Probability Theory, Discrete and Continuous Distributions, and Standard Normal Calculations

Foundations of Probability Theory

• Definition of Probability:

  • Probability is the mathematical study of randomness and random processes.

  • Randomness does not imply that all outcomes are equally likely; rather, individual outcomes carry distinct probabilities of occurrence.

  • In real-world scenarios where exact forecasts are impossible, probabilities express relative likelihoods based on known historical frequencies (e.g., forecasting whether it will rain on a specific day in Greensboro next September, where non-rainy days outnumber rainy days).

• Numerical Representation:

  • Probabilities are expressed as proportions or decimals ranging strictly between $0$ and $1$:

  • $0$: Indicates absolute impossibility (the event definitely will not occur).

  • $1$: Indicates absolute certainty (the event definitely will occur).

  • $0.5$: Indicates equal likelihood ($50\%$ chance of occurrence).

  • While colloquial communication uses percentages, formal mathematical probability strictly utilizes proportions and decimal values.

• Classical Formalization of Probability:

  • For a well-defined outcome $A$:     $P(A) = \frac{\text{Number of outcomes corresponding to } A}{\text{Total number of possible outcomes}}$

• Standard Playing Card Deck Applications:

  • Standard deck composition:

  • Total cards: $52$

  • Suits: $4$ (Diamonds, Clubs, Hearts, Spades)

  • Cards per suit: $13$ (Ace, numbers $2$ through $10$, and $3$ face cards: Jack, Queen, King)

  • Probability of drawing an Ace:

  • Total possible outcomes: $52$

  • Favorable outcomes ($A = \text{Ace}$): $4$     $P(\text{Ace}) = \frac{4}{52} \approx 0.0769 \approx 0.077 \quad (7.7\%$\text{ chance})\n * **Probability of drawing a Spade**:\n * Favorable outcomes (A = \text{Spade}$):$13\n    P(\text{Spade}) = \frac{13}{52} = \frac{1}{4} = 0.25 \quad (25\%$\text{ chance})$

  • Probability of drawing the Jack of Diamonds:

