86 CHAPTER 4 RESEARCH METHODOLOGY 4.1 INTRODUCTION This chapter describes the research design adopted in the study to fulfil the objectives set forth for the study. The research design adopted is a combination of qualitative and quantitative methodologies in which primary and secondary sources are very thoughtfully tapped for information and perspectives. The main focuses of the present study are: Household water sources, access to water, quality of water, quantity of water, sanitation, hygiene, health and the willingness to pay. The study examines each of these in relation to Chennai city, especially in regard to the north and south zones. Affordability to pay in the two locations, that is, north zone and south zone of Chennai city. Hence, the two zones with various household options for water have been selected to conduct interviews with high, middle and low income groups of people. Data have been collected for two periods, April and June in 2010 in both the zones of study area. The data collected through interview schedule have been properly coded, master tabulated and analyzed using the Statistical Package for Social Sciences (SPSS). For the purpose of this study water use at the household level has been obtained from households (primary sources) through an extensive household survey using a questionnaire designed for the purpose. The
23
Embed
CHAPTER 4 RESEARCH METHODOLOGY - Shodhgangashodhganga.inflibnet.ac.in/bitstream/10603/24959/9/09_chapter 4.pdf · potentially useful to humans. Uses of water include agricultural,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
86
CHAPTER 4
RESEARCH METHODOLOGY
4.1 INTRODUCTION
This chapter describes the research design adopted in the study to
fulfil the objectives set forth for the study. The research design adopted is a
combination of qualitative and quantitative methodologies in which primary
and secondary sources are very thoughtfully tapped for information and
perspectives. The main focuses of the present study are: Household water
sources, access to water, quality of water, quantity of water, sanitation,
hygiene, health and the willingness to pay. The study examines each of these
in relation to Chennai city, especially in regard to the north and south zones.
Affordability to pay in the two locations, that is, north zone and south zone of
Chennai city. Hence, the two zones with various household options for water
have been selected to conduct interviews with high, middle and low income
groups of people. Data have been collected for two periods, April and June in
2010 in both the zones of study area.
The data collected through interview schedule have been properly
coded, master tabulated and analyzed using the Statistical Package for Social
Sciences (SPSS).
For the purpose of this study water use at the household level has
been obtained from households (primary sources) through an extensive
household survey using a questionnaire designed for the purpose. The
87
secondary data needed for the study have been collected from various relevant
government departments. The collected data have been analyzed using
appropriate computer assisted and analytical procedures using SPSS and Arc
GIS.
Primary data collected through questionnaire survey have been
coded, tabulated and analyzed using SPSS software. The analysed data have
been presented with the help of a series of maps, tables and graphs using
digital cartographic techniques. This analysis has included assessment of
households using water from different sources for water uses, assessment of
water, water quality and willingness to pay.
4.2 OPERATIONAL DEFINITIONS
The operational definitions of various terms, particularly water
sources used in the study are presented, which may be trivial but are
necessary for clarity of thought and understanding.
Water Resources: Water resources are sources of water that are useful or
potentially useful to humans. Uses of water include agricultural, industrial,
household, recreational and environmental activities. Virtually all of these
human uses require freshwater. It is estimated that 8 percent of the water use
worldwide is for household purposes. These include drinking water, bathing
water, cooking water, water for sanitation and gardening. Basic household
water requirements have been estimated by Peter Gleick at around 50 litres
per person per day, excluding water for gardens. Drinking water is water that
is of sufficiently high quality so that it can be consumed or used without risk
of immediate or long term harm. Such water is commonly called potable
water. In most developed countries, the water supplied to households,
commerce and industry is all of drinking water standard eventhough only a
very small proportion is actually consumed or used in food preparation.
88
Water Supply: Basic needs include access to a safe supply of water for
domestic use, meaning for drinking, food preparation, bathing, laundry,
dishwashing and cleaning. In many cases domestic water may also be used for
watering animals and vegetable plots or gardens.
Water Access: Definitions of ‘access’ (Distance to the nearest water – point
and per capita availability) and ‘safe’ (Water Quality) may vary from country
to country.
WTP: WTP is the maximum amount that an individual status they are willing
to pay for a good or service DfID (1997).
ATP: ATP is the extent to which prices (for example, water supply and
sanitation) are within the financial means of users. An important consideration in
service planning relating to choice of service level and pricing.
