An Introduction to t-test

Presented by:
P.N.V.Sriram 2010231
Swati Jain 2010238
Tushar Manghrani 2010243
Udit Gupta 2010245
Ashikaben Patel 2010270
 Suppose we conducted a study to compare two strategies
for teaching spelling.
Group A had a mean score of 19. The range of scores was 16
to 22, and the standard deviation was 1.5.

Group B had a mean score of 20. The range of scores was 17

to 23, and the standard deviation was 1.5.

How confident can we be that the difference we found

occurredthe
between because
meansofofdifferences and
Group A in our reading strategies,
Group B by chance?
rather than

10
9
8
7
6
5
4
3
2
1

12 13 14 15 16 17 18 19 20 21 22 23 24 25
2 t-test 12/8/21
Spelling Test Scores
 A t-test allows us to compare the means of two groups and
determine how likely the difference between the two means
occurred by chance.

The calculations for a t-test requires three pieces of

information:
- the difference between the means (mean difference)
- the standard deviation for each group
- and the number of subjects in each group.

10
9
8
7
6
5
4
3
2
1

12 13 14 15 16 17 18 19 20 21 22 23 24 25
3 t-test 12/8/21
Spelling Test Scores
 All other factors being equal, large differences between
means are less likely to occur by chance
than small differences.

10
9
8
7
6
5
4
3
2
1

12 13 14 15 16 17 18 19 20 21 22 23 24 25
Spelling Test Scores

10
9
8
7
6
5
4
3
2
1

12 13 14 15 16 17 18 19 20 21 22 23 24 25
4 t-test 12/8/21
Spelling Test Scores
 The size of the standard deviation also influences the
outcome of a t-test.
Given the same difference in means, groups withare more
smaller
likely
standard
to report adeviations
significant difference than groups with larger
standard deviations.

10
9
8
7
6
5
4
3
2
1

12 13 14 15 16 17 18 19 20 21 22 23 24 25
Spelling Test Scores
10
9
8
7
6
5
4
3
2
1

12 13 14 15 16 17 18 19 20 21 22 23 24 25
5 t-test 12/8/21
Spelling Test Scores
 From a practical standpoint, we can see that smaller
standard
than larger
deviations Less overlap
standardproduce
deviations. would
less overlap indicate
between thethat the
groups
groups are more different from each other.

10
9
8
7
6
5
4
3
2
1

12 13 14 15 16 17 18 19 20 21 22 23 24 25
Spelling Test Scores
10
9
8
7
6
5
4
3
2
1

12 13 14 15 16 17 18 19 20 21 22 23 24 25
6 t-test 12/8/21
Spelling Test Scores
The size of our sample is also important. The more
subjects that are involved in a study,
the more confident we can be that the differences we find
between our groups did not occur by chance.
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1

12 13 14 15 16 17 18 19 20 21 22 23 24 25
Spelling Test Scores

10
9
8
7
6
5
4
3
2
1

12 13 14 15 16 17 18 19 20 21 22 23 24 25
7 t-test 12/8/21
Spelling Test Scores
Once we calculate the outcome of
the t-test(which produces a
t-value), we check that value
with the appropriate degrees of
freedom on a critical value table to
determine how likely the
difference between the means
occurred by chance.

8 t-test 12/8/21
There are three types of t-tests and each is calculated slightly
differently.

A correlated (or paired) t-test is concerned with the difference

between the average scores of a single sample of individuals who
is assessed at two different times (such as before treatment and
after treatment) or on two different measures. It can also
compare average scores of samples of individuals who are paired
in some way (such as siblings, mothers and daughters, persons
who are matched in terms of a particular characteristics).

An independent t-test compares the averages of two samples

that are selected independently of each other (the subjects in
the two groups are not the same people). There are two types of
independent t-tests: equal variance and unequal variance.
9 t-test 12/8/21
 An equal variance (pooled variance) t-test is used when the
number of subjects in the two groups is the same
OR the variance of the two groups is similar.

An unequal variance (separate variance) t-test is used when the

number of subjects in the two groups is different AND the
variance of the two groups is different.

