7 the t-test
TRANSCRIPT
-
8/2/2019 7 The t-test
1/22
The t-test
Inferences about Population Means
-
8/2/2019 7 The t-test
2/22
Questions
How are the distributions ofz and trelated?
Given that
construct a rejection region. Draw a pictureto illustrate.
What is the standard error of the difference
between means? What are the factors that
influence its size?
01.2;49;14;75:;75: )48,05(.10 tNsHH y
-
8/2/2019 7 The t-test
3/22
Questions (2)
What are the main uses of the t-test?
Give a concrete example of the use of the
{one sample, independent samples, dependent
samples} t-test. State why the particular testis the right one to choose.
What is the importance of variance accounted
for?
-
8/2/2019 7 The t-test
4/22
Confidence intervals in z
For large samples (N>100) can use z.
Suppose
Then
If
M
Mest
yz
.
)(
N
N
yy
N
sest
y
M1
)(
.
2
200;5;10:;10:10
NsHH y
35.
14.14
5
200
5.
N
sest
y
M
05.96.183.2;83.235.
)1011(11
pzy
-
8/2/2019 7 The t-test
5/22
The tDistributionWe use twhen the population variance is unknown (the
usual case) and sample size is small (N
-
8/2/2019 7 The t-test
6/22
Degrees of Freedom
For the tdistribution, degrees of freedom are always asimple function of the sample size, e.g., (N-1).
One way of explaining dfis that if we know the total or
mean, and all but one score, the last (N-1) score is not free to
vary. It is fixed by the other scores. 4+3+2+X = 10. X=1.
-
8/2/2019 7 The t-test
7/22
Confidence Intervals in t
With a small sample size, we compute the same numbers
as we did for z, but we compare them to the tdistribution
instead of thez distribution.
25;5;10:;10: 10 NsHH y
125
5.
N
sest
y
M1
1
)1011(11
ty
064.2)24,05(. t 1
-
8/2/2019 7 The t-test
8/22
Review
How are the distributions ofz and trelated?
Given that
construct a rejection region. Draw a pictureto illustrate.
01.2;49;14;75:;75: )48,05(.10 tNsHH y
-
8/2/2019 7 The t-test
9/22
Difference Between Means (1)
Most studies have at least 2 groups(e.g., M vs. F, Exp vs. Control)[1 v 2sample]
If we want to know diff in populationmeans, best guess is diff in samplemeans.
Unbiased:
Variance of the Difference:
Standard Error:
2
2
2
121)var(
MMyy
212121 )()()( yEyEyyE
2
2
2
1 MMdiff
-
8/2/2019 7 The t-test
10/22
Difference Between Means (2)
We can estimate the standard error of
the difference between means.
For large samples, can use z
2
2
2
1... MMdiff estestest
diffest
yy
diffz 2121 )(
3;100;12
2;100;10
0:;0:
222
111
211210
SDNy
SDNy
HH
36.100
13
100
9
100
4. diffest
05.;56.5
36.
2
36.
0)1210(
pzdiff
-
8/2/2019 7 The t-test
11/22
Independent Samples t (1)
Looks just like z:
df=N1-1+N2-1=N1+N2-2
If SDs are equal, estimate is:
diffestyy
difft 2121 )(
21
2
2
2
1
211
NNNNdiff
Pooled variance estimate is weighted average:
)]2/[(])1()1[( 212222112 NNsNsNestPooled Standard Error of the Difference (computed):
21
21
21
2
22
2
11
2
)1()1(.
NN
NN
NN
sNsN
est diff
-
8/2/2019 7 The t-test
12/22
Independent Samples t(2)
21
21
21
2
22
2
11
2
)1()1(. NN
NN
NN
sNsNest diff
diffest
yy
difft 2121 )(
7;83.5;20
5;7;18
0:;0:
2222
1
2
11
211210
Nsy
Nsy
HH
47.135
12
275
)83.5(6)7(4.
diffest
..;36.147.1
2
47.1
0)2018(sntdiff
tcrit= t(.05,10)=2.23
-
8/2/2019 7 The t-test
13/22
Assumptions
The t-test is based on assumptions of
normality and homogeneity of variance.
You can test for both these (make sure
you learn the SAS methods).
As long as the samples in each group
are large and nearly equal, the t-test is
robust, that is, still good, even thoassumptions are not met.
-
8/2/2019 7 The t-test
14/22
Review
What is the standard error of the
difference between means? What are
the factors that influence its size?
What are the assumptions of the t-test?
-
8/2/2019 7 The t-test
15/22
Strength of Association (1)
Scientific purpose is to predict orexplain variation.
Our variable Y has some variance that
we would like to account for. There arestatistical indexes of how well our IVaccounts for variance in the DV. Theseare measures of how strongly or closely
associated our IVs and DVs are. Variance accounted for:
2
2
21
2
2
|
2
2
4
)(
YY
XYY
-
8/2/2019 7 The t-test
16/22
Strength of Association (2)
How much of variance in Y is
associated with the IV?2
2
21
2
2
|
2
2
4
)(
YY
XYY
6420-2-4
0.4
0.3
0.2
0.1
0.0
Compare the 1st (left-most) curve with the curve in the
middle and the one on the right.
In each case, how
much of the variance
in Y is associated
with the IV, groupmembership? More
in the second
comparison. As
mean diff gets big, so
does variance acct.
-
8/2/2019 7 The t-test
17/22
Association & Significance
Power increaseswith association
(effect size) and
sample size.
Effect size:
Significance =
effect size X sample
size.
pyy /)( 21
21
2
21
11
)(
NN
yyt
p
Increasing sample size does not increase effect size
(strength of association). It decreases the standard
error so power is greater. Widely misunderstood.
N
yt
2
)(
pooledSD
XXd 21
-
8/2/2019 7 The t-test
18/22
Estimating Power (1)
If the null is false, the statistic is nolonger distributed as t, but rather as
noncentral t. This makes power
computation difficult. Hays (p. 334) presents an alternative
method based on strength of
association, that is, on
2
2
21
2
2
|
2
2
4
)(
YY
XYY
-
8/2/2019 7 The t-test
19/22
Estimating Power (2)
Based on Hayss method, we find:
35.22)25(.2
)75(.]58.228.1[2
gn
Suppose alpha is .01, power
desired is .90, and variance
accounted for is .25. What is
n per group? Its 24 (23?) per
group or 48 all together.
(Hays says add one moreperson for luck. its wise
28.1)90(.)1( zz
58.2)005(.)2/( zz
Same problem, but variance a/c is .10, need 68/group.
Same again, but .15, need 43 per group. What if alpha =
.05?
2
22
)2/()1(
2
)1(][
zzng
-
8/2/2019 7 The t-test
20/22
Dependent t(1)
Observations come in pairs. Brother, sister, repeated measure.
),cov(2 212
2
2
1
2 yyMMdiff
Problem solved by finding diffs between pairs Di=yi1-yi2.
1
)(2
2
N
DDs
i
D
N
sest DMD .N
DD
i )(
MDest
DEDt
.
)( df=N(pairs)-1
-
8/2/2019 7 The t-test
21/22
Dependent t (2)
Brother Sister
5 7
7 8
3 35y 6y
Diff2 1
1 0
0 1
1D
58.3/1. MD
est
72.158.
1
.
)(
MDest
DEDt
1
1
)(2
N
DDsD
2)( DD
df=2; n.s.
-
8/2/2019 7 The t-test
22/22
Review
What are the main uses of the t-test?
Give a concrete example of the use of the
{one sample, independent samples, dependent
samples} t-test. State why the particular testis the right one to choose.
What is the importance of variance accounted
for?