  • Favorable outcomes ($A = \text{Jack of Diamonds}$): $1$     $P(\text{Jack of Diamonds}) = \frac{1}{52} \approx 0.0192 \quad (1.92\%$\text{ chance})\n\n* **Frequentist and Long-Run Relative Frequency Interpretation**:\n * On any single draw, an outcome yields a binary reality (100\%$or$0\% occurrence once drawn).\n * In the long run, across an infinite or large repetition of draws under identical conditions with replacement and reshuffling, the proportion of draws yielding outcome A$converges to its theoretical probability ($7.7\% for drawing an Ace).\n\n* **Sampling Methodologies and Assumptions**:\n * Probability values strictly depend on the sampling approach.\n * **Random Sampling Assumption**: Assumes every possible outcome or card has an equal probability of selection (e.g., well-shuffled deck without looking).\n * **Application to Psychological Research**:\n * Statistical calculations assume that study data represent a random sample drawn from the broader target population.\n * In practice, obtaining a truly random sample of clinical populations (e.g., all depressed individuals globally) is practically impossible.\n * Determining whether research conclusions generalizability holds depends on whether the sample adequately represents the target population—a research methodology question rather than a mathematical/statistical calculation.\n * **Two Critical Roles of Randomness in Studies**:\n 1. *Random Sampling*: How participants are selected from a population.\n 2. *Random Assignment*: How participants are allocated to conditions (e.g., placebo vs. active treatment) in experimental designs.\n\n# Properties and Requirements of Valid Probability Distributions\n\n* **Bowl of Marbles Example**:\n * Contents: 10$blue marbles,$20$red marbles,$30$green marbles ($60 total marbles).\n * **Probability of drawing a Red marble**:\n * Favorable outcomes: 20\n * Total outcomes: 10 + 20 + 30 = 60\n    P(\text{Red}) = \frac{20}{60} = \frac{1}{3} \approx 0.3333…\n * Expressed rounded to two decimal places: 0.33\n\n* **Definition of a Probability Distribution**:\n * A probability distribution consists of the set of all possible outcomes of a random process alongside the specific probability associated with each outcome.\n\n* **Core Requirements for Validity**:\n 1. **Exhaustive**: The listed outcomes must cover every single possible result. No valid outcome can be omitted.\n 2. **Mutually Exclusive**: Outcomes cannot overlap; no two listed outcomes can occur simultaneously on a single trial.\n 3. **Summation to Unity**: Because outcomes are exhaustive and mutually exclusive, the sum of all individual probabilities across the distribution must equal exactly 1$($100\%):\n     \sum P(X) = 1\n\n* **Random Draw Modeling**:\n * Outcomes are analytically treated as unpredictable draws from the probability distribution.\n\n* **Comparison to Empirical Sample Distributions**:\n * *Frequency Tables and Histograms*: Summarize observed counts or proportions from a finite sample.\n * *Probability Distributions*: Provide theoretical long-run expected proportions. Repeated sampling (e.g., a million trials) generates sample histograms that mirror the theoretical probability distribution shape.\n\n# The Binomial Distribution and Its Parameters\n\n* **Definition of the Binomial Distribution**:\n * A discrete probability distribution describing a process repeated a fixed number of times (n trials), where each trial yields exactly one of two possible binary outcomes (often designated as Success/Failure or Heads/Tails).\n\n* **Coin Flipping Example (n = 3$trials, Fair Coin$p = 0.5)**:\n * Variable X$: The number of heads obtained ($X \in {0, 1, 2, 3}).\n * **Mathematical Breakdown of Probabilities**:\n * P(\text{3 Heads}) = 0.5 \times 0.5 \times 0.5 = 0.125\n * P(\text{1 Head})$: Occurs via three distinct sequence paths (HTT, THT, TTH). Each sequence has a probability of$0.5 \times 0.5 \times 0.5 = 0.125. Summing these paths yields:\n      P(\text{1 Head}) = 3 \times 0.125 = 0.375\n * **Complete Probability Distribution Table for n = 3, p = 0.5**:\n * P(X = 0) = 0.125\n * P(X = 1) = 0.375\n * P(X = 2) = 0.375\n * P(X = 3) = 0.125\n * Total sum: 0.125 + 0.375 + 0.375 + 0.125 = 1.000\n\n* **Defining Parameters of the Binomial Distribution**:\n * Every specific binomial distribution is fully defined by two parameters:\n 1. n: The total number of independent trials.\n 2. p: The constant probability of success on a single trial.\n\n* **Parameter Shifts and Distribution Shape Changes**:\n * Example with weighted coin (n = 3, p = 0.6):\n * P(X = 0)$becomes less likely ($0.064).\n * P(X = 3)$becomes more likely ($0.216).\n * P(X = 2)$($0.432$) becomes more likely than$P(X = 1)$($0.288).\n * The distribution shifts its central tendency (mode becomes 2) and becomes negatively skewed.\n\n* **Distribution Statistics**:\n * Probability distributions possess formal parameters for central tendency (mean, median, mode), variability (variance, standard deviation), and skewness, analogous to observed empirical sample statistics.\n\n# Calculating Binomial Probabilities\n\n* **Computational Inputs for Binomial Calculations**:\n * Parameter inputs required for statistical tools (e.g., StatTrek Binomial Calculator):\n * Probability of success on a single trial (p).\n * Number of trials (n).\n * Number of successes evaluated (x).\n * Output categories provided:\n * Exact single outcome probability: P(X = x)\n * Cumulative lower probabilities: P(X < x)$,$P(X \le x)\n * Cumulative upper probabilities: P(X > x)$,$P(X \ge x)\n\n* **Applied Example 1: Baseball Series**:\n * Problem: An average baseball team (p = 0.5$) plays a$4$-game series ($n = 4$). What is the probability they win at least$3$games ($X \ge 3) to qualify for the playoffs?\n * Parameter settings: n = 4$,$p = 0.5$,$x = 3\n * Evaluated expression: P(X \ge 3)\n * Calculation result: 0.3125$($31.25\% probability of making the playoffs).\n\n* **Applied Example 2: Greensboro July Rainfall**:\n * Problem: In Greensboro, it rains on 28\%$of days in July ($p = 0.28$). If a friend visits for$3$days ($n = 3$), what is the probability that it does not rain during the visit ($X = 0)?\n * Parameter settings: n = 3$,$p = 0.28$,$x = 0\n * Evaluated expression: P(X = 0)\n * Result computation: (1 - 0.28)^3 = (0.72)^3 = 0.373248 \quad (37.32\% probability of zero rain).\n\n* **Applied Example 3: Defective Lightbulbs**:\n * Problem: The defect rate for lightbulbs is 1\%$($p = 0.01$). A pack contains$10$lightbulbs ($n = 10$). What is the probability that$1$or more lightbulbs are defective ($X \ge 1)?\n * Parameter settings: n = 10$,$p = 0.01$,$x = 1\n * Evaluated expression: P(X \ge 1)\n * Calculation result: 0.09562$(approximately$9.56\% probability).\n\n# Continuous Probability Distributions and the Normal Distribution\n\n* **Discrete vs. Continuous Distributions**:\n * **Discrete Probability Distributions**: Model discrete variables with finite or countable outcomes (e.g., coin flips). Non-zero probabilities are assigned to distinct discrete numbers.\n * **Continuous Probability Distributions**: Model continuous variables with infinitely fine gradations (e.g., height, weight, time, speed).\n\n* **The Zero-Probability Point Property of Continuous Variables**:\n * For a continuous variable, the probability of drawing an exact value defined to infinite precision is precisely zero:\n    P(X = c) = 0\n * *Explanation*: Selecting a height of exactly 69.000000…$inches (out to infinite decimal places) yields a probability of zero. Probabilities for continuous variables can only be evaluated across intervals/ranges of values (e.g., height between$68.5$and$69.5 inches).\n\n* **Characteristics of the Normal Distribution (Bell Curve)**:\n * The most prominent continuous probability distribution in statistical theory.\n * **Bell-shaped Density**: Maximal probability density occurs at the mean; probability density smoothly decreases as distance from the mean increases.\n * **Symmetry**: Completely symmetric around the mean with no skew (\text{Mean} = \text{Median} = \text{Mode}).\n * **Low Outlier Frequency**: Extreme observations drop off rapidly. Values beyond 3 standard deviations from the mean are extremely rare.\n\n* **Parameters of the Normal Distribution**:\n * Defined by two parameters:\n 1. Mean (\mu): Determines distribution location/center.\n 2. Variance (\sigma^2$) or Standard Deviation ($\sigma): Determines distribution width/spread.\n\n* **The Standard Normal Distribution**:\n * A specific normal distribution standardized to have a mean of zero and a standard deviation of one:\n    \mu = 0\n    \sigma = 1\n * Maps directly to Z-score units:\n    Z = \frac{X - \mu}{\sigma}\n * Allows standardized probability comparisons across diverse continuous variables.\n\n# The Standard Normal Distribution and the Empirical Rule\n\n* **The Empirical Rule (68-95-99.7 Rule)**:\n * **\pm 1\sigma$Range**:$68.26\%$(approx.$68\%$) of observations lie within one standard deviation of the mean ($Z \in [-1, 1]).\n * **\pm 2\sigma$Range**:$95.44\%$(approx.$95\%$) of observations lie within two standard deviations of the mean ($Z \in [-2, 2]).\n * **\pm 3\sigma$Range**:$99.73\%$(approx.$99.7\%$) of observations lie within three standard deviations of the mean ($Z \in [-3, 3]).\n\n* **Standard Normal Region Proportions**:\n * Area between Z = 0$and$Z = 1$:$34.13\%\n * Area between Z = 1$and$Z = 2$:$13.59\%\n * Area above Z = 2$:$2.28\%\n * Due to symmetry, exact identical percentages apply to negative Z-scores:\n * Area between Z = -1$and$Z = 0$:$34.13\%\n * Area between Z = -2$and$Z = -1$:$13.59\%\n * Area below Z = -2$:$2.28\%\n\n* **Worked Standard Normal Examples**:\n\n * **Example 1: Exam Grade Distribution**:\n * Conditions: Grades normally distributed with \mu = 80$and$\sigma = 10.\n * Question: What percentage of students earn a C grade (score between 70$and$80)?\n * Z-score calculations:\n      Z_{70} = \frac{70 - 80}{10} = \frac{-10}{10} = -1.0\n      Z_{80} = \frac{80 - 80}{10} = \frac{0}{10} = 0.0\n * Evaluation: The region between Z = -1.0$and$Z = 0.0$contains$34.13\%$(approx.$34\%$) of the population.

