Normal Distributions, Sampling Distributions, and the Central Limit Theorem

Course Administration, Schedule, and Classroom Policies

• Mobile devices must be silenced and stored away from desks during all class sessions to eliminate distractions.

• Students entering the lecture hall are required to fill seating toward the front and avoid sitting in the back three rows.

• Course schedule progression:

  • Current academic calendar point: Week 7.
  • Upcoming schedule: Week 8 precedes the fall break.
  • Material covering the normal distribution must be completed prior to the fall break.
  • Chapter 7 (Estimation) commences on Monday and Wednesday following the return from break.
  • Week 9 schedule:
  • Monday: Dedicated review day for assignments and the practice exam; students do not report to the physical classroom.
  • Wednesday: Administration of the formal examination.

• Subsequent curriculum mapping:

  • Chapter 7 covers statistical estimation.
  • Chapters 8 and 9 cover hypothesis testing.
  • Calculations in Chapters 7, 8, and 9 build directly upon normal distribution techniques (finding areas to the left of $z$, to the right of $z$, and between two $z$-values).

• Supplemental learning resources:

  • Dedicated Word document uploaded on Canvas providing organized video links for analytical review.
  • Direct links provided within the syllabus tab.
  • Recommended supplementary searching on YouTube for reinforced problem walk-throughs.

• Academic integrity and attendance policies:

  • Signing attendance sheets on behalf of absent peers constitutes academic dishonesty.
  • Discovery of proxy attendance signing results in a mandatory post-class conference and an immediate deduction of $10\%$ from the offending student's final course grade.

Normal Probability Distribution for Individual Data Values

• The standard normal distribution is characterized by a mean of $\mu = 0$ and a standard deviation of $\sigma = 1$.

• For any continuous variable $x$ that is normally distributed with mean $\mu$ and standard deviation $\sigma$, individual raw scores are standardized into a $z$-score via the formula:   $z = \frac{x - \mu}{\sigma}$

• The $z$-score represents the directed distance of a data point from the mean measured in units of standard deviation:

  • A score below the mean yields a negative $z$-score.
  • The center of the distribution corresponds to $z = 0$.
  • A score above the mean yields a positive $z$-score.

• Computational probability regions under the normal curve:

  • Area to the left: cumulative probability below a specified value ($P(X < x)$).
  • Area to the right: cumulative probability above a specified value ($P(X > x)$).
  • Area in between: probability contained between two boundaries ($P(x_1 < X < x_2)$).

• Use of computational software (StatCrunch):

  • Navigation path: Stat > Calculators > Normal.
  • Direct input allows entering the population mean $\mu$ and standard deviation $\sigma$ directly into the calculator rather than manually computing intermediate $z$-scores.
  • The area under the entire normal curve equals strictly $1$, meaning probability inputs and outputs must always remain between $0$ and $1$.

Example: Household Newspaper Recycling

• Problem scenario: Monthly newspaper waste generated by American households is approximately normally distributed with an average of $28\,\text{pounds}$ and a standard deviation of $2\,\text{pounds}$. A household is selected at random. The objective is to determine the probability that the household generates between $27$ and $31\,\text{pounds}$ per month.

• Identification of parameters:

  • Population mean: $\mu = 28\,\text{pounds}$
  • Population standard deviation: $\sigma = 2\,\text{pounds}$
  • Lower raw data bound: $x_1 = 27\,\text{pounds}$
  • Upper raw data bound: $x_2 = 31\,\text{pounds}$

• Standardization of bounds:

  • Lower $z$-score: $z_1 = \frac{27 - 28}{2} = -0.5$
  • Upper $z$-score: $z_2 = \frac{31 - 28}{2} = 1.5$

• Software execution steps in StatCrunch:

  • Select the Between mode option in the normal calculator.
  • Input Mean: $28$.
  • Input Standard Deviation: $2$.
  • Set the lower inequality box to $27$ and the upper inequality box to $31$.
  • Select Compute to generate the probability value directly.

