A general guidelines for applying different statis-
tical techniques is given below:
For further clarification, let’s consider one exam-
ple of each type of case:
Case 1: H0 (Null Hypothesis): There is no signif-
icant association between performance and
training (when both performance and training
are measured in nominal or ordinal scale) – Chi
square test
Multivariate data analysis are classified into two
basic groups –Dependence method and Interde-
pendence method.
Dependence Methods- When hypothesis con-
tains dependent and independent variables
“Dependence Techniques” are used. Again the
technique varies depending on metric or non-
metric data and on number of dependent varia-
Case 2: H0 (Null Hypothesis): There is no signifi-
cant association between income of the parents
and IQ level of the students (when both income
and IQ are measured in Interval or ratio scale) –
Pearson’s correlation test
Case 3: Independent samples
H0 (Null Hypothesis): There is no significant dif-
ference of financial awareness among male and
female. – t test (if sample size <30) OR Z test ((if
sample size >= 30)
Dependent samples: H0 (Null Hypothesis): There
is no significant difference of sales of the product
before and after sales promotion. – Pearson’s
correlation test
Case 4: H0 (Null Hypothesis): There is no signifi-
cant difference in the mileage of the three types
of cars. – One way ANOVA
For all the above test if the value of p < 0.05,
null hypothesis should be rejected at 5% level of
significance.
Multivariate Data Analysis- Research involving
three or more variables or that is concerned with
underlying dimensions among multiple variables,
will require multivariate statistical techniques to
analyse data. Suppose the case discussed at the
beginning of the article, impact of Servqual di-
mensions on Patients’ loyalty in case of Hospital.
12
bles. They are—
Interdependence Techniques: Any research ex-
amines questions that do not distinguish be-
tween independent and dependent variables will
be analysed by interdependency techniques.
• For metric inputs -- Factor Analysis, Clus-
ter Analysis and Metric-Multidimensional
Scaling techniques are used.
• For non-metric inputs- Non Metric multidi-
mensional scaling technique is used.
All the above data analysis techniques have var-
ious assumptions and applicability. Each of the
statistical technique again needs to be discussed
elaborately. This article is only an attempt to give
an overview of most commonly used techniques
in business research. A research has to decide
about when to use which techniques, depending
on the objective of the research, types of data
and distribution of the data and the assumptions
underlying in each technique.
For further clarification please contact at shampa_nandi@
yahoo.co.in
DYNAMISM(E)