Sanitation and Hygiene: The word ‘sanitation’ alone is taken to mean the
safe management of human excreta. It therefore includes both the ‘hardware’
(e.g. Latrines and sewers) and the ‘software’ (regulation, Hygiene promotion)
needed to reduce fecal-oral disease transmission. It encompasses the
re-use of and ultimate disposal of human excreta.
4.3 SELECTION OF STUDY AREA AND SAMPLES
With respect to time and resource constraints, it has been decided to
conduct the field study in Chennai city. The city has a total of 10 Zones and
155 wards. This study has selected two northern and two southern zones for
study. In the northern part, 2 zones, namely, Tondiarpet and Pulianthope have
been selected for interview based questionnaire survey. In the southern part
also two 2 zones, namely, Saidapet and Adyar have been selected for
questionnaire survey (Figure 4.1).
89
In the northern part, and in Tondiarpet zone, Kodungaiyur East,
Dr. Radhe Krishnan Nagar North, Jeeva Nagar North and Cheriyan Nagar
South have been included in the study. In Pulianthope zone, Vyasarpet South,
Vyasarpet North, Perambur North, and Elango Nagar have been selected in
4 wards. In the southern part, and in Saidapet zone, G.D Naidu Nagar East,
G.D Naidu Nagar West, Guindy West, Guindy East and in Adyar zone, Adyar
West, Adyar East, Thiruvanmiyur West, and Thiruvanmiyur East have been
selected for questionnaire-based study.
The questionnaire survey has covered 320 samples with each ward
accounting for 21 samples. The survey has been conducted with 3 groups of
people, namely, high, medium and low income. The levels have been
identified using land values in Chennai city. This is thus a stratified sample
survey.
4.3.1 Analytical Framework for the Study
The research framework is integral to any research and indeed
forms the heart of the research work. Devising a research framework requires
a proper and clear understanding of the objectives of research. The framework
for the present study (Figure 4.2) is broadly divided into three
aspects- Theorectical aspects, Data aspects and Analytical aspects. The three
aspects are interlinked and contribute to each other and are meaningless
without the links which connect them. The theoretical aspects refer the
theoretical base in the present research. It consist of theories on which the
research is grounded. In order to gaining a deeper understanding of the
theoretical aspects as discussed earlier and the interplay of the economic,
cultural and environmental factors, the study was carried out in urban
neighbourhoods, in the north and south Chennai. Data was collected from
both primary and secondary data sources. The analytical aspects range from
the formulation of the research questions to using appropriate analytical tools
to answer them.
90
Figure 4.2 presents the analytical framework adopted for the study.
It is evident from the schematic that how theoretical, data and analytical
aspects of the study are treated in the context of the study, and the schematic
is rather self-explanatory indicating how primary and secondary data have
been used in the analysis and writing up of the thesis, towards developing
recommendations for resolving problems connected with the water demand
and supply, including WTP and ATP. Figure 4.3 shows elaborately as to how
primary and secondary data have been collected and analysed for
interpretation and inferences.
Figure 4.1 Field Setting: Sample Areas, Sampling and Samples
155 Wards (16 Wards 10 %)
(319 samples)
North: 2 Zones (159 samples) South: 2 Zones (160 samples)
Saidapet Tondiarpet
Pulianthope Adayar
151. Adyar West
152. Adyar East
154. Thiruvanmiyur
West
155. Thiruvanmiyur
East
138. G.D. Naidu Nagar
East
139. G.D. Naidu Nagar
West
140. Guindy West
141. Guindy East
2. Kodungaiyur East
3. Dr. Radha Krishnan Nagar
North
5. Jeeva nagar North
6. Cheriyan Nagar South
32. Vyasarpet South
33. Vyasarpet North
34. Perambur North
36. Elango Nagar
91
Figure 4.2 Analytical Framework for the Study
SPPSS analysis
SPSS
Analysis
SYSTHESIS
Conceptual frame work
Water, Hygiene, Sanitation and Health
Two stages of Water Evaluation process
1. Willingness to pay and 2. Affordable to pay
Population
Water Distribution
Water connection, Assesses, Consumer
Water Infrastructure
Demand and supply gap
Review of studies
Socio-Economic Assessment, Water Supply, Sources of water,
Metro water connection, Metro water storages, Cost, Domestic
water sector, Toilets Hygiene, Clinging Sanitary Infrastructures and Diseases and Health, Indoor and outdoor.
