Scribd
Upload
72 views4 pages

2017-1-31 Lecture 3a - Central Limit Theorem v1.0

The document discusses key concepts in parametric statistics including: 1) Parametric statistical tests require certain assumptions about the population distribution and sampling methodology. 2) The Central Limit Theorem states that as sample size increases, the sampling distribution of the sample mean will approximate a normal distribution, regardless of the shape of the original population distribution. 3) For a normally distributed population, the sampling distribution of the sample mean will also be normally distributed, allowing use of parametric tests to compare means. But for non-normal populations, sample sizes must be large enough for this approximation to hold.

Uploaded by

bahadar94
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
72 views4 pages

2017-1-31 Lecture 3a - Central Limit Theorem v1.0

The document discusses key concepts in parametric statistics including: 1) Parametric statistical tests require certain assumptions about the population distribution and sampling methodology. 2) The Central Limit Theorem states that as sample size increases, the sampling distribution of the sample mean will approximate a normal distribution, regardless of the shape of the original population distribution. 3) For a normally distributed population, the sampling distribution of the sample mean will also be normally distributed, allowing use of parametric tests to compare means. But for non-normal populations, sample sizes must be large enough for this approximation to hold.

Uploaded by

bahadar94
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

1/31/2017

 
Steven Tung

January 24 Lecture 1a: Data Types, Sampling Methodology, and Sampling Error

January 26 Lecture 2a: Descriptive Statistics, Probability Distribution, and Z score

January 31 Lecture 3a: Central Limit Theorem

February 2 Lecture 4a: Hypothesis Testing

February 7 Lecture 5a: Student’s t-test

February 9 Lecture 6a: Analysis of Variance

February 14 Lecture 7a: Non-Parametric Statistics and Chi-Square

February 16 Lecture 8a: Binomial and Poisson

February 21 Lecture 9a: Simple Linear Regression and Multiple Regression

February 23 Reserved

February 28 Reserved

March 2 Reserved – Possible Final Exam

1
1/31/2017

 Parametric statistics are used to make inferences about


population parameters

 In order to apply parametric statistical test, certain


assumptions must hold true:
o Samples come from a normal distribution
o Dependent variable is continuous (interval or ratio)
o Samples are random (selection of sample members are
independent of one another)
o Homogeneity of Variance (Variances are equal)

 Should these assumption fail to hold true, and “fixes” can’t be


made, you must use the less powerful non-parametric
statistical tests

 The Central Limit Theorem is the foundation of all


parametric statistics

 For both normal and non-normal distributed data, the


sampling distribution of the sample means will
approximate the normal distribution as the sample size
gets large, regardless of what the original distribution
looks like.

2
1/31/2017

10  If we were to plot the


9
8
frequency of sample means
7 (x̄), it would appear to look like
6 a normal distribution
5
4
3
2
1
0
1 2 3 4 5 6

Sample size = 4

S1 = [1,4,5,6] x̄1 = 4
S2 = [3,4,4,6] x̄2 = 4.25
S3 = [1,5,6,6] x̄3 = 4.5
.  The frequency table can be
. referred to as the “Sampling
Sn = [?,?,?,?] x̄n = ?? Distribution of the Sample
Mean”

Gmacro Gmacro
Population SamplingDistribution
#generate 10000 random numbers from the #k1 is sample size
#normal distribution with mean of 3 and Let k1 = 4
#stdev of 1 and place them into column 1 #k2 is number of samples
Random 10000 c1; Let k2 = 100
Normal 3 1.
#generate 10000 random numbers from the #Calculate mean of sample
#uniform distribution with a of 0 and b of 6 Do k3 = 1:k2
#and place them into column 3 #c5 stores index location based on
Random 10000 c3; #sample size requested
Uniform 0 6. Random k1 c5;
#plot the histograms of both my "population" integer 1 10000.
Histogram c1 c3; #c6 stores the actual value based on
Bar. #indexed location stored in c5
Endmacro Do k4 = 1:k1
#k5 is needed as a temporary
#index holder
Let k5 = c5(k4)
Let c6(k4) = c1(k5)
Let c7(k4) = c3(k5)
Enddo
#store sample mean from c6
Let c9(k3) = mean(c6)
Let c10(k3) = mean(c7)
Enddo
#plot the histograms of both my "sample distribution of
#the sample mean"
Histogram c9 c10;
Bar.
Endmacro

3
1/31/2017

 Mean
o “Sampling Distribution of the Sample Mean” will have the same mean as the
original distribution:

 Variance and Standard Deviation


o “Sampling Distribution of the Sample Mean” variance will follow the relationship:

o “Sampling Distribution of the Sample Mean” standard deviation will follow the
relationship:

• The standard deviation of the sampling distribution of the sample mean is


sometimes called the standard deviation of the mean or standard error of the mean
(SEM)

 If a characteristic is distributed normally in the population, you


can automatically use parametric statistical test to compare
the means (This is because the sampling distribution of the
sample mean drawn from normally distributed data always
has a normal distribution)

 If a characteristic is NOT distributed normally in the


population, you CAN’T automatically use parametric statistical
test to compare sample means (This is because sampling
distribution of the sample mean drawn from non-normally
distributed data doesn’t always have normal distribution)
o In this case, the sample size must be sufficiently large for the
sampling distribution of the sample mean to approach a normal
distribution

You might also like