  • Example 2: Test Scores above $38$:

  • Conditions: Population test scores normally distributed with $\mu = 30$ and $\sigma = 8$.

  • Question: What percentage of scores drop above $38$?

  • $Z$-score calculation:       $Z_{38} = \frac{38 - 30}{8} = \frac{8}{8} = 1.0$

  • Evaluation via Empirical Rule:

    • Region between $Z = -1$ and $Z = 1$ captures $68\%$ of observations.

    • Total area outside this range is $100\% - 68\% = 32\%$.

    • By symmetry, upper tail area ($Z > 1.0$) is half of $32\%$:         $\frac{32\%}{2} = 16\%$

    • Exact breakdown sum: $13.59\% + 2.28\% = 15.87\% \approx 16\%$

  • Example 3: Non-Integer $Z$-Score Bounds ($X > 42$):

  • Conditions: Population test scores with $\mu = 30$ and $\sigma = 8$.

  • Question: What percentage of scores fall above $42$?

  • $Z$-score calculation:       $Z_{42} = \frac{42 - 30}{8} = \frac{12}{8} = 1.5$

  • Deductive Estimation:

    • Area above $Z = 1.0$ is $16\%$.

    • Area above $Z = 2.0$ is $2.5\%$ (calculated as $\frac{100\% - 95\%}{2} = 2.5\%$

    • Because area decreases non-linearly across the interval, the region above $Z = 1.5$ lies between $2.5\%$ and $16\%$ (roughly estimated around $6\%$ to $7\%$).

  • Rule Limitations: The Empirical Rule provides rapid rough estimates, but precise continuous probabilities for fractional $Z$-scores require standard normal distribution lookup tables or statistical computer software.