GIS
&Excel
Theory
Analysis
nd
Social Surveys
and
Approaches
Factor analysis
Socio-Economic Dimension
Household size and Water Consumption
Dimension
Attitude Dimension
Potable and Bathing water Dimension
Factor Analysis
Socio-Economics of Food and Body Hygiene
Dimension
Hospital Visitation for Health Problems Dimension
Sanitation and Cleanliness Dimension
Social Component Dimension
Implication and Recommendation for Water Supply
and Sanitary Infrastructure, Closed and Well
Maintained Sewer Drains, Sewage and Wastewater
Treatment Plant, Provision of Adequate Health
Services.
92
4.3.2 Sources of Data
Since the present study seeks to identify the people’s attitudes
towards paying for Water with WTP and ATP, it has been decided that
obtaining first hand information directly from the respondents would be the
most reliable sources through which the main focus of the study could be
analysed. Even though the study primarily relies upon the primary sources, it
still seeks data for some supportive and supplemental information, which are
secondary in nature.
The secondary data have been collected from the Government
records such as those of the Metro Water Board and maps have been collected
from the Survey and Land Records Department of the Government of Tamil
Nadu, Chennai.
Figure 4.3 Data Collection and Data Analysis in the Study
Result
Water Sources, Supply, Consumer Access,
Demand and Gap
CVM: WTP,
ATP
Sanitation
and Health
Indoors and
Outdoor Risk
Factor
Statistical
Analysis GIS Maps
Frequency and Percentage
Analysis Factor
Analysis
Data Collection
Primary Data Collection Secondary Data Collection
Data Analysis
93
4.3.3 Tools of Data Collection (Contingency Valuation Method
(CVM)
A comprehensive literature was carried out during the initial stages
and also on regular intervals, throughout the research process. Initial stage of
review, paved the way for the background theme of the research and enable
the determination of the research problem and the objectives of the research.
Knowledge gained from the earlier studies, carried out in relation to the
research title, enhanced a better understanding of the research problem.
Review also helped the researcher in order to get wider perspective of related
issues and methodologies employed for the research. Updating the literature
during various phases of the research, helped to keep track of the latest
developments in research areas and to develop the research at each stage.
Further, the ongoing review showed the way to describe the research problem
to be examined during the study. After deriving the research problem, precise
research objectives were set down. The research objectives were framed to
devise the research questions. A research framework was developed on the
study to describe in detail of the research theme, so as to satisfy the objectives
of the research. This has just been discussed briefly and the framework is
given in Figures 4.2 and 4.3.
In order to achieve the objectives of the study, combinations of
quantitative method have been used to gather information. The people’s
attitude towards paying for water, in some cases such as water sources,
access, quantity, quality, distributor, willingness to pay, affordability to pay,
sanitation, hygiene and health impact and risk factor, have been considered in
the survey through the interview schedule. The information drawn were
analyzed and presented in the respective sections.
94
The quantitative tool selected for study is an interview method
through which the required primary data could be elicited from the
respondents. Since the respondents constitute both literate and illiterate
categories, the methods adopted have been found to be most appropriate.
Interviews have been held with women and men from the households in the 4
zones and 16 wards where interviews have been conducted separately for
respondents and their responses have been immediately noted in the schedule.
There have been some difficulties in the intial stages in getting information
about incomes of the households (respondents,) use of water, sanitation and
health particulars but later when the purpose of the study has been explained
in detail, the respondents felt free to give answers to all the statements
incorporated in the schedule.
The economic concept that Contingent Valuation (CV) surveys
trying to capture is the maximum amount that a respondent would be willing
to pay for the improved water services in the context of the existing
institutional regime within which households are free to allocate their
financial resources (Wedgwood 2003).
The CVM has been increasingly advocated by economists and
water sector specialists as a useful tool for gathering reasonably accurate data
about how much a household can afford and is willing to pay for particular
water and sanitation options presented to them. Early researches in the 1980s
have found that ‘when the CV method is used to estimate the use of goods
and services with which the individuals are familiar, these methods can yield
accurate and useful information on households’ ‘preferences’ when these are
carefully designed and administered. (Wedgwood 2003)
Good quality contingent valuation surveys require both a series of
skills and knowledge, including: survey skills, knowledge of how to develop
95
different CV scenarios for population groups with different water supply or
housing/income conditions, and how to train enumerators and analyse results.
