Scribd Logo
Upload

Welcome to Scribd!

  • 15 views

    STAT 5 Week 9 Notes Part 1

    Uploaded by

    Amber Margalit
    This document summarizes key concepts taught by TA Jacob including: 1) Single variable summary statistics like mean, median, and standard deviation that characterize aspects of data distribution. 2) Box models that represent populations and how random sampling relates to expected values and standard errors. 3) Statistical inference techniques like confidence intervals that allow estimating population parameters like the mean from random samples when the true population is unknown.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    STAT 5 Week 9 Notes Part 1

    Uploaded by

    Amber Margalit
    15 views3 pages
    This document summarizes key concepts taught by TA Jacob including: 1) Single variable summary statistics like mean, median, and standard deviation that characterize aspects of data distribution. 2) Box models that represent populations and how random sampling relates to expected values and standard errors. 3) Statistical inference techniques like confidence intervals that allow estimating population parameters like the mean from random samples when the true population is unknown.

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document summarizes key concepts taught by TA Jacob including: 1) Single variable summary statistics like mean, median, and standard deviation that characterize aspects of data distribution. 2) Box models that represent populations and how random sampling relates to expected values and standard errors. 3) Statistical inference techniques like confidence intervals that allow estimating population parameters like the mean from random samples when the true population is unknown.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    15 views3 pages

    STAT 5 Week 9 Notes Part 1

    Uploaded by

    Amber Margalit
    This document summarizes key concepts taught by TA Jacob including: 1) Single variable summary statistics like mean, median, and standard deviation that characterize aspects of data distribution. 2) Box models that represent populations and how random sampling relates to expected values and standard errors. 3) Statistical inference techniques like confidence intervals that allow estimating population parameters like the mean from random samples when the true population is unknown.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 3

    May 24th, 2022

    Taught by TA Jacob. Their slides are typed below.

    Review:
    Summary Statistics
    ● Idea: Characterize an aspect of the data in a single value
    ● Single Variable Summary Stats
    Mean: Characterizes center of the distribution
    Median: Also characterizes the center, robust to outliers
    Mode: Most common value(s)
    Standard Deviation: Characterizes spread of distribution around the mean
    Histograms
    ● Two Variable Summary Statistics
    Correlation Coefficient: characterizes linear relationship between variables
    Scatterplots
    Regression Line (can also be used to make predictions)

    Box Models:
    ● Box represents the overall population
    ● We take random draws from the box
    ● Can find the expected value and standard error

    Expected (new average) SE (New SD)

    Sum (sample size) x (box avg) 𝑠𝑎𝑚𝑝𝑙𝑒 𝑠𝑖𝑧𝑒 x (Box SD)

    Average Box average Box SD÷ 𝑠𝑎𝑚𝑝𝑙𝑒 𝑠𝑖𝑧𝑒

    Percentage Box average x 100% (Box SD÷ 𝑠𝑎𝑚𝑝𝑙𝑒 𝑠𝑖𝑧𝑒)x 100%

    Box Models Continued:


    Law of Large Numbers:
    ● Expected value of average/percentage -› box average
    Central Limit Theorem
    ● Sums/Average of draws from the box -› N (Expected Value, Standard Error)
    ● Assumes independent draws
    ● Number of draws needed depends on probability distribution of the box
    ● General rule of thumb: the more the better
    ● Only true for sums/averages, not products

    Statistical Inference
    Thus far, we have assumed that we know what's inside the box
    ● Then we can describe what happens for random samples from the box (CLT, LLN)
    But what if we don't know the contents of the box?
    ● This is the core problem of statistics
    ● We have a random sample from the box, and want to guess what is inside
    ● In other words, can we find the box average/standard deviation from the sample?
    ● Can we quantify how uncertain we are?

    Estimating Averages/Percentages
    ● Sample Mean/Percentage = Population Mean/Percentage + Chance Error
    ● Idea: We can approximate the population mean by the sample mean
    ● We know from the Law of Large Numbers that this estimate will get better as the sample
    size gets bigger
    ● But how certain are we?

    Bootstrapping
    ● To quantify our uncertainty, we need the SE
    ● To get the standard error, we need the box SD
    ○ But we don't have the box standard deviation!
    ● Idea: Use the SD of the sample to approximate the SD of the box
    ● This is called the "Bootstrap”

    Confidence Intervals:
    ● We want a concise summary of how accurate our estimate is
    ● We can use the SD, and Normal approximation
    ● We know that 68% of the time, the true mean will be within 1 SE of the sample mean
    ● We know 95% of the time, the true mean will be within 2 SE of the sample mean
    (actually 1.96 SE to be exact)
    ● We know that 99% of the time, the true mean will be within 2.58 SE of the sample mean

    Confidence Intervals Continued


    Sample Mean ± 1.96 SE
    ● First find sample average and sample SD
    ● Next compute the SE
    ● Then find the lower and upper bounds of the Confidence Interval

    Important reminders for SEs:


    SE for sums= # 𝑜𝑓 𝑑𝑟𝑎𝑤𝑠 x SD of Box
    𝑆𝐸 𝑓𝑜𝑟 𝑠𝑢𝑚
    SE for average= # 𝑜𝑓 𝑑𝑟𝑎𝑤𝑠
    SE for count= SE for sum, from a 0-1 box
    𝑆𝐸 𝑓𝑜𝑟 𝑐𝑜𝑢𝑛𝑡
    SE for percent= # 𝑜𝑓 𝑑𝑟𝑎𝑤𝑠
    x 100%

    Interpreting Confidence Intervals:


    ● The randomness is in the Sample, not the underlying parameter (ie. mean/percentage)
    ● The true parameter is treated as fixed, the randomness comes from the change error in
    the sampling procedure
    ● We build the confidence interval from the sample, which means that the confidence
    interval is random
    ● The interval for any two samples will be different

    Interpreting Confidence Intervals:


    ● A common incorrect interpretation:
    ○ The true mean will fall inside the 95% Confidence interval with 95% probability
    (Wrong!)
    ○ The true mean is fixed. It's either inside or outside the Confidence Interval

    ● The correct Interpretation:


    ○ Confidence intervals are random objects generated by the sample
    ○ For each sample, you'll have a difference confidence interval
    ○ If you take 100 samples, the 95% of the confidence intervals will contain the true
    mean

    You might also like