes a value
is:

i.
e. Cumulative Poisson probabilities
i. P(X≤ k) = P(0) + P(1) + P(2)...+P(k)
 ii. Table 2 of appendix B gives the cumulative probabilities for
P(X≤ k)
10-18-24
f. Poisson approximations to binomial probabilities
i. Consider a binomial probability distribution:
1. X = # of successes in n trials
2. n = # of trials
3. P = probability of success
4. q = 1 - p = probability of failure
5. M(mu) = E(P) = np
ii. Under certain conditions, we can use a poisson distribution to approximate a
binomial distribution:
1. n ≥ 20
2. p ≤ 0.5
iii. Example:
1. 50 computers
2. P = probability of a computer benign defective: 0.02
3. Find the probability that 5 computers are defective
a. X - binomial random variable, # of defective computers
b. n = 50
c. P = 0.02
d. M = E(x) = np = (5)(.02) = 0.1
e. Binomial distribution: P(5) = (50!/5! x 45!)(0.02)^5 (0.98)^45

CHAPTER 8 - Continuous Probability Distributions

1. Continuous random variables and continuous probability distributions
a. Continuous probability distribution = distribution of a continuous random variable
b. Continuous random variable = variable that assumes an uncountable number of values
c. There are an infinite number of values in any given interval
i. Ex: time, distance, weight, height
d. Point probabilities(continuous distributions)
i. Because the continuous random variable assumes an uncountable number of
values, it’s impossible to list them all
ii. Because there is an infinite amount of values, the probability of each value is
virtually zero
iii. Therefore, we can determine the probability of ranges of values only
1. P(x = a) = 0 not meaningful
2. P(x ≤ a) = P(x < a) meaningful
3. P(a ≤ x ≤ b) = P(a < x < b) meaningful
e. Probability density functions
i. Every continuous probability distribution is defined over its probability density
function(pdf)
ii. Probability density function(curve) = smooth function(curve), f(x) (over a range
of x between a and b), that satisfies 2 conditions:
 iii. f(x) ≥ 0, for a ≤ x ≤ b
iv. Total area under curve = 1
1. Integral from a to b of f(x)dx = 1
f. The probability that x assumes a range of values is given by the area under the pdf over
range of values
g. Continuous distributions covered in chapter 8
i. Uniform distribution
ii. Exponential distribution
iii. Normal distribution
iv. Student t’s distribution
v. Chi-squared distribution
vi. F distribution
2. The uniform distribution
a. Definition
i. Also called the rectangular distribution
ii. A continuous probability distribution whose pdf is f(x) = 1/b-a, a
≤x≤b
b. Expected value and variance
i. M = E(x) = (a + b)/2
ii. σ² = v(x) = 1/12 (b-a)^2
iii. σ = s(x) = (b-a)/square root of 12
10-25-24
3. Finding values of Z
a. Find some value of Z, Za, for a given probability(area under the curve)
i. Example: find Z.025
1. A = 0.025
2. Z.025 = P(Z > 0.025) = 0.025
a. Go to Z-score table *appendix B
3. Z.025 = 1.96
4. Student’s T distribution
a. Continuous probability distribution
b. It’s pdf is given by:

i.
c. The student’s T distribution is defined by the degrees of freedom(V - greek letter nu)
d. Distribution is bell shaped around it’s mean = 0 (very similar to standard normal
distribution)
e. As V → infinity, the student’s T distribution → standard normal distribution
 f. The mean and variance are:
i. E(t) = 0
ii. V(t) = v/(v-2)

g.
h. Example:

i.
i. Finding T probabilities:
i. Table 4 of Appendix B can be used to infer about(find) T probabilities
1. Example:
a. V = 12
b. P(T > 2.5)
c. < 2.5 <
d. Go to table
e. 0.010 < P(T > 2.5) < 0.025
10-30-24
5. Chi-Squared Distribution
a. The parameter V is the degrees of freedom
b. Finding values of X^2:
i. Look for values of x^2 for a given probability
ii. x^2 values corresponding to an area under the curve to the right of the curve
iii. Table 5 of appendix b gives the values of X^2a, V
1. Example:
a. X^2.05, 20 = P(X^2 > X^2.05, 20) = 0.05
b. Use table 5
c. X^2.05, 20 = 31.4
c. Finding x^2 probabilities
i. Work backwards
1. Example: P(X^2 > 17), V = 23
 a. X^20.10 = 32 < 34 < 35.2 = X^2.05
b. 0.05 < P(X^2 > 34 < 0.10
c. X^2.90 = 14.8 < 17 < 32 = X^2.10
d. .10 < P(X^2 > 17) < .90
6. The F Distribution
a. Definition: another continuous probability distribution
b. It’s pdf is given by a very complicated function
c. Two parameters define this distribution, degrees of freedom.
i. V1 is the “numerator” degrees of freedom
ii. V2 is the “denominator” degrees of freedom
iii. *Mu 1 and Mu 2*
d. Distribution is skewed to the right

