PIA2012 - Social Research Methods
Standard Errors
Le Bao
Department of Public and International Affairs
2024-10-18
BAO, Le · PIA2012 · https://baole.io/ 1
Sampling Distribution
Sampling Distribution: If we took many random samples from a population,
computed the mean for each, and plotted these means, they would form a
sampling distribution.
The standard deviation of this sampling distribution is the standard error.
Example: If we measure the test scores of 100 randomly chosen students
from the university multiple times, the average test score of each sample
may slightly differ. The standard error measures how much these sample
means differ from the true population mean (e.g., the actual mean score of
all students).
BAO, Le · PIA2012 · https://baole.io/ 2
Expected Value
Expected Value: The mean of the sampling distribution of the mean is called
the expected value of the mean.
It represents the population mean (μ), meaning that, on average, the
sample mean will equal the population mean if we take an infinite
number of random samples.
Example: Suppose the average score on a standardized test for all
students in a school district is 75. This 75 is the expected value of the
sample mean because, over many random samples of students, the
average score should approach 75.
BAO, Le · PIA2012 · https://baole.io/ 3
Standard Error
The standard error (SE) is the standard deviation of the sampling
distribution of a statistic (e.g., mean, proportion).
SE measures the extent to which a sample statistic (e.g., sample mean) is
expected to vary from sample to sample due to random chance.
σ
SE =
‾
√n
σ = standard deviation of the population
n = sample size
Standard error gives us an idea of how much our sample statistic is likely to
fluctuate from the true population parameter due to random chance in the
sampling process.
BAO, Le · PIA2012 · https://baole.io/ 4
Standard Error
In practice, we usually use the sample estimate of the standard deviation.
s
SE =
√n‾
s = standard deviation of the sample
As the sample size grows, s should approximate σ
BAO, Le · PIA2012 · https://baole.io/ 5
Example
Suppose we want to estimate the average test score of students in a large
school. We take a random sample of 100 students, and the sample mean is
77, with a sample standard deviation (s) of 8.
s 8
SE = = = 0.8
√n‾ √100
‾‾‾‾
The standard error of the sample mean is 0.8 points. This tells us that, on
average, the sample mean will vary by about 0.8 points from the true
population mean due to random variation.
BAO, Le · PIA2012 · https://baole.io/ 6
What Affects Standard Erro
σ
SE =
‾
√n
Sample Size (n):
Larger samples lead to a smaller standard error because they better
approximate the population.
Inverse Relationship: As sample size increases, the standard error
decreases.
Population Variability (σ ):
The more variability in the population, the larger the standard error. High
variability means the sample is less likely to represent the population
well, increasing the SE.
BAO, Le · PIA2012 · https://baole.io/ 7
The Central Limit Theorem and SE
Central Limit Theorem (CLT): As long as the sample size is sufficiently large
(*n ≥ 30), the sampling distribution of the sample mean will be normally
distributed, even if the population distribution is not.
The normality of the sampling distribution allows us to make probabilistic
inferences using the standard error, even when the population distribution
is unknown.
In the next: CLT justifies the use of standard error in constructing confidence
intervals and hypothesis testing.
BAO, Le · PIA2012 · https://baole.io/ 8
SE of Proportions
Polling Example (Wheelan):
Suppose 46% of a poll of 1,000 Americans disapprove of the president’s
job performance, with a ±3% margin of error.
This margin is derived from the standard error. The poll tells us that if we
repeat the polls 100 times, approximately 95 time the true disapproval
rate lies between 43% and 49%.
SE of a Proportion:
‾p(1
‾‾‾‾‾‾‾
− p)‾
√
SE =
n
p is the sample proportion.
BAO, Le · PIA2012 · https://baole.io/ 9
Example
A researcher wants to know the proportion of students who passed a test. In
a sample of 200 students, 70% passed.
‾p(1
‾‾‾‾‾‾‾
− p)‾ ‾‾‾‾‾‾‾‾‾‾‾
− 0.7)‾
√ √
0.7(1
SE = = = 0.032
n 200
The standard error is approximately 0.032. This means that the sample
proportion (70%) is expected to vary by about 3.2 percentage points due to
random chance.
BAO, Le · PIA2012 · https://baole.io/ 10
Standard Errors in Inferential Statistics
Inferential Statistics: Used to determine whether the patterns observed in a
sample reflect true patterns in the population or are simply due to random
sampling variation.
Standard Error: Measures how much a sample statistic (e.g., mean,
proportion) is expected to vary from the population parameter due to
random variation.
BAO, Le · PIA2012 · https://baole.io/ 11
Comparing Group Means
Compares the average test scores of A class and B class. A class average
score is 85, while B class average score is 80.
Is this 5-point difference meaningful, or could it be due to random
sampling error?
Suppose the standard error of the difference between the means is 1.5
points
Sample statistic = 5, standard error = 1.5
5
Ratio: 1.50
= 3.33
The larger the value is, it’s more unlikely it’s due to random chance.
BAO, Le · PIA2012 · https://baole.io/ 12
