Data Analysis
Data analysis embraces a whole range of activities of both the
qualitative and quantitative type. It is usual tendency in behavioral
research that much use of quantative analysis is made and statistical
methods and techniques are employed. The statistical methods and
techniques are employed. The statistical methods and techniques have
got a special position in research because they provide answers to the
problems.
Kaul defines data analysis as, ”Studying the organized material
in order to discover inherent facts. The data are studied from as many
angles as possible to explore the new facts.”
Purpose
The following are the main purposes of data analysis:
(i) Description:
It involves a set of activities that are as essential first step in the
development of most fields. A researcher must be able to identify a
topic about which much was not known; he must be able to convince
others about its importance and must be able to collect data.
(ii) Construction of Measurement Scale:
The researcher should construct a measurement scale. All numbers
generated by measuring instruments can be placed into one of four
categories:
(a) Nominal: The number serves as nothing more than labels. For example no 1 was not less than no 2 .Similarly no 2 was neither more than no 1 and nor less than no 3.
(b) Ordinal: Such numbers are used to designate an ordering along some dimensions such as from less to more, from small to large, from sooner to later.
(c) Interval: The interval provides more précised information than ordinal one. By this type of measurement the researcher can make exact and meaningful decisions. For example if A,B and C are of 150 cm, 145cm and 140 cm height, the researcher can say that A is 5 cm taller than B and B is 5 cm taller than C.
(d) Ratio Scale: It has two unique characteristics. The intervals between points can be demonstrated to be precisely the same and the scale has a conceptually meaningful zero point.
(iii) Generating empirical relationships:
Another purpose of analysis of data is identification of regularities and
relationships among data. The researcher has no clear idea about the
relationship which will be found from the collected data. If the data
were available in details it will be easier to determine the relationship.
The researcher can develop theories if he is able to recognize pattern
and order of data. The pattern may be showing association among
variables, which may be done by calculating correlation among variables
or showing order, precedence or priority. The derivation of empirical
laws may be made in the form of simple equations relating one interval
or ratio scaled variable to a few others through graph methods.
(iv) Explanation and prediction:
Generally knowledge and research are equated with the identification
of causal relationships and all research activities are directed to it. But in
many fields the research has not been developed to the level where
causal explanation is possible or valid predictions can be made. In such a
situation explanation and prediction is construct as enabling the values
of one set of variables to be derived given the values of another.
Functions
The following are the main functions of data analysis:
(i) The researcher should analyze the available data for examining
the statement of the problem.
(ii) The researcher should analyze the available data for examining
each hypothesis of the problem.
(iii) The researcher should study the original records of the data
before data analysis.
(iv) The researcher should analyze the data for thinking about the
research problem in lay man’s term.
(v) The researcher should analyze the data by attacking it through
statistical calculations.
(vi) The researcher should think in terms of significant tables that
the available data permits for the analysis of data.
Statisctical Calculations
The researcher will have to use either descriptive statistics or inferential
statistics for the purpose of the analysis.
(i) The descriptive statistics may be on any of the following forms:
(a) Measures of Central Tendency:
These measures are mean, median, mode geometric mean and harmonic mean. In behavioral statistics the last two measures are not used. Which of the first three will be used in social statistics depends upon the nature of the problem.
(b) Measures of Variability:
These measures are range, mean deviation, quartile deviation and standard deviation. In social statistics the first two measures are rarely used. The use of standard deviation is very frequently made for the purpose of analysis.
(c) Measures of Relative Position:
These measures are standard scores (Z or T scores), percentiles and percentile ranks .All of them are used in educational statistics for data analysis.
(d) Measures of Relationship:
There measures are Co-efficient of Correlation, partial correlation and multiple correlations. All of them are used in educational statistics for the analysis of data. However the use of rank method is made more in comparison to Karl pearson method.
(ii)The inferential statistics may be in any one of the following forms:
(a) Significance of Difference between Means:
It is used to determine whether a true difference exists between population means of two samples.
(b) Analysis of Variance:
The Z or t tests are used to determine whether there was any significant difference between the means of two random samples. The F test enables the researcher to determine whether the sample means differ from one another to a greater extent then the test scores differ from their own sample means using the F ratio.
(c) Analysis of Co-Variance:
It is an extension of analysis of variance to test the significance of difference between means of final experimental data by taking into account the Correlation between the dependent variable and one or more Co-variates or control variables and by adjusting initial mean differences in the group.
(d) Correlation Methods:
Either of two methods of correlation can be used for the purpose of calculating the significance of the difference between Co-efficient of Correlation.
(e) Chi Square Test:
It is used to estimate the like hood that some factor other than chance accounts to the observed relationship. In this test the expected frequency and observed frequency are used for evaluating Chi Square.
(f) Regression Analysis:
For calculating the probability of occurrence of any phenomenon or for predicting the phenomenon or relationship between different variables regression analysis is cone. Continue to Index