e.
11-01-24
7. Using normal distribution to approximate binomial distributions
a. X = binomial random variable
b. M = E(x) = np
c. SD: v(x) = npq = np(1-p)
d. Var: s(x) = square root of (npq)
e. If we have the following conditions:
i. N ≥ 20
ii. P ≥ 0.05
iii. Np ≥ 5
iv. Nq ≥ 5
f. ORRRR:
i. M +/- 30 = mp +/- 3(square root of npq) within (0,n)
g. ORRR:
i. P +/- 3(var)
ii. P +/- 3(square root of (pq/n)) within (0,1)
h. Then, we can use the normal distribution to approximate the binomial distribution
i. Both poisson and binomial distributions are discrete distributions
ii. Therefore, when we used the poisson approximation to binomial probabilities, we
did not need to make any continuity correction
iii. Now, when we use the normal distribution(continuous) to approximate binomial
distribution(discrete), we MUST make a continuity correction
1. Z = (x-m)/SD = (x-(np))/square root of npq
11-13-24
8. Sampling distribution of difference between two sample means
a. x̄ 1 - x̄ 2
b. Independent random samples must be drawn from each of two normal populations
i. If this condition is met, the sampling distribution of the difference between the
two sample means will be normally distributed
c. Describing the sampling distribution of x̄ 1 - x̄ 2
i. The expected value and variance of the sampling distribution of x̄ 1 - x̄ 2 are given
by:

1.
2. (another name for standard deviation is standard error)
ii. Shape → central limit theorem
1. Theorem #1: if the original population are normally distributed, then the
sampling distribution of x̄ 1 - x̄ 2 is also normally distributed for any
sample sizes
2. Theorem #2: if the original populations have unknown distributions, then
the sampling distribution of x̄ 1 - x̄ 2 is at least approximately normally
distributed IF the sample sizes of each population are larger than 30
(n1>30 and n2>30)
iii. We can compute Z(standard normal random variable) in this way:

1.
iv. QUESTION LIKE THIS ON NEXT EXAM
d. EXAMPLE:
 i.
ii. We want P(x̄ 1 > x̄ 2) = P(x̄ 1 - x̄ 2 > 0)
iii. x1 - N(62,000, 14,500)
iv. x2 - N(60,000, 18,300)
v. = by CLT, (x̄ 1 - x̄ 2) - N(2000, (square root of (14,500^2/50 + 18,300^2/60))
vi. Standard error = 3,128. 34
vii. (P(x̄ 1 - x̄ 2) - (Mx̄ 1 - x̄ 2))/(SD x1 - x2) > (0 - 2,000/3,128.34) = P(Z > -0.64)
1. = P(Z < 0.64)
viii. P(Z < 0.64) = 0.7389

CHAPTER 10 - Estimation
I. Introduction
A. Statistical inference - the process by which we acquire information and draw conclusions
about populations from samples
B. From chapters 7 & 8 - we looked at probability distributions, which made statements
about individual values of random variables. Took population and parameters and made
statements about individuals from the population.
C. From chapter 9 - we introduced sampling distributions and used those distributions to
make statements about values of statistics.
D. In this chapter - in realistic situations, parameters are unknown. We use the sampling
distributions to make inferences about unknown parameters.
1. Statistic → population parameters
2. Two types of inferences:
a) Estimation
b) Hypothesis testing notes start below
II. Estimation
A. Estimation definition: determine the approximate value(estimation) of an unknown
population parameter on the basis of a sample statistic.
B. Types of estimators:
1. Point estimator: draws inferences about a population by estimating the value of
the unknown parameter by using a single point.
2. Interval estimator: draws inferences about a population by estimating the value of
an unknown parameter using an interval.
a) The chance that a single value will reflect the population parameter is
virtually zero, hence why we prefer to use an interval estimator
 b) Going to build an interval around the point estimator
C. Qualities of estimators: (an estimator SHOULD have:)
1. Unbiasedness: an unbiased estimator of a population parameter is the estimator
whose expected value is the value of the population parameter
2. Consistency: an unbiased estimator of a population parameter is said to be
consistent if the difference between the estimator and the parameter grows small
as the sample size increases.
a) Standard error of estimator decreases as N increases
3. Relative efficiency: if there are two unbiased and consistent estimators for the
unknown population parameter, the estimator with the lower variance/standard
deviation is said to be relatively (more) efficient
a) X bar is the best estimator of a population mean µ
11-15-24
III. Estimating µ when σ is known
A. Assumptions
1. Random sample
2. σ is known
B. (1 - α )100% confidence interval estimator for