A number of steps are advocated in the guidelines for an effective
CVM survey and dissemination of results. The following are the various
stages and steps involved in the estimation of willingness to pay, using the
contingent method (Figure 4.4).
Figure 4.4 Steps in Contingency Value Method in the Study
Step 1: Select interview technique
Step 3: Develop the CVM Scenario
Step 2: Develop a Sampling strategy
Step 5: Costing the options
Step 4: Decide which elicitation method is good to use
Step 6: Write household survey and CVM questionnaire
Step 7: Implement Survey
Step 8: Data Entry and Analysis
Step 9: Using CVM results
Preparation
Implementation
Data Analysis
96
Application of CVM in environmental economics includes
estimation of non-use values (for example, Walsh et al 1984; Brookshire
et al 1983), non-market use values (for example, Choe et al 1996; Loomis and
du Vair 1993) or both (for example, Niklitschek and Leon 1996; Desvousges
et al 1993) of environmental resources. Though it is a popular non-market
valuation method, a group of academics criticise the method severely for not
being a proper method of estimating the non-market values ( Hausman 1993).
Summaries of CVM studies by different authors reveal that the major
criticism of results of CVM revolves mainly around two aspects, namely:
(a) their validity and (b) their reliability (Smith 1993; Freeman 1993; NOAA
1993).
The validity of CVM is of three types: content validity, criterion
validity and construct validity (Mitchell and Carson 1989; Bateman et al
2002). The content validity in a CV experiment simply refers to the ability of
the instruments included in the scenario to measure the value in an
appropriate manner. Criterion validity of the CV method may be assessed in
terms of another measure, say a ‘market price’ for the same commodity which
may be considered as a criterion. Construct validity has two forms:
convergent validity and theoretical validity. The convergent validity refers to
the correspondence between two measures of the same theoretical construct.
CV results can be said to be ‘theoretically valid’ if the results conform to the
underlying principles of economic theory. In other words, the theoretical
validity involves assessing the willingness to pay (WTP) values of the CV
method by way of regressing the WTP value against standard economic
variables (Mitchell and Carson 1989). Most important, the CVM has however
been used in several studies in India, particularly in relation to water
resources and water uses. Some authors put emphasis on the need to use
97
CVMs in regard to use of water and the willingness to pay for it, in the case of
both irrigation and drinking (for example, Venkatachalam 2004).
4.3.3.1 Contingent valuation method to measure willingness to pay
The WTP value of a good or service may de elicited: (1) directly by
asking consumers, through carefully orchestrated elicitation methods; or
(2) indirectly by examining market prices. The CV method is a survey based
elicitation technique to estimate WTP values of a good that is not traded in the
conventional market.
The CV method is often referred to as stated preference method,
in contrast to revealed preference method, which uses actual revealed
behaviour of consumers in the market. The CV method directly asks
consumers’ WTP for non-marketed goods under a given condition or a
prescribed circumstance. To elicit consumers’ WTP values for non-marketed
goods, a hypothetical market scenario should be formulated and described to
the survey respondents. Thus, the elicited WTP values of a goods are
“contingent upon” the hypothetical market prescribed in the survey
instrument.
Since a CV survey always asks WTP questions, it has been
commonly called a “WTP study.” Subsequently, the key fundamentals of
“contingent” market scenarios are often overlooked by practitioners as the
term “WTP” predominates over “CV method.”
4.3.3.2 Willingness to pay
There are various definitions of willingness to pay, but the most
common one states that: ‘WTP is the maximum amount that an individual
states that he is willing to pay for a good or service’ (DFID 1997).
98
In economics, consumer’s WTP means the maximum amount that a
person would be willing to pay for a service rather than do without it. The
demand curve is: (a) based on the idea that the lower the price of a good, the
more consumers will be willing to pay. The area below the demand curve as
shown in Figure 4.5 (a) represents WTP. The total WTP is not simply the
amount plus the “consumers’ surplus” (Figure 4.5(b)). In this case, the supply
curve shows the production cost of various quantities of the good. The price
times the quantity equals the water system revenue. The shaded triangle
represents the consumer’s surplus, which is not revealed (Ghuraiz and
Enshassi 2005).
Method of Determination of WTP
WTP can be estimated indirectly by observing the behavioural
pattern of the people and their stated preferences.
