0rclauIon t een i - t e s t ana Fos LCSt DCores

Alternatives to the t-testbrogob ano ataroh onoaidinl(

a2 The t-test can be used to test the equality of the means of two normal
populations with unknown, but equal, variance. To relax the normality
assumption, a non-parametric alternative to the t-test can be used and the
usual choices are (a) for independent samples, the Mann-Whitney U test
and (b) for related samples, either the binomial test or the
Wilcoxin signed-
rank test. To test the equality of the means of more than two normal
populations, an analysis of variance can be performed. or 1o6tdu
ho(B) ANOVA : Analysis of variance (ANOVA) is used for comparing
more
than two groups on a
statistical
single variable. It is a collection of
models and their associated procedures, in which the observed variance
is partitioned into components due to different
explanatory variables. In
its simplest form, ANOVA gives a statistical test of whether
the means of
several groups are all equal, and therefore generalizes Student's two
sample t-test to more than two groups.
vOMA in0
There are three conceptual classes of such models: np
1. Fixed-effects model assumes that the data came from normal
populations, which may differ only in their means. The fixed-effects
bo model of analysis of variance applies to situations in which the
o experimenter applies several treatments to the subjects of the
bn experiment to see if the response variable's values change. This allows
nthe experimenter to estimate the ranges of response variable values
that the treatment would generate in the population as a whole.
 402 Educational Research

Random-effects model assumes that the data describe a hierarchy of

different populations whose differences are constrained by the
hierarchy. Random effects models are used when the treatments are
not fixed. This occurs when the various treatments (also known
as
factor levels) are
sampled from a larger population. Because the
treatments themselves are random variables, some assumptions and
the method of
contrasting the treatments differ from ANOVA.
3. Mixed-effect model describes situations where both fixed and
random
effects are present.
2+Ma2)
Types of ANOVA : These are as follows:
(a) One-way ANOVA : It is used to test for
differences among two or
more
independent groups. Typically, however, the one-way ANOVA
used to test for
differences among three or more
groups, since the
two-group be covered by student's t-test. When
case can
there are
only two means to compare, the t-test and the F-test are
with F = t. For example, a researcher wants to equivalent
attitude towards the school based on school compare students
ICSE). In this case, there is one dependent types (SSC, CBSE and
variable, namely, attitude
towards the school and three
groups, namely, SSC, CBSE and ICSE
schools. Here, the one-way ANOVA is used to test
for differences in
students' attitude towards the school
among the three groups.
(b) One-way ANOVA for
repeated measures : It is used when the
subjects subjected to repeated measures. This means that
are

subjects are used for each treatment. Note that this methodthecan
same
be
subject to carryover effects. This technique is often
experimental research in which we want to
used in
compare three or more
groups on one
dependent variable which is measured twice i.e. as
pre-test and post-test.
(c) Two-way ANOVA : It is used when the
researcher wants to study the
effects of two
known
or more
independent or treatment variables. It is also
asfactorial ANOVA. The most
factorial ANOVA is the 2x2
commonly used type of
(read as "two by two'", as you would
matrix) design, where there are two a
independent variables and each
variable has two levels or distinct values.
be multi-level such as 3x3, etc. Two-way ANOVA can also
or higher order such as
but analyses with 2x2x2, etc.
higher numbers of factors are rarely done hand by
because the calculations are
lengthy. However, since the
of dataanalytic software, the utilisation of higher order introduction
designs and
analyses has become quite common. For
to compare students' example, a researcher wants
attitude towards the school based on (i) school
 Quantitative Data Analysis
403

types (SSC, CBSE and ICSE) and (ii) gender. In this case, there is
one dependent variable, namely, attitude towards the school and two
independent variables, viz., (G) school types including three levels,
namely, SSC, CBSE and ICSE schools and (ii) gender including two
levels, namely, boys and girls. Here, the two-way ANOVA is used to
test for differences in students' attitude towards the school on the
2basis of (i) school types and (il) gender. This is an example of 3x2
two-way ANOVA as there are three levels of school types, namely,
SSC, CBSE and ICSE schools and two levels of gender, namely,
boys and girls. It is known as two-way ANOVA as it involves
comparing one dependent variable (attitude towards the school) on
the basis of two independent variables, viz., (i) school
types and (ii)
gender. Jsno odt yd boielqo no nshsy
(d) MANOVA : When one wants to compare two or more independent
groups in which the sample is subjected to repeated measures such
as pre-test and post-test in an experimental study, one may perform a
factorial mixed-design ANOVA ie. Multivariate Analysis of Variance
or MANOVA in which one factor is a between-subjects variable and
the other is within-subjects variable. This is a type of
mixed-effect
model. It is used when there is more than one dependent
variable.
(e) ANCOVA: While comparing two groups on a dependent variable, if
it is found that they differ on some other variable
such as their IQ
SES or pre-test, it is necessary to remove these initial
This can be done through using the technique of
differences.
ANCOVA.od
Assumptions of Using ANOVA
1. Independence ofcases
2 Normality of the distributions in each of the groups.
Equality or homogeneity of variances, known as homoscedasticity
i.e. the variance of data in groups should be the same. Levene's
test for
homogeneity of variances is typically used to confirm homoscedasticity.
The Kolmogorov-Smirnov or the Shapiro-Wilk test may be
used to confirm
normality. According to Lindman, F-test is unreliable if there are
deviations from normality whereas Ferguson and Takanes claim thatthe
Ftest is robust. The Kruskal-Wallistest is a non-parametric alternative
which does not rely on an assumption of normality. These
togetherform
. Lindman, H. R. Analysis of Variance in Compler Experimental Designs.WH
Freeman& Co, San Francisco, 1974.
2. Ferguson, G. A. and Takane, Y. Statistical Analysis in Psychology and
Education. (Sixth Edition). Montréal, Quebec: McGraw-Hill RyersonLimited.
2005. pp.261-2
404 ieEducational Research
thecommon assumption that the errors are independently, identically
andnormally distributed for fixed effects models.
(CEAfect Size Measures