1.
2. The probability (1 - α ) is the confidence level, which is a measure of how
frequently the interval will actually include µ

3.
4. EXAMPLE: The Doll Computer Company makes its own computers and
delivers them directly to customers who order them via the internet. To achieve
its objective of speed, Doll makes each of its most popular computers and
transports them to warehouses from which it generally takes 1 day to deliver a
computer to the customer. To lower these costs the operations manager wants to
use an inventory model, he notes demand during lead time is normally distributed
and he needs to know the mean to compute the optimum inventory level. He
observes 25 lead time periods and records the demand during each period, the
manager would like a 95% confidence interval estimate of the mean demand
 during lead time. Assume that the manager knows that the standard deviation is
75 computers.
a) X = demand during lead time
b) n = 25
c) Σ = 75
d) 95% confidence interval estimate of mu:

(1)
e) In order to use out confidence interval estimator, we need the following
data:

(1)
C. Interpreting confidence interval
1. The CI is NOT about the interval computed, but it is about the process
2. If we were to build all possible CIs using all possible values of x bar, then (1 - α)
x 100% of those intervals will contain mu, and α x 100% of those intervals will
not contain mu
3. There is no way of knowing if the interval computed is one of the (1 - α ) x 100%
that contain my, or one of the α x 100% that does not contain mu
D. Effects of changing confidence level, σ and n
1. Confidence level
a) CL = (1 - α)
b) CL increases, α decreases → Zα/2 increases, → B increases *B is interval
width
2. Standard deviation
a) σ increases, B increases → interval is wider
3. Sample size(n)
a) n increases, B decreases → interval is narrower
E. Estimating the sample size 11-20-24
1. Sampling error: estimator - parameter
2. It is also called “error of estimation”
3. The bound on the error of estimation is:
 a)
b) B will be given
c) CL will be given
d) σ not always given

e) EXAMPLE:

11-15-24
REVIEW SESSION - CHAPTERS 7, 8 AND 9
1. Chapter 7 - discrete probability distributions
a. Discrete random variable - countable values
i. Ex: # of times heads come up when you flip a coin a certain number of times
b. Normal distribution, poisson distribution
2. Chapter 8 - continuous probability distributions
a. Continuous random variable - uncountable values
i. Ex: height, time, weight, distance
3. Chapter 9 - sampling distributions
a. Probability distributions of statistics
i. Ex: x bar, m, s^2
b. We know about the shape of these distributions and use the distributions to make
statements about the individual values of x bar, m and s^2
4. Problem 1: for some value z0 (mean of z is 0), the probability that a standard normal variable is
between z0 and 0 is 0.4798(or P(z0 < Z < 0)). What is the value of z0?
a. P(Z<0) - P(Z<z0)
b. = 0.5 - P(Z<z0) = 0.4798
c. P(Z < z0)= 0.0202
i. Go to z table and find value that would produce this area(table appendix b)
 ii.
iii. Corresponding z is -2.05 which is final answer
iv. z0 = -2.05
5. Problem 2: find the value of F.01, 9, 18
a. P(F > F.01, 9, 18 ) = .01
b. Go to table 6 of appendix B (look for F: A = .01)
i. Mu 1 = 9, mu 2 = 18 *numerator, denominator*
ii. Final answer F = 3.6
6. Problem 3: Let X be a binomial random variable with n = 60 and p = .04. Use the appropriate
poisson distribution to approximate P(X = 7)
a. Check to see if we can use poisson approximation (n must be more than 20 and p must be
less than .05)
i. N = 60 (60>20) yes
ii. P = .04 (.04 < .05) yes
iii. Therefore, we can use poisson approximation
b. Np = 60 x .04 = 2.4
c. P(X = 7) = P(X = 7, mu = 2.4)
d. Poisson approximation formula:

i.
e. (e^-2.4 x 2.4^7)/7! = .0083 = final answer
7. Problem 4: “a continuous random variable is one that can have only certain clearly separated
values resulting from a count of some item of interest”, true or false?
a. False, impossible to count, count be infinite values
8. Problem 5: A minister has to work every saturday on which there is a wedding, church records
show that the number of weddings that take place on saturdays has a poisson distribution with
mean 1. What is the probability that the minister will have to work this saturday?
a. X is poisson distributed
b. X = # of weddings
c. Mu = E(x) = 1 *expected value will always be given in poisson*
d. Looking for P(X>1, mu = 1) *greater than or equal to*
e. = 1 - P(X = 0)
f. Put in poisson formula, get 0.6321
 g.
9. Problem 6: the probability distribution of a continuous random variable is represented by a _____
that is a smooth curve
a. Pdf - probability density function
10. Problem 7: consider a random variable X with the following probability distribution:
a. X = -4, 0, 1, 2
b. P(X) = .2, .3, .4, .1
c. Find P(0 < or equal to X < or equal to 1)
d. = P(0) + P(1)

CHAPTER 11: INTRODUCTION TO HYPOTHESIS TESTING

I. Concepts of hypothesis testing
A. Statistical inference: process by which we gather information to draw inferences about
population on basis of samples
B. 2 types of statistical inferences:
a) Estimation (chapter 10)
b) Hypothesis testing (chapter 11)
(1) In hypothesis testing, the objective is to test value of a
population parameter using a sample
C. Non statistical hypothesis testing:
1. EXAMPLE: “innocent until proven guilty”:
a) Innocent = H0 = null hypothesis
b) Guilty = H1 = alternative hypothesis
2. Type I error = when we reject a true null hypothesis (jury convicts innocent
person)
a) P(Type I error) = α
3. Type II error = when we don’t reject a false null (jury does NOT convict a guilty
defendant)
a) P(Type II error) = B(beta)
4. Procedure:
a) Start with null and alternative hypotheses
b) Assume the null is true
c) Goal is to determine whether there is enough evidence to infer that the
alternative hypothesis is true
d) Two possible decisions:
(1) Conclude that there is enough evidence to support the alternative
hypothesis
(2) Conclude that there is not enough evidence to support the
alternative hypothesis
D. Elements of hypothesis testing: 7 elements
1. H0, null hypothesis
a) Status quo
b) Assumed to be true
c) It must contain “=” sign
2. H1 or Ha, alternative(research) hypothesis
a) Contradicts H0
3. Types of tests
a) Two tailed test
(1) H0: Mu = M0 (hypothesized value of Mu)
(2) H1: Mu does NOT = M0
b) Lower tailed test
(1) H0 : Mu ≥ M0
(2) H1: Mu < M0
c) Upper tailed test
(1) H0: Mu ≤ M0
(2) H1: Mu > M0
4. Assumptions/requirements
a) Ex: test about Mu
(1) Random sample
(2) CLT applies
5. Test statistic
a) Ex: test about Mu

b)
6. Rejection region
a) Region in which H0 will be rejected

b)

7. Computation of observed test statistic

a) n, x bar, σ → Z score
 8. Conclusion: If ZObserved falls in the rejection region, then reflect H0 at alpha level
of significance(in favor of H1). There is enough evidence to support H1, else we
fail to reject H0 at alpha level of significance, there is not enough evidence to
support H1. P-value of a test: the probability of observing a test statistic at least
as extreme as the one computed given that the null hypothesis is TRUE.

a)
b) Interpreting P-value: compare p-value with the selected level of
significance, α:
(1) If p-value < α, reject H0 in favor of H1
(2) If p-value > α, do not reject H0 in favor of H1
(3) 0<p<.01 → overwhelming evidence to support H1(high
significant)
(4) .01<p<.05 → strong evidence to support H1(significant)
(5) .05<p<.10 → weak evidence to support H1(not significant)
(6) p>.10 → no evidence to support H1(definitely not significant)
II. Test of hypothesis about mu, (sigma is known)
A. Elements of test
1. H0
2. H1
3. Assumptions/requirements:
a) Random sample
b) Ensure that CLT holds
4. Test statistic:

a)
5. Rejection region; upper tailed test
6. Computation of observed test statistic:
 a)
7. Conclusion: 2 types
a) P-value approach:
(1) P-value = P(Z > 2.46)
(2) = P(Z < -2.46) = 0.0069
(3) .0069 < α(.05) → reject H0 in favor of H1
(a) Overwhelming evidence
b) Rejection region approach:
(1) Since z = 2.46 > 1.645(Z.05), we reject H0 in favor of H1