Observe the prices that people pay for goods in various
markets (that is, water vending, buying from neighbours, and
paying local taxes).
Observe individual expenditures of money, time, and labour to
obtain goods or to avoid their loss. This method might involve
an assessment of coping strategies and involves observations,
focus group discussions and even household surveys.
Ask people directly what they are willing to pay for goods or
services in the future.
99
(a) (b)
Figure 4.5 Demand and Supply Curve Related to Individual
Consumers’ WTP and Consumer Surplus
4.3.3.3 Affordability to pay
Affordability to pay is the extent to which prices (for example,
water supply and sanitation) are within the financial means of users. An
important consideration in service planning relating to choice of service level
and pricing. Fajita et al. (2005) conducted estimates of willingness to pay and
affordability to pay of beneficiaries for water and sanitation services in
Iquitos city, Japan. The main findings of the study are of two categories: WTP
is approximately twice of the current average payment level, and ATP is
higher than the current payment level.
Hicks (1946) classified the ‘consumer surplus measure’ into two
different categories, namely, the compensating variation and the equivalent
variation. For a ‘proposed welfare gain’ due to provision of public good, the
compensating variation refers to the amount of money income that has to be
given up by the consumer to attain increased level of utility (that is, the WTP
measure). The equivalent variation refers to the amount of compensation
required to be provided to the individual so that he/she could attain an
improved utility level in case the provision of the public good does not take
100
place (that is, WTA). For a welfare loss, the compensating variation refers to
the amount of money income that is required to compensate the individual for
the welfare loss experienced (that is, WTA) and equivalent variation refers to
the amount of money income to be sacrificed by the consumer to prevent the
loss from occurring in future (that is, WTP; see Bateman and Turner 1993).
4.3.4 Components of Interview Schedule
The interview starts with questions related to some demographic
characteristics of the households and moves on to family background.
A detailed list of contents in seven parts of the schedule (Appendix 2) is given
below:
I) Profile of Respondents
Name, age, education, community, family size, occupation, type of
family, and ownership of the house.
II) Information on Water
Well details, sources of supply for Metrowater, sufficiency or
otherwise of water sources, water distributor, can water, water problems,
sources of water and uses.
III) Sources of water
Drinking water, cooking water, bathing water, washing clothes in
water sources, distance, quality, quantity, time duration, number of pots,
satisfaction level.
101
IV) Willingness to Pay and Affordability to pay
Regular water supply, WTP, ATP and operation and maintaining
cost.
V) Health details
Health problem, diseases, source of water, expenditures of water
buying, boiling and non-boiling, hospital and treatment cost.
VI) Sanitation and Hygiene
Type of sanitation facility, type of toilet, sewage collected and
treated, septic tank to the nearest well, hands wash before and after eating and
preparing food, hand wash after using the toilet or latrine and body hygiene,
indoors and out door risk factors.
4.4 STATISTICAL TECHNIQUES
4.4.1 Frequency and Percentage Analysis
The analysis of data often begins with what is called a “frequency
and percentage analysis.” For the purpose of description of sample and
respondent related characteristics, a frequency and percentage analysis has
been done for all variables extracted from the questionnaire and put into the
dataset. First, a simple frequency of each of the fields with column
percentages has been made and then two-way tables using certain select pairs
of variables have been carried out, in order to measure variations. The analyst
begins to explore the data, by measuring the central tendency of the data, and
more importantly, the dispersion of the data around this central tendency.
102
Frequency analysis is particularly useful for describing discrete
categories of data having multiple choice or yes-no response formats. This
analysis involves constructing a frequency distribution. The frequency
distribution is a record of the number of scores that fall within each response
category. The frequency distribution, then, has two elements: (1) the
categories of response, and (2) the frequency with which respondents are
identified with each category.
The only technical requirement of the frequency analysis is that the
categories of response be mutually exclusive and exhaustive. This means that
the same observation cannot be counted as belonging to more than one
response category. The frequency analysis must be exhaustive in the sense
that all respondents must fit into a category. The tables so generated are
numerous, only select tables are therefore included in the text while others are
interpreted so as to show the variations therein.
4.4.2 Factor Analysis
Factor analysis is a statistical technique designed to analyse the
interrelationships within a set of variables by reducing the complex data to an
easily interpretable form (Davis 2002). In multivariate analysis, the
bi-variate techniques are extended so that more than two variables can be
considered, the ‘m’ variable becoming the ‘m’ axes of the test space.
