Notes on Simple Random Sampling & Margin of Error (Ch. 3.3-3.4)

Simple Random Samples (SRS)

  • Definition

    • A simple random sample (SRS) of size nn is a sample chosen in such a way that every group of nn individuals in the population has an equal chance of being selected as the sample.
    • Equivalently, each sample of size nn has the same chance to be chosen, which implies that each member of the population has the same chance of being included in the sample.
  • How to choose an SRS with technology

    • Label: Give each individual in the population a distinct numerical label from 11 to NN, where NN is the number of individuals in the population.
    • Randomize: Use a random number generator to obtain nn different integers from 11 to NN.
    • Select: Choose the individuals that correspond to the randomly selected integers.
  • Without replacement vs. with replacement

    • When choosing an SRS, selections are made without replacement: once an individual is selected for the sample, that individual cannot be selected again.
    • Many random number generators sample numbers with replacement, so repeated numbers should be ignored when using technology to select an SRS.
  • Random number generators and options

    • Websites and tools mentioned: www.random.org, calculators, Siri (as RNG options).
    • Practical note: In SRS, ignore repeats if the RNG outputs the same label more than once.
  • Real-world prompt related to SRS (Landfill → Park problem)

    • Problem setup: When landfills run out of room, they are sometimes converted into parks. Does the presence of a landfill underneath affect the trees that grow in the park?
    • Data example: A park has 283283 trees. Researchers want to inspect a sample of n=20n = 20 trees.
    • Tasks:
    • (a) Describe how to select an SRS of n=20n = 20 trees using slips of paper.
    • (b) Describe how to select an SRS of n=20n = 20 trees using a random number generator.
  • Sampling Variability

    • Definition: The fact that different random samples of the same size from the same population produce different estimates is called sampling variability.
    • Consequence: Sample means vary, even though the samples are drawn from the same population.
    • Mitigation: We can reduce sampling variability by using a larger sample size, i.e., increasing nn reduces the variability of the sample statistic.
  • Median income in Virginia (illustrative problem)

    • Problem: The median household income in a random sample of 371371 Virginia households is \61{,}784\.
    • (a) Do you think that the median income of all Virginia households is equal to \61{,}784\? Explain.
    • (b) If the sample size were increased to 1000 households, what effect would this have on the estimated median income?
    • Educational takeaway: Larger samples tend to yield more precise estimates of the population median; increased sample size generally reduces sampling variability and stabilizes the estimate toward the population value.
  • Inference: drawing conclusions about a population from sample data

    • The process of drawing conclusions about a population on the basis of sample data is called inference because we infer information about the population from what we know about the sample.
    • Key point: Inference underpins estimation (point estimates and margins of error) and hypothesis testing.

Estimating a Margin of Error (MOE)

  • Definition

    • The margin of error of an estimate describes how far, at most, we expect the estimate to vary from the true population value.
  • Estimating MOE for a proportion (procedure)

    • To estimate the MOE for a sample proportion resulting from a random sample of size n:
    1. Simulate the distribution of the sample proportion for many random samples of size n assuming that the true population value is equal to the sample proportion from the original sample.
    2. Multiply the standard deviation of the simulated distribution by 2 to get the margin of error.
  • Interpretation notes for MOE

    • The MOE provides a range around the sample statistic within which the true population value is expected to lie, given the stated method and assumptions (e.g., randomness, independence, etc.).
    • When the MOE is computed via simulation, it reflects the variability under the assumed population parameter and the sampling process used in the simulation.
  • Example problem (MOE from a summary of simulations)

    • Given data (for a proportion-based MOE method):
    • Number of samples: 300
    • Mean: 0.42
    • Standard Deviation: 0.013
    • (a) Use the results to approximate the MOE.
    • (b) Interpret the MOE.
    • Calculation: MOE \approx 2 \times 0.013 = 0.026.
    • Interpretation: The true population proportion is expected to lie within ±0.026 of the observed sample proportion (0.42) under the stated model and assumptions.
  • MOE for mean (alternative context)

    • If the data summarize a mean instead of a proportion, the same 2×SD rule often serves as a rough MOE for the mean estimate when the sampling distribution is approximately normal or large-sample approximation holds.
    • Example from the notes:
    • Number of samples: 300
    • Mean: 4.778
    • Standard Deviation: 0.246
    • (a) Approximate MOE: MOE \approx 2 \times 0.246 = 0.492.
    • (b) Interpretation: The true population mean is expected to lie within ±0.492 of the sample mean (4.778) under the same assumptions.
  • Connections to foundational principles

    • The MOE quantifies the precision of an estimate and depends on sample size, variability, and the sampling method.
    • Larger samples (increasing n) generally reduce MOE by reducing sampling variability, improving reliability of inference.
    • The margin of error is typically paired with a confidence level in standard practice (e.g., a 95% CI), though the exact confidence level isn’t explicitly stated in the provided material. The MOE here is presented as a practical, simulation-based approximation.

Practical notes and real-world relevance

  • SRS is designed to yield unbiased and representative samples by giving every group of size n the same probability of selection.
  • When using RNGs, be mindful of replacement vs. non-replacement to avoid biased samples due to duplicate selections.
  • In practice, larger samples reduce sampling variability and yield more stable estimates (e.g., for medians, means, or proportions).
  • Margin of error provides a tangible measure of reliability for an estimate and guides decision-making in survey design and interpretation of results.

Quick reference: key formulas and terms

  • Simple Random Sample (SRS) definition: every group of n individuals has equal probability of being chosen.
  • Without replacement: once selected, individuals cannot be chosen again; with replacement: selections are independent and duplicates may occur (ignored for SRS).
  • Margin of Error (MOE) for a proportion via simulation:
    • Step 1: simulate distribution of sample proportion under true value equal to original sample proportion.
    • Step 2: MOE \approx 2 \times \mathrm{SD}_{\text{sim}}.
  • MOE for a mean (approximate):
    • MOE \approx 2 \times \mathrm{SD}$$ given the sampling distribution is approximately normal.

Notes on the sources and scope

  • These notes synthesize content from Statistics and Probability with Applications, 3rd Edition (Starnes & Tabor), covering Sections 3.3–3.4 (Simple Random Samples & Estimating a Margin of Error).
  • Emphasis on conceptual understanding, practical procedures, and interpretation alongside concrete numerical examples.