Comparison of Two Population Mean

Comparison of 2 Population
Means
• Goal: To compare 2 populations/treatments
wrt a numeric outcome
• Sampling Design: Independent Samples
(Parallel Groups) vs Paired Samples
(Crossover Design)
• Data Structure: Normal vs Non-normal
• Sample Sizes: Large (n1,n2>20) vs Small
Independent Samples
• Units in the two samples are different
• Sample sizes may or may not be equal
• Large-sample inference based on Normal
Distribution (Central Limit Theorem)
• Small-sample inference depends on
distribution of individual outcomes (Normal
vs non-Normal)
Parameters/Estimates
(Independent Samples)
• Parameter:     
• Estimator: Y 1  Y 2
S12 S 22
• 
Estimated standard error: n1 n2
• Shape of sampling distribution:
– Normal if data are normal
– Approximately normal if n1,n2>20
– Non-normal otherwise (typically)
Large-Sample Test of 
• Null hypothesis: The population means differ by

0 (which is typically 0): H 0 : 1   2   0
• Alternative Hypotheses:
– 1-Sided: H A : 1   2   0
– 2-Sided: H A : 1   2   0
• Test Statistic: ( y1  y 2 )   0
zobs 
S12 S 22

n1 n2
• Decision Rule:
– 1-sided alternative H A : 1   2   0
• If zobs  z ==> Conclude 

• If zobs < z ==> Do not reject 
H A : 1   2   0
– 2-sided alternative
• If zobs  z ==> Conclude 

• If zobs  -z ==> Conclude 
• If -z < zobs < z ==> Do not reject 
• Observed Significance Level (P-Value)

– 1-sided alternative H A : 1   2   0
• P=P(z  zobs) (From the std. Normal distribution)
– 2-sided alternative H A : 1   2   0
• P=2P( z |zobs| ) (From the std. Normal distribution)
• If P-Value  then reject the null hypothesis

Large-Sample (1-100% Confidence
Interval for 
• Confidence Coefficient (1-) refers to the proportion
of times this rule would provide an interval that
contains the true parameter value if it were
applied over all possible samples
• Rule:
y 1 
 y 2  z / 2
S12 S 22

n1 n2
Large-Sample (1-100% Confidence
Interval for 
• For 95% Confidence Intervals, z.025=1.96
• Confidence Intervals and 2-sided tests give
identical conclusions at same -level:
– If entire interval is above 0, conclude 
– If entire interval is below 0, conclude 
– If interval contains 0, do not reject ≠ 
Example: Vitamin C for Common Cold
• Outcome: Number of Colds During Study Period
for Each Student
• Group 1: Given Placebo
y1  2.2 s1  0.12 n1  155
• Group 2: Given Ascorbic Acid (Vitamin C)
y 2  1.9 s2  0.10 n2  208
Source: Pauling (1971)

2-Sided Test to Compare Groups
• H0: 12No difference in trt effects)

• HA: 12≠Difference in trt effects)
(2.2  1.9)  0 0.3

• Test Statistic: zobs 
(0.12) 2 (0.10) 2

0.0119
 25.3

155 208
• Decision Rule (=0.05)

– Conclude > 0 since zobs = 25.3 > z.025 = 1.96
95% Confidence Interval for 
• Point Estimate: y1  y 2  2.2  1.9  0.3
• Estimated Std. Error: (0.12) 2 (0.10) 2

  0.0119
155 208
• Critical Value: z.025 = 1.96
• 95% CI: 0.30 ± 1.96(0.0119)  0.30 ± 0.023

 (0.277 , 0.323) Entire interval > 0
Small-Sample Test for 
Normal Populations
• Case 1: Common Variances (12 = 22 = 2)
• Null Hypothesis: H 0 : 1   2   0
• Alternative Hypotheses:
– 1-Sided: H A : 1   2   0
– 2-Sided: H A : 1   2   0
• Test Statistic:(where Sp2 is a “pooled” estimate of 2)

2 2
( y1  y 2 )   0 ( n  1) S  ( n  1) S
t obs  S p2  1 1 2 2
 1 1  n1  n2  2
S p2   
 n1 n2 
Normal Populations
• Decision Rule: (Based on t-distribution with =n1+n2-2 df)
• If tobs  t, ==> Conclude 
• If tobs < t ==> Do not reject 
• If tobs  t , ==> Conclude 
• If tobs  -t ==> Conclude 
• If -t < tobs < t ==> Do not reject 
Normal Populations
• Observed Significance Level (P-Value)
• Special Tables Needed, Printed by Statistical Software
Packages
• P=P(t  tobs) (From the t distribution)
• P=2P( t  |tobs| ) (From the t distribution)
• If P-Value  then reject the null hypothesis
Small-Sample (1-100% Confidence Interval
for Normal Populations
• Confidence Coefficient (1-) refers to the proportion of
times this rule would provide an interval that contains the
true parameter value if it were applied over all
possible samples
• Rule: y 1 
 y 2  t / 2, 2 1 1
S   
p
 n1 n2 
• Interpretations same as for large-sample CI’s
Small-Sample Inference for 
Normal Populations
• Case 2: 12  22
S12 S 22
• Don’t pool variances: Sy y  
1 2
n1 n2
• Use “adjusted” degrees of freedom (Satterthwaites’

Approximation) : S
 
S 
 1
2 2
2
2
 n n2 
*   1 
S2 
2
 S 22 
2

 1 
 n1 
 
 n2 
 
 n 1 n2  1 
 1

 
 

Comparison of Two Population Mean

Uploaded by

Copyright:

Available Formats

Comparison of Two Population Mean

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Comparison of Two Population Mean

Uploaded by

Copyright:

Available Formats

Comparison of 2 Population

• Null hypothesis: The population means differ by

• If zobs  z ==> Conclude 

• If zobs  z ==> Conclude 

• Observed Significance Level (P-Value)

• If P-Value  then reject the null hypothesis

y1  2.2 s1  0.12 n1  155

• Group 2: Given Ascorbic Acid (Vitamin C)

y 2  1.9 s2  0.10 n2  208

Source: Pauling (1971)

• H0: 12No difference in trt effects)

(2.2  1.9)  0 0.3

• Decision Rule (=0.05)

• Point Estimate: y1  y 2  2.2  1.9  0.3

• Estimated Std. Error: (0.12) 2 (0.10) 2

• Critical Value: z.025 = 1.96

• 95% CI: 0.30 ± 1.96(0.0119)  0.30 ± 0.023

• Test Statistic:(where Sp2 is a “pooled” estimate of 2)

• Use “adjusted” degrees of freedom (Satterthwaites’

You might also like