Procedures of multivariate analysis are often concerned with the problem of
reducing the original test space to the minimum number of dimensions needed
to describe the relevant information contained in the original observations.
Multivariate procedures differ in the types of original information they
preserve. Some understanding of matrix algebra is essential to using and
understanding the multivariate analysis.
103
A popular multivariate procedure in social science analysis is the
Common Factor Analysis, for which variants are available and are in use in
social sciences as well as in geography. It is a particular psychometric model
that has been in wide use in social sciences. This helps in the study of the
logical implications of systematic inter-correlations within sets of tests.
However, the social sciences follow just one of the many approaches to the
reduction of dimensionality in correlated systems of measurements and the
rotation (varimax, a short form for maximizing variance, for example) of a
reduced number of axes to more meaningful positions.
The Factor Analysis (FA) is also a classification procedure in that it
may be usefully applied to multivariate situations to classifying the N
individuals, on the basis of ‘m’ variables. One particular feature of the FA is
that ‘p’ underlying factors in the multivariate sample space model is always
less than the ‘m’ variables: p < m. The underlying factor dimensions are
drawn from the use of inter-correlations system by generating ‘p’ number of
scores each for the ‘N’ individuals. The scores may however be drawn from
the varimax rotation, which stands for maximizing variance. If we can
measure ‘m’ variables with respect to areal units, the scores may be assigned
to these areal units for constructing one or more maps showing real areal
differences (or regional variations) in respect of ‘p’ reduced dimensions.
The purpose of factor analysis is to interpret the structure within the
variance-covariance matrices of the multivariate data collection made on the
different indicators related to an eco-city related aspects, including
ecofriendliness and sustainability of the city of Gulbarga, in Karnataka, India.
The basic mathematical operations in factor analysis may be stated as follows:
Zj = aj1 P + aj2 P2 + …….+ ajm Pm
104
where
Zj = Xj X mean /Oj or standardized variable
Pi =(I = 1,2,…m) are the principal components and
aj = (j= 1,2,…n) are the coefficients or factor loadings of
(I = 1,2,…m) jth
variable relating to the ith
component.
In other words, each factor is nothing but a linear combination of
weighted variables which can also be expressed as:
P1 = aj Xj where
Thus, in factor analysis, a data matrix containing measurements on
‘m’ variables for each of ‘N’ observations is analysed. The technique uses
extraction of the eigenvalues and eigenvectors from the matrices of
correlations or covariance. The basic mathematical operations in factor
analysis are done with many embellishments on the procedures. FA is a deep
and complex methodology. It is one of the most widely used multivariate
procedures. The model is based on several unique assumptions. For one, the
precise number of factor is assumed prior to the analysis. The factors
extracted, or rather the number of factors, are validated by the variance each
of them explain to the total. There is a progressive decline in the value of
variances with the increasing number of factor dimensions. The first or the
main factor dimension has the highest of the total variance explained and the
bipolar the next highest and so on, resulting in progressively declining
variance.
The analysis begins with the standardization of data. In this
procedure, the data is first converted to standardized, or unitless, form by
subtracting from each observation the mean of the data set and dividing by the
105
standard deviation. The new or the transformed variables will then have a
mean of 0.0 and a variance of 1.0, which is useful in comparing the
distribution of one variable to that of another when the two variables are
expressed in different units of measurement. It provides, in a manner of
speaking, a way of comparing disparate variables.
Since the variables used in any given application are not
immediately comparable, it is necessary to standardize each individual item of
the data, before computing the variance-covariance matrix. The covariance
matrix of standardized variables is nothing more than the correlation matrix,
which in this analysis is referred to as the inter-correlation (similarity) matrix.
Standardization does not have a tremendous influence on the structure of the
variance-covariance matrix and consequently on the results of the FA. In
social sciences, we have no alternatives but to standardize our data, because
the raw matrices of variances and covariances would contain hodge-podge of
measurement units that logical interpretation would be difficult. Hence, there
is a good reason to standardize.
The FA employs principal components, the eigenvectors of a
variance-covariance matrix, as the starting point for analysis. It belongs to the
category of techniques in which utility is judged by performance and not by
theoretical considerations. It relies on a set of assumptions about the nature of
the parent population from which the samples are drawn. These assumptions
provide the rationale for the operations, which are performed and the manner
in which the results are interpreted.