Practical Applications: Municipal Resource Management

• Statistical distributions serve as decision-making tools for public administration and management (e.g., municipal leadership in Bloomington Town).

• Applications of household waste probability data:

  • Operational budgeting: Forecasting municipal expenditures required for waste processing and landfill management based on expected tonnage.
  • Utility pricing: Formulating appropriate disposal fees and surcharge structures for residential households.
  • Capital allocation: Calculating the optimal size and composition of the city's collection truck fleet.
  • Route scheduling: Establishing the logistical frequency of residential trash pickup schedules across districts.

Percentiles and Inverse Normal Calculations

• Percentiles represent cumulative relative frequencies that increase strictly from left to right across the distribution.

• A percentile ranking indicates the area under the standard normal curve lying entirely to the left of the corresponding score (e.g., the $65\text{th}$ percentile represents an area of $0.65$ to the left; the $95\text{th}$ percentile represents an area of $0.95$ to the left).

• When given a target probability or percentile, the corresponding standard score $z$ or raw value $x$ is solved in reverse:

  • To solve for $x$ manually:     $x = \mu + z \times \sigma$
  • In StatCrunch, enter the mean $\mu$, standard deviation $\sigma$, specify the direction of the inequality, enter the desired probability (e.g., $0.95$) in the right-hand output box, and click Compute to determine $x$.

Example: Aircraft Cabin Ceiling Height Design

• Problem scenario: Aircraft cabin designers require a ceiling height sufficient to permit $95\%$ of men to stand upright without making contact with the ceiling. Men's heights are normally distributed with a mean of $\mu = 69.5\,\text{inches}$ and a standard deviation of $\sigma = 2.4\,\text{inches}$.

• Parameter configuration:

  • Mean: $\mu = 69.5\,\text{inches}$
  • Standard deviation: $\sigma = 2.4\,\text{inches}$
  • Target cumulative non-exceedance probability: $0.95$

• Mathematical derivation:

  • The $z$-score corresponding to a cumulative left-tail area of $0.95$ is approximately $z = 1.645$.
  • Linear transformation equation:     $x = 69.5 + 1.645 \times 2.4$
  • Calculation result: $x \approx 73.448\,\text{inches}$, rounding to approximately $73\,\text{inches}$.

• Commercial and risk implications:

  • Setting clearance at $73\,\text{inches}$ accommodates $95\%$ of customers while excluding the top $5\%$ of tallest men.
  • Inadequate clearance exposes airlines to personal injury lawsuits from passengers bumping their heads.
  • Increasing clearance beyond $73\,\text{inches}$ expands comfort for passengers taller than six feet one inch, preventing the loss of that customer demographic.

Critical Values in Statistical Inference

• A critical value is a boundary $z$-score that delineates significantly low or significantly high outcomes from non-significant (typical) outcomes.

• Connection to significance concepts:

  • Typical or expected values occupy the central region of the distribution.
  • Values in the extreme tails represent statistically significant outcomes.
  • The critical value sits precisely at the borderline dividing the typical region from the significant regions.

• Structural forms of critical values in hypothesis testing:

  • Right-tailed test: Utilizes a single positive critical value bounding the upper critical rejection region.
  • Left-tailed test: Utilizes a single negative critical value bounding the lower critical rejection region.
  • Two-tailed test: Utilizes two symmetric critical values ($\pm z$) bounding the lower and upper critical regions simultaneously.

Fundamentals of Sampling Distributions

• Definition of a sample statistic: A numerical descriptive measure calculated from sample data (such as sample mean $\bar{x}$, sample variance $s^2$, or sample standard deviation $s$).

• Sampling distribution of a statistic: The probability distribution formed by plotting all possible values of that statistic obtained from all possible random samples of equal size $n$ drawn from the same underlying population.

