FORMULAS IN INFERENTIAL STATISTICS
Number of Groups Being Compared
Parameter single two (independent) More than two (independent)
Two
(dependent)
Three or
more
(dependent)
Mean(s)
(1, 2, 3,,k)
z-test for single mean
n
x
z
o
=
t-test for single mean
n
s
x
t
= , df = n-1
z-test for two means
2
2
2
1
2
1
2 1
n
x x
z
+
=
t-test (with equal variances) for two means
( ) ( )
|
|
.
|
\
|
+
|
|
.
|
\
|
+
+
=
2 1 2 1
2
2 2
2
1 1
2 1
n
1
n
1
2 n n
s 1 n s 1 n
x x
t , df = n1 + n2 2
t-test (with unequal variances) for two means
2
2
2
1
2
1
2 1
n
s
n
s
x x
t
+
= , df =
Analysis of Variance (ANOVA): Single-Factor
MSE
MSC
(within) error squares mean
(between) columns squares mean
F = =
Analysis of Covariance (ANCOVA)
**Post hoc Analyses:
Tukeys HSD,
Duncan Multiple Range Test (DMRT)
Fishers LSD,
Dunnetts Test (if with control group),
Bonferonni,
Scheffe,
Student-Newman Keuls (S-N-K)
*Kruskal-Wallis Test
**Post hoc Analysis:
Mann-Whitney (U)
t-test
n
s
d
t
d
= df =
n-1
Analysis of
Variance for
Repeated
Measures
(ANOVARM)
*Friedman
Test
Variance(s)
(o
2
1, o
2
2, o
2
3,,
o
2
k)
Chi-square test for single variance
( )
2
2
2
1
o
_
s n
= , df = n-1
F test for two variances
2 1
2
2
2
1
s s where ,
s
s
F > =
*Levenes Test
Bartletts Test
Chi-square test
for single
variance Within
(
)
Proportion(s)
(p1, p2, p3,, pk)
z-test (n > 30)
( )
n
p 1 p
p p
z
=
where
n
x
p =
*Clopper-Pearson (neN)
z-test
( ) ( ) 2 1
2 1
2 1
2 1
n n
x x
p where ,
n
p 1 p
n
p 1 p
p p
z
+
+
=
=
Chi-square test of homogeneity
( )
= ,
E
E O
2
2
df = (rows 1)(columns 1)
**Post hoc analysis:
Marascuilo Test
*Mc Nemars
Test
* Non-parametric Test (does not assume normality/randomization and constancy of variance(s))
** Post hoc analyses are applied if data provides sufficient evidence that the means or proportions across two or more groups have significant difference
(
(
(
(
(
|
|
.
|
\
|
+
|
|
.
|
\
|
|
|
.
|
\
|
+
1 n
n
s
1 n
n
s
n
s
n
s
2
2
2
2
2
1
2
1
2
1
2
2
2
2
1
2
1
Correlation Techniques (Tests of Dependence)
Levels of Measurement Interval/Ratio Ordinal Nominal (non-dichotomous) Nominal (dichotomous)
Interval/Ratio
Pearson Correlation
*Spearman Correlation
*Spearman Correlation
***Point-Biserial Correlation
Ordinal *Spearman Correlation *Spearman Correlation
Rank-Biserial Correlation
Chi-square Test of
Independence
Nominal (dichotomous) n/a n/a
Chi-square test of
Independence
Chi-square test of
Independence
Nominal (dichotomous)
***Point-Biserial Correlation
Chi-square test of
Independence
Chi-square test of
Independence
Chi-square Test of
Independence
Pearson Correlation
Phi coefficient (|)
* Non-parametric
*** Correlation Technique derived from Pearson Correlation
Pearson Correlation coefficient may be computed as
( ) | | ( ) | |
=
2
2
2
2
Y Y n X X n
Y X XY n
r ,
where its test of significance may be computed using
1 = n r z for n > 30 or 2 ;
1
2
2
n df
r
n r
t =
= for n < 30
Spearman Correlation coefficient may be computed as
) n(n
d
1
6
1
2
2
=
,
where its test of significance may be computed using 2 ;
1
2
2
n df
r
n r
t =
=
Chi-square test of Independence test statistic may be computed as
( )
= ,
E
E O
2
2
df = (rows 1)(columns 1) if all 5 >
ij
E
, where
( )( )
GrandTotal
total Column total Row
E
j i
ij
= .
Otherwise, collapse or remove rows/columns.
If df = 1, Fishers exact test, where
( ) ( ) ( ) ( )
! ! ! ! !
! ! ! !
N D C B A
D B C A D C B A
p
+ + + +
=
Simple Regression Analysis
Y = a + bX +e or Y = 0 + 1X + e ; where
x
1 0
| | = , and
( )
( )
=
2
2
1
x x n
y x xy n
|
- Test of significance of
1
| may be performed to determine if 1 = 0
( )
2
2
1
0
x x
s
t
x y
|
, with df = n 2
CRITICAL VALUES
FOR Z-TESTS: FOR CHI-SQUARE (_
2
)TESTS:
Two-tailed test One-tailed test
.01 2.575 2.33 or 2.33
.05 1.96 1.645 or 1.645
.10 1.645 1.28 or 1.28
FOR t-TESTS:
FOR F TESTS:
= 0.05
= 0.01
