T Distribution,
Paired and Unpaired T test
Mean Comparison
between 2 groups
Prepared by:
Bikram Adhikari (136)
A bit of History
• W.A. Gassit (1905) first published a t-test.
• He worked at the Guiness Brewery in Dublin and
published under the name Student.
• The test was called Student Test later shortened
to t test
T Distribution
• Aka Student’s T distribution
• A probability distribution that is used to estimate
population parameters when the sample size is
small and/or when the population variance is
unknown.
T Distribution
• If X is normally distributed and a sample of size n is
randomly chosen from this underlying population,
the probability distribution of the random variable
is known as Student's t distribution with n- 1
degrees of freedom.
Properties of t Distribution
• The mean of distribution is ZERO.
• It is symmetrical about the mean.
• In general, it has a variance greater than 1, but the
variance approaches 1 as the sample size becomes
large.
• The variable t ranges from -∞ to +∞
Ho
rejectio
n
region
Similarities with Normal
Distribution
• Both are uni-modal
• Symmetric around is mean of zero
• Total area of the curve is One.
Difference between t and normal
distribution
• Compared to the normal distribution, the t
distribution is less peaked in the center and
has thicker tails.
When to Use the t Distribution?
• The t distribution can be used with any statistic
having a bell-shaped distribution
• Quantative data, random samples
• The population distribution is normal.
• The population distribution is symmetric, unimodal,
without outiers, and the sample size is at most 30.
• The population distribution is moderately skewed,
unimodal, without outliers, and the sample size is at
most 40.
Application of t test
• T test can be applied if:
• Samples are randomly selected from population
• There is homogeneity of variance in sample
• It is applied to find the significance of difference
between two means as:
• Unpaired t-test
• Paired t-test.
Statistical methods for differences or
paired samples
Statistical methods for comparing two
independent groups or samples
Standard Error of Mean
• Difference between sample estimates of statistics
and population parameter is measured by standard
error
• Standard error is a measure of chance variation and
it does not mean error or mistake
• Calculation of standard error of mean:
Standard Error of difference
between Means
• Frequency distribution of difference give a normal
curve.
• The standard deviation of a distribution of
differences is known as standard error of
difference between two means.
• Calculation of standard error of difference between
two means:
Degree of Freedom (df)
• Degrees of freedom depends on two factors:
• No. of groups we wish to compare
• No. of parameters we need to estimate to calculate the
standard deviation of the contrast of interest.
• For paired t test: df=n-1
• For one sample t test: df=n-1
• For two sample t test: df=2n-2
Paired t test
• Two groups of paired observations, x11, x12, . . . ,
x1n in Group 1 and x21, x22, . . . , x2n in Group 2
such that x1i is paired with x2i and the difference
between them, di = x1i - x2i.
• Assumptions
• The di’s are plausibly Normally distributed. It is not
essential for the original observations to be Normally distributed
• The di’s are independent of each other.
Paired t test
• Steps
• Hypothesis generation:
• Null Hypothesis: mean difference is zero
• Alternative Hypothesis: mean difference in the population is not zero.
• Calculate the differences di = x1i - x2i, i = 1 to n.
• Calculate the mean difference (d) and standard deviation,
(Sd) of the differences di.
• Calculate the standard error of the mean difference
• Calculate the test statistic
Paired t test
• Steps
• Find the degrees of freedom.
• df=n-1
• Refer 't' table and find the probability of the
calculated ‘t’ corresponding to n – 1 degrees of
freedom.
• Result
Paired t test
Problem
• Systolic blood pressure (SBP) of 9 normal
individuals, who had been recumbent for 5 minutes
was taken. Then 2 ml of 0.5% solution of
hypotensive drug was given and blood pressure
recorded again. Did the injection of drug lower the
blood pressure?
Paired t test
• Hypothesis:
• Ho: no mean difference
• Ha: mean difference
• Calculate differences
• Calculation of mean
difference:
• Calculate Sd of mean
difference:
Paired t test
• Calculate standard error of mean difference:
• Calculate test statistic
• Degree of freedom: n-1=9-1=8
Paired t test
• Find the probability of the calculated ‘t’
corresponding to 8 degrees of freedom and 5%
significance limit.
• T=2.31
2.20
Paired t test
• Test statistic lies in
rejection region.
• Ho is rejected Ho Ho
rejection rejection
• It means there is region region
Ho
significant difference acceptance
between means at 5% region
cl.
2.31 2.31 5.
Unpaired t test
• One Sample t test
• Independent two Sample t test
Independent sample t test
• The independent samples t-test is used to test for a
difference in the mean value of a continuous
variable between two groups.
• Assumptions
• The groups are independent.
• The variables of interest are continuous
• The data in both groups have similar standard deviations
• Homogeneity of variance
• The data is Normally distributed in both groups.
Homogeneity of variance test
• Levene’s test
• This tests whether the variances of two samples are
approximately equal or not.
• We want levene’s test to be not significant
• Ho: Variance are equal
• Ha: Variances are not equal
• Note: As long as N>30, n1 nearly=to n2, there is
robust to violations of homogeneity of variance
Independent sample t test
• Steps
• Generate hypothesis
• Ho: no difference in two means
• Ha: there is difference in two means
• Calculate mean difference between groups
• Calculate pooled standard deviations
• Calculate the standard error of the difference
between two means
Independent sample t test
• Calculate the test statistic
• Calculate degree of freedom:
df=n1+n2-2
• Refer 't' table and find the probability of the
calculated ‘t’ corresponding to 2n – 2 degrees of
freedom.
• Result interpretation
Independent sample t test
Independent sample t test
• Degree of freedom =9+13-2 =20
• T20=2.086 (from table)
• Calculated t statistic=2.63
What if the variances in two
groups are not Equal?
• Use of modification of Independent sample t test.
• Instead of using sp 2 as an estimate of the common
variance σ 2, we substitute s1 2 for ai σ 2
and s2 2 for σ2 2. Therefore, the appropriate test
statistic is
Degree of
freedom
One Sample t test
• It is used for test of the null hypothesis that our
data are a sample from a population with a specific
'hypothesized' mean.
• Test statistic(t):
Ho: there is no difference in dietary intake from recommended level.
Ha: there is difference in dietary intake from recommended level.
• 95% CI
• df=10
• t=2.23
One Sample t test
• Result interpretation
• The dietary intake of female is significantly lower than
recommended level
Ho Ho
rejection rejection
region region
Ho
acceptance
region
2.23 2.23
References
• Altman, A practical statistics for medical research,
8th edition, 1999
• Mahajan’s Methods in Biostatistics for medical
students and research works. 8th edition,2016
• Medical statistics-Text book for health sciences, 4th
edition, 2007
• Principles of Biostatistics, Marcello Pagano, 2nd
Edition
• Few internet sites.
Any Queries????
Thank you