10 t-test 12/8/21
 Overview

• Two paired samples: Within-Subject Designs

-Hypothesis test
-Confidence Interval
-Effect Size

• Two independent samples: Between-Subject Designs

- Hypothesis test
- Confidence interval
- Effect Size

11 t-test 12/8/21
 Comparing Two Populations
Until this point, all the inferential statistics we have considered
involve using one sample as the basis for drawing conclusion
about one population.

Although these single sample techniques are used occasionally in

real research, most research studies aim to compare of two (or
more) sets of data in order to make inferences about the
differences between two (or more) populations.

12 t-test 12/8/21
 Two kinds of studies

There are two general research strategies that can be used to

obtain the two sets of data to be compared:
1. The two sets of data could come from two independent
populations (e.g. women and men, or students from
section A and from section B) <- between-subjects design
2. The two sets of data could come from related
populations (e.g. “before treatment” and “after
treatment”) <- within-subjects design

13 t-test 12/8/21
How do we Are the scores for the two means
from the same subject (or related
determine which t- subjects)?
test to use…
Yes No
Paired t-test (Dependent Are there the same
t-test; Correlated t-test) number of people in the
two groups?

No

Yes
Equal Variance Are the variances of the
Independent t-test two groups different?
(Pooled Variance
Independent t-test)
Yes
No (Significance Level for
(Significance Level for Levene (or F-Max) is p
Equal Variance Levene (or F-Max) is p <.05
Independent t-test >.05
(Pooled Variance Unequal Variance
Independent t-test) Independent t-test
(Separate Variance
12/8/21 t-test Independent t-test)
 Part I

• Two paired samples: Within-Subject Designs

-Hypothesis test
-Confidence Interval
-Effect Size

15 t-test 12/8/21
 Paired T-Test for Within-Subjects Designs

Our hypotheses:
Ho: D = 0
HA: D  0

To test the null hypothesis, we’ll again compute a t

statistic and look it up in the t table.

Paired Samples t
s
t = D - D sD =
16 t-test sD n 12/8/21
 Confidence Intervals for Paired Samples

General formula
X  t (SE)

Paired Samples t
D  t (sD)

17 t-test 12/8/21
 Effect Size for Dependent Samples
One Sample d

ˆ X  H0
d
s

Paired Samples d

ˆ D
d
sD

18 t-test 12/8/21
 Example
In Everitt’s study (1994), 17 girls being treated for
anorexia were weighed before and after treatment.
Difference scores were calculated for each participant.

Change in Weight
n = 17
D = 7.26
sD = 7.16

Test the null hypothesis that there was no change in weight.

Compute a 95% confidence interval for the mean difference.
Calculate the effect size
19 t-test 12/8/21
 Change in Weight
Exercise
n = 17
D = 7.26
sD = 7.16

7.16
T-test SE   1.74
17

7.26  0
t (16)   4.17
1.74

p  .01

20 t-test 12/8/21
 Part II

• Two independent samples: Between-Subject Designs

-Hypothesis test
-Confidence Interval
-Effect Size

21 t-test 12/8/21
 T-Test for Independent Samples
The goal of a between-subjects research study is to evaluate
the mean difference between two populations (or between
two treatment conditions).

We can’t compute difference scores, so …

Ho: 1 = 2
HA: 1  2

22 t-test 12/8/21
 T-Test for Independent Samples

We can re-write these hypotheses as follows:

Ho: 1 - 2 = 0
HA: 1 - 2  0

To test the null hypothesis, we’ll again compute a t

statistic and look it up in the t table.

23 t-test 12/8/21
 T-Test for Independent Samples

General t formula
t = sample statistic - hypothesized population parameter
estimated standard error

One Sample t
XH0
ttest
sX

Independent samples t
(X 
X
t 1 2
)
( 1
2) s X1  X 2  ?
sX
1
X2

24 t-test 12/8/21
 T-Test for Independent Samples
Standard Error for a Difference in Means
s X1  X 2

The single-sample standard error ( sx ) measures how

much error expected between X and .

The independent-samples standard error (sx1-x2)

measures how much error is expected when you are
using a sample mean difference (X1 – X2) to represent a
population mean difference.