For the purpose of computation in FA, the relationship within a set
of ‘m’ variables is regarded as reflecting the correlation of each of the
variables with ‘p’ mutually un-correlated underlying factors. The usual
assumption is that p < m. Variance of the ‘m’ variable is therefore derived
106
from variance in the ‘p’ factors, but in addition a contribution is made by
unique sources which independently affect the ‘m’ original variables. The FA
refers to the ‘p’ underlying factors as common factors and summarize the
independent contribution as a unique factor.
The FA requires that ‘p’, the number of factors, be known prior to
analysis. This implies that the investigator has some insight into the probable
nature of the factors and can predict a suitable number of factors to be
extracted. The eigenvalue operation in factor analysis is performed on a
standardized variance-covariance or correlation matrix. Hence, the FA used
here is said to be R-mode factor analysis. This assumes not only that all
variables are weighted equally, but also allows us to convert the principal
component vectors into factors. In larger matrices such as ours, the
eigenvalues usually are more uniform for standardized data than for raw data.
And to perform the FA, it is necessary that we convert our unit, or normalize
eigenvalue. The result is a factor, a vector, which is weighted proportionally
to the amount of total variance it represents.
The elements in the factors are referred to as factor loadings. The
eigenvalues represent the proportion of the total variance accounted for by the
eigenvectors. The factor loadings on the other hand are the correlation values
between the old and the new, transformed variables.
If we arrange the factor loadings in a matrix form, we have then a
factor matrix. If we square the elements in the factor matrix and sum within
each variable, the totals are the amount of variance of each variable retained
in the factors. These sums are referred to as the communalities and are
symbolically represented as hj2. The communalities are equal to the original
variances.
107
A specific rule that most factor analysts suggest in the extraction of
factor is that of retaining all factors, which have eigenvalues greater than one.
That is, retain all factors, which contain greater variance than the original
standardized variables. But of course in most instances only a few of the
factors will contain most of the variances in the dataset and hence this
recommendation is useful. If factor theory is applicable to any given dataset, a
few factors should account for a very high percentage of the variance and the
communalities of the variables found under each factor dimension is high.
The FA is said to be reducing the dimensionality of a problem to a
manageable size. However, the meaning of the factors may be difficult to
deduce. This problem is overcome by resorting to maximization of the
variance of the loadings on the factors. This in other words is maximizing the
range of the loadings. This is done in the analysis here by a rotation procedure
called Kaiser’s varimax rotation. The rotation of the factor axes is performed,
iteratively. The analysis also results in factor scores, which represent
estimates of the contribution of various factors to each original observation
(residents). In fact, factors themselves are estimated from these same data.
Thus the computation of factor scores is somewhat a circular process and the
results are not unique. Factor analysis explains in a sense the
interrelationships in a large number of variables by the presence of a few
factors (Kaiser 1958; Harman 1960; Lawrence and Upchurch 1982).
The factor extraction is done with a minimum acceptable
eigenvalue of >1.0 (Kaiser 1958; Harman 1960). The factor loading matrix is
rotated to an orthogonal simple structure, according to varimax rotation. It
results in maximization of variance of factor loadings of the variables. This
procedure renders a new rotated factor matrix in which each factor is
described in terms of only those variables and affords greater ease for
interpretation. Factor loading is a measure of the degree of closeness between
108
the variables and the factor. The largest loading, either positive or negative,
suggests the meaning of the dimension; positive loading indicates that the
contribution of the variables increases with the increasing loadings in a
dimension; and negative loading indicates a decrease (Lawrence and
Upchurch 1982).
4.5 CONCLUSION
The methodology used in the study is an integrated methodology,
where traditional schedule based data collection and processing is integrated
with the modern, statistical as well as qualitative analysis. The former
complements the latter. The methodology which follows the traditions of
social science research and the latest developments in economic research have
the following components:
1. Field survey (primary data).
2. Collection of documented data (secondary data).
3. Statistical approach (simple frequency and percentage, factor
analysis)
4. Analysis and interpretation of domestic water consumers’ data
from Chennai city.
In selecting the most appropriate tool, the following considerations
were useful: the Uses, and the Resources, Familiarity, Significance and
Industry involved. There are several ways of collecting the appropriate data
which differs considerably in the context of money, cost, time and other
resources at the disposal of the researcher. For the present study, both primary
and secondary data have been collected and used for analysis.