• Discrete uniform population demonstration:

  • Population test score values: $2, 4, 6, 8$ ($N = 4$).
  • Population mean calculation:     $\mu = \frac{2 + 4 + 6 + 8}{4} = 5$
  • Population standard deviation calculation:     $\sigma = \sqrt{5} \approx 2.236$
  • Complete set of all samples of size $n = 2$ drawn with replacement ($4^2 = 16$ total samples):
  • Sample $(2, 2)$: mean = $2$ (frequency: $1/16$)
  • Samples $(2, 4), (4, 2)$: mean = $3$ (frequency: $2/16$)
  • Samples $(2, 6), (4, 4), (6, 2)$: mean = $4$ (frequency: $3/16$)
  • Samples $(2, 8), (4, 6), (6, 4), (8, 2)$: mean = $5$ (frequency: $4/16$)
  • Samples $(4, 8), (6, 6), (8, 4)$: mean = $6$ (frequency: $3/16$)
  • Samples $(6, 8), (8, 6)$: mean = $7$ (frequency: $2/16$)
  • Sample $(8, 8)$: mean = $8$ (frequency: $1/16$)
  • Structural properties observed:
  • Although the parent population is uniform, the resulting distribution of sample means ($\bar{x}$) forms a symmetric, bell-shaped distribution centered at $5$.
  • The mean of all sample means equals the population mean ($\mu_{\bar{x}} = \mu = 5$).

The Central Limit Theorem (CLT)

• Fundamental statement of the Central Limit Theorem: As the sample size $n$ increases, the sampling distribution of the sample mean ($\bar{x}$) approaches a normal distribution, regardless of the shape of the original population distribution.

• Core properties of the sampling distribution of sample means:

  • The mean of the sample means is identical to the population mean:     $\mu_{\bar{x}} = \mu$
  • The standard deviation of the sample means, formally called the standard error of the mean, is equal to the population standard deviation divided by the square root of the sample size:     $\sigma_{\bar{x}} = \frac{\sigma}{\sqrt{n}}$
  • The standard error $\sigma_{\bar{x}}$ is strictly smaller than the population standard deviation $\sigma$ for all sample sizes $n > 1$.

• Operational rules for normality:

  • If the original population variable is normally distributed, the sampling distribution of $\bar{x}$ is normally distributed for any sample size $n$.
  • If the original population variable is not normally distributed (or of unknown distribution), a sample size strictly greater than thirty ($n > 30$) is required to ensure that the sampling distribution of $\bar{x}$ is approximately normal.

• Standardization formula for sample means:   $z = \frac{\bar{x} - \mu}{\frac{\sigma}{\sqrt{n}}}$

• Software implementation for standard error in StatCrunch:

  • In the standard deviation field, input sigma/sqrt(n) (e.g., 3/sqrt(20)).

Example: Children's Television Viewing Habits

• Problem scenario: A study states that children between the ages of two and five watch an average of $25\,\text{hours}$ of television per week with a standard deviation of $3\,\text{hours}$. The variable is normally distributed. If a random sample of $20$ children between the ages of two and five is selected, find the probability that their mean viewing time exceeds $26.3\,\text{hours}$.

• Distinguishing parameters from contextual background:

  • The child age span "two to five years" is descriptive demographic context and is omitted from calculation formulas.
  • Population mean: $\mu = 25\,\text{hours}$
  • Population standard deviation: $\sigma = 3\,\text{hours}$
  • Sample size: $n = 20$
  • Sample mean threshold: $\bar{x} = 26.3\,\text{hours}$

• Analytical standardization:   $z = \frac{26.3 - 25}{\frac{3}{\sqrt{20}}}$

• Software execution in StatCrunch:

  • Input Mean: $25$.
  • Input Standard Deviation: 3/sqrt(20).
  • Select the greater-than-or-equal inequality sign ($\ge$) for $26.3$.
  • Compute to yield the right-tail probability ($P(\bar{x} > 26.3)$).