25 t-test 12/8/21
 T-Test for Independent Samples
Standard Error for a Difference in Means
s12 s22
s X1  X 2  
n1 n2

Each of the two sample means represents its own population mean, but
in each case there is some error.
The amount of error associated with each sample mean can be measured
by computing the standard errors.
To calculate the total amount of error involved in using two sample
means to approximate two population means, we will find the error from
each sample separately and then add the two errors together.
26 t-test 12/8/21
 T-Test for Independent Samples
Standard Error for a Difference in Means
s12 s22
s X1  X 2  
n1 n2
But…
This formula only works when n1 = n2. When the two
samples are different sizes, this formula is biased.
This comes from the fact that the formula above treats the
two sample variances equally. But we know that the
statistics obtained from large samples are better estimates,
so we need to give larger sample more weight in our
t-test 12/8/21
27
estimated standard error.
 T-Test for Independent Samples
Standard Error for a Difference in Means
s 2p s 2p
s X1  X 2  
n1 n2
We are going to change the formula slightly so that we use
the pooled sample variance instead of the individual sample
variances.

This pooled variance is going to be a weighted estimate of

the variance derived from the two samples.
2 SS1  SS2
s 
p
28 t-test df1  df 2 12/8/21
 Steps for Calculating a Test Statistic

Independent Samples T
1. Calculate X1-X2
SS1  SS2
2. Calculate pooled variance s2p 
2 2
df1  df 2
s
p s p
 3. Calculate standard error
n1 n2
(X
X) (
)
4. Calculate T and d.f. t1 2 1 2
sx
1x
2
d.f. = (n1 - 1) + (n2 - 1)
5. Use the Table

29 t-test 12/8/21
 Illustration
A developmental psychologist would like to examine the
difference in verbal skills for 8-year-old boys versus 8-
year-old girls. A sample of 10 boys and 10 girls is
obtained, and each child is given a standardized verbal
abilities test. The data for this experiment are as follows:

Girls Boys

n1 = 10 n2 = 10
X1 = 37 X 2 = 31
SS1 = 150 SS2 = 210

30 t-test 12/8/21
 Girls Boys
Illustration
n1 = 10 n2 = 10
X1 = 37 X 2 = 31
SS1 = 150 SS2 = 210

STEP 1: get mean difference

X1  X 2  6

31 t-test 12/8/21
 Girls Boys
Illustration
n1 = 10 n2 = 10
X1 = 37 X 2 = 31
SS1 = 150 SS2 = 210

STEP 2: Compute Pooled Variance

2SS1  SS2 150  210 360

s p    20
df1  df 2 (10  1)  (10 1) 18

32 t-test 12/8/21
 Girls Boys
Illustration
n1 = 10 n2 = 10
X1 = 37 X 2 = 31
SS1 = 150 SS2 = 210

STEP 3: Compute Standard Error

2 2
sp sp 20 20
SE      4 2
n1 n2 10 10

33 t-test 12/8/21
 Girls Boys
Illustration
n1 = 10 n2 = 10
X1 = 37 X 2 = 31
SS1 = 150 SS2 = 210

STEP 4: Compute T statistic and df

(X1  X 2 )  (1  2 ) (37  31)  0

t  3
sx1  x2 2

d.f. = (n1 - 1) + (n2 - 1) = (10-1) + (10-1) = 18

34 t-test 12/8/21
 Girls Boys
Illustration
n1 = 10 n2 = 10
X1 = 37 X 2 = 31
SS1 = 150 SS2 = 210

STEP 5: Use the table

T = 3 with 18 degrees of freedom

For alpha = .01, critical value of t is 2.878

Our T is more extreme, so we reject the null
There is a significant difference between boys and girls
35 t-test 12/8/21
Confidence Intervals for Independent Samples

General formula
X  t (SE)

One Sample t
X  t (sx)

Independent Sample t
(X1-X2)  t (sx1-x2)
36 t-test 12/8/21
 Sample Hypothesized Sample Estimated t-statistic
Data Population Variance Standard
Parameter Error

One sample
SS s2 X

2
X s  t
t-statistic df sx
n

Paired
samples t- SS
s  D
2 s2 D  D
statistic D D df t
n sD

Independent SS  SS2
1  2
2 2
samples X1  X 2 s  1
2 sp sp
p
df1  df 2 
t-statistic n1 n2

(X1  X 2 )  (1  2 )
t
37 t-test sx1  x212/8/21
Questions??????