NORMALITY CHECK, ONE SAMPLE t
TEST
MUHAMMAD NIZAMUDDIN
MS in Biostatistics and Epidemiology (Dow)
BS MT (Surgical Technology)
Medical Technologist
Ophthalmology Department
Dow University Hospital
Dow International Medical College
SUMARIZED SNAP SHOT OF THE LAST
LECTURE:
• Statistical Inference: The act of generalization of statistic from
representative sample data to population parameter with
calculated probability (confidence level).
• The “Probability Distribution” of sample statistic is called Sampling
Distribution.
• Central Limit Theorem: If Sample size is large enough then
sampling distribution of statistic follows normal distribution
• ◦ Sample Mean
• ◦ Sample Proportion
• Point Estimate:
• Summarize the sample by a single number that is an estimate of the
population parameter
• Interval Estimate:
• A range of values within which, we believe the true parameter lies
with high probability
SUMARIZED SNAP SHOT OF THE LAST
LECTURE:
• Hypothesis testing: Make statement(s) regarding population
parameter values based on sample data and verify/nullify it.
• 1.Statement of Hypotheses: (Null (=, ≥, or ≤ ) and Alternative
(≠, >, or < ))
• 2.Level of Significance: α = 0.05/0.01/0.1
• 3.Test Statistic: Checked Mean of Normal Data
• 4.Critical Region: P Value < α
• 5.Computation Analyze -> Compare Means -> One sample t-
test.
• 6.Result & Conclusion: Based on P Value
SUMARIZED SNAP SHOT OF THE LAST
LECTURE:
• Hypothesis testing: Make statement(s) regarding population
parameter values based on sample data and verify/nullify it.
• 1.Statement of Hypotheses: [Null (=, ≥, or ≤ ) and Alternative
(≠, >, or < )]
• 2.Level of Significance: α = 0.05/0.01/0.1
• 3.Test Statistic: Checked Mean of Normal Data DID WE CHECK
NORMALITY?
• 4.Critical Region: P Value < α
• 5.Computation Analyze -> Compare Means -> One sample t-
test.
• 6.Result & Conclusion: Based on P Value
IS MY DATA “NORMAL”?
1. Through Measure of Central Tendency
2. Through histogram (For Transform Data)
3. Through Skewness and Kurtosis
(Both Skewness and Kurtosis should be between -1 & 1)
4. Through Probability (OR Q-Q) Plot
5. Through Normality Test
EXAMPLE 1
• Suppose we have ten samples from one of
experiment.
• 100, 98, 101, 93, 123, 112, 85, 76, 119, 111
• We want to know if we can reasonably
assume that these data come from a normal
distribution through Normality Test.
THE NORMALITY TEST
THE NORMALITY TEST
• We use Shapiro Wilk Test when observations
are ≤ 2000. For n>2000 we use P-value of
Kalmogrov-Simrnov Test
• Statement of Hypothesis:
Ho : Data follow normal distribution
Ha : Data do not follow normal distribution
P-value = 0.915 (From SPSS)
• Conclusion:
Since p-value is greater than 0.05 (level of significance)
therefore we fail to reject Ho and conclude that data follow
normal distribution.
Hypothesis Testing Procedures for
Mean
Example
• According to the College Entrance Board, the mean verbal score on
the Scholastic Aptitude Test (SAT) in 1983 was 425 points out of
possible 800.
• A random sample of 25 verbal scores for last year yielded the
following results:
346 496 352 378 315 491 360 385 500 558 381 303 434 562
420 485 446 479 422 494 289 436 516 615 496.
At 5% significant level, does it appear that last year’s mean verbal SAT
scores has changed over the 1983 mean score.
• [Hint: Before applying this parametric test (one sample t-test) first
check normality of the data.
1. One Sample T-test
• Two Steps for Performing One Samples t Test
i. Checking for Normality Assumption
ii. Test for Mean
Checking Normality Assumption
Hypothesis Testing Steps of Normality
Assumption:
• Statement of Hypothesis:
Ho : Sat score Data follows normal distribution
v/s Ha : Sat Score Data does not follow normal distribution
P-value =0.566 (From SPSS)
• Conclusion:
Since p-value is greater than i.e.(0.566 > 0.05)
therefore we fails to reject Ho and conclude that SAT scores
follows normal distribution. Therefore we can proceed to apply a
Parametric Test i.e. one sample t test.
ONE SAMPLE t TEST FOR POPULATION MEAN:
ONE SAMPLE t TESTING FOR POPULATION
MEAN:
• Statement of Hypothesis:
1. Ho : Last year’s mean verbal SAT scores has not changed over
the 1983 mean score. (µ= 425) v/s Ha : Last year’s mean verbal
SAT scores has changed over the 1983 mean score (µ ≠ 425)
2. Level of Significance: 0.05
3. Choose Test: One Sample t Test
4. Computation: Abs. Mean difference = 17.4, P-value =0.319
5. Critical Region: Reject Ho if P-value ≤ 0.05
6. Conclusion: Since p-value is greater than 0.05 i.e.(0.319 > 0.05)
therefore we fails to reject Ho and conclude that last year’s
mean
• Verbal SAT scores has no changed over the 1983 mean score.
Result Writing
• Since we have a score variable that is why we are writing here
Minimum and Maximum. If it is a continuous variable than we
will use C.I (See here).
N Mean ±SD Min -Max Test Value P-value
Score 25 442.4 ±85.2 289 -615 425 0.1595
N Mean ±SD 95% C.I Test Value P-value
Name
ASSIGNMENT:
Q:1) A random sample of 15 has the following results.
At 5% significant level, does it appear that mean age
has changed the last year population mean age 50
years? Write proper six steps of hypothesis testing and
check also the assumption of the test.
Age (years)
35 54 44.7 33.2 37 45.9 41.7 47.2
57 58.5 63 66.8 68 45 52
Q:2) The first 28 patients admitted for diagnosis and
evaluation to a newly opened center of a general hospital
experienced the following height (inches). Test the
hypothesis at 5% significant level; does it appear that
mean height has changed compare to the population
mean height of 65 inches? Write proper six steps of
hypothesis testing and check also the assumption of the
test.
Height (Inches)
65 66 69.5 68.5 62 70 69.5
63 62 70 66 67.5 71 68.5
66 73 71 67.5 63.5 75 66
66 70 76 62 66 69.5 63