• Comparative probability variations:

  • Left-tail scenario: Probability of the sample mean being less than $26\,\text{hours}$ ($P(\bar{x} < 26)$) requires setting the inequality direction to $\le 26$.
  • Between scenario: Probability of the sample mean falling between $23$ and $26\,\text{hours}$ ($P(23 < \bar{x} < 26)$) requires selecting the Between mode with limits $23$ and $26$.

• Societal applications: Data utilized by pediatric researchers, parental organizations, and legal advocacy groups evaluating screen time exposure limits.

Comparative Analysis: Individual Values Versus Sample Means

• Determining which formula to apply depends on whether the question concerns an individual data observation or a sample mean of a group:

  • Individual observation ($n = 1$): Standardization uses the population standard deviation:     $z = \frac{x - \mu}{\sigma}$
  • Sample mean ($n > 1$): Standardization uses the standard error of the mean:     $z = \frac{\bar{x} - \mu}{\frac{\sigma}{\sqrt{n}}}$

• Problem comparison using adult male weights:

  • Given population parameters: $\mu = 172\,\text{pounds}$, $\sigma = 29\,\text{pounds}$, with normal distribution.

• Part A: Probability that a single randomly selected man has a weight greater than $167\,\text{pounds}$.

  • Applies individual value formula:     $z = \frac{167 - 172}{29}$
  • Computes the area under the curve to the right of $x = 167\,\text{pounds}$.

• Part B: Probability that a random sample of $12$ men has a mean weight greater than $167\,\text{pounds}$.

  • Parameter additions: $n = 12$, $\bar{x} = 167\,\text{pounds}$.
  • Applies sample mean formula:     $z = \frac{167 - 172}{\frac{29}{\sqrt{12}}}$
  • Because the standard error ($\frac{29}{\sqrt{12}}$) is smaller than the standard deviation ($29$), the resulting $z$-score has a larger absolute magnitude, significantly altering the tail probability.

Example: Registered Boat Ages

• Problem scenario: The average age of boats registered in the United States is $96\,\text{months}$. A random sample of $36$ boats is selected. Find the probability that the sample mean age of these boats is between $90$ and $100\,\text{months}$.

• Evaluation of normality condition:

  • Even if the population distribution of boat ages is not specified as normal, the sample size is $n = 36$.
  • Because $n = 36 > 30$, the Central Limit Theorem satisfies the requirement for the sampling distribution of sample means to be approximately normally distributed.

• Parameter values:

  • Population mean: $\mu = 96\,\text{months}$
  • Sample size: $n = 36$
  • Lower sample mean limit: $\bar{x}_1 = 90\,\text{months}$
  • Upper sample mean limit: $\bar{x}_2 = 100\,\text{months}$

• Software calculation structure in StatCrunch:

  • Navigation: Stat > Calculators > Normal, click Between.
  • Enter Mean: $96$.
  • Enter Standard Deviation: calculated standard error using sigma/sqrt(36).
  • Enter interval bounds: $90$ and $100$.
  • Select Compute to determine the probability that the sample mean falls within that range.

Questions and Discussion

• Discussion prompt: How can municipal executives utilize statistical waste generation distributions to establish public works policies?

  • Student response (Avan): Utilizing percentages of waste generation to determine whether to increase or decrease operational capacity and adjust citizen waste fees.
  • Student response (Mark): Evaluating revenue generation by charging households higher rates for excess waste disposal beyond standard deviations.
  • Student response (Smith): Determining operational fleet requirements (number of garbage trucks) and establishing route collection frequencies.

• Discussion prompt: In the children's television study, what role do the numbers "two and five" play in setting up the calculation?

  • Student clarification: The age range of two to five years identifies the sample population group but is not a numerical parameter included in mathematical calculations.

• Clarification on distinguishing $\mu$ from $\bar{x}$ in problem statements:

  • When an average describes the entire population (e.g., "all women," "all children," "average age of boats in the US"), it represents the population mean $\mu$.
  • When an average describes a subset extracted from the population (e.g., a sample of $20$ children, $12$ men, or $36$ boats), it represents the sample mean $\bar{x}$.