Chapter 3

RESEARCH METHODOLOGY AND DESIGN

3.1 Rationality and scope of the research problem

The ‗new economy‘ is characterized by its substantial use of intangible assets (Bond & Cummins,
2000). Studies done by Blundell, et al., (1999) and Oriani (2005) have shown great progress
comparing the empirical relationship between technological innovation and market value of the
firm. Geroski et al. (1993) posited that there is a positive relationship between the number of
commercially successful innovations in any one year and profit margins of a company. Barney
(1991) argued that in business, successful companies have realized that investing in ‗intangibles‘
of their business is essential to their ability to create higher value products and services. Daum
(2001) advocated that companies in today‘s new economy should not primarily invest in fixed
assets, but in intangibles, since these are today‘s value drivers.

Gould and Gruben (1996) in their study examined the relationship between intellectual property
rights and economic growth and concluded that economic growth is more prominent in countries
with open and developed economies than in countries with closed and developing economies.
However Felvey et al. (2006) posited that in less developed countries, intellectual property rights
do not clearly affect economic growth and most industries in these countries rely on pirated and
imitated technologies. In fact Janjua and Samad (2007) in their study concluded that intellectual
property impact on economic growth is sometimes negative in less developed countries.

Oliner et al. (2007), Marrocu et al. (2009), and O‘Mahony and Vecchi (2009) in their studies for
example, found a positive contribution of intangible assets to both firm and industry productivity.
Studies have also shown that intangible assets significantly contribute to company values in
financial market (Sandner and Block, 2011). Barney (1991) however argued that intangible assets
are becoming a critical source of competitive advantage, but very few empirical works have
actually investigated the factors that may lead firms to undertake this type

124
of technological investment. Galbreath (2005) studied the relationship between tangible and
intangible resources and its contribution to company‘s market performance and suggested that
intangible assets appear to have a greater impact on superior performance than tangible assets.
Prashar and Aggarwal (2009) in their study observed that globally companies are realizing that
intangible assets have the capability of generating revenues, decreasing costs, expanding and
protecting competitive positions, enhancing customer value propositions and increasing the
attractiveness of businesses. Other studies also provide evidence that ―intangible assets are a
fundamental source of competitive advantage for companies in most industries, and investments
made in intangibles consistently improved the financial performance of companies‖ (Garcia-
Ayuso, 2003).

Research study done by Bollen, et al. (2005) linking intellectual capital and intellectual property
to company performance in the German pharmaceutical industry concluded that including
intellectual property in models linking intellectual capital to firm performance enhances the
statistical validity of such models. Namvar, et al. (2010) conducted a similar study, exploring the
impact of intellectual property on intellectual capital and company performance for the Iranian
computer and electronic organizations. Their study though focused on the key dimensions of
intellectual capital like human capital, relational capital and structural capital.

Chang (2006) in his study examined the correlation between firms‘ intellectual capital (including
human capital, structural capital, and social capital), R&D expenditure, intellectual property, and
market performance, based on 2001 - 2005 annual reports from listed Taiwan IT companies
(hardware companies) and found significantly positive or negative correlations with firms‘ market
value (M/B and P/E ratio) and profitability (PM, BEP, ROA, and ROE) in different IT fields.
Literature review revealed that no significant study has been done on the impact of intellectual
property on company performance in the IT industry in India.

Intellectual Property (IP) consisting mainly of intangible assets and their contributions are difficult
to measure. Product based companies are in a comparatively better position to quantify the
contribution of intellectual property (IP) to brand portfolios and the new products launched in the
market by measuring revenues from these products, their market share and profitability. In

125
comparison, service industries without any products, find it difficult to measure the IP contribution
to top-line and bottom-line of the company. Therefore top management in service based companies
does not consider IP management as a strategic issue, but relegate it to the legal department or the
R&D cell. This research study addresses this problem of bringing IP to the strategic level function
for top management decision making. Issues become strategic only when it creates value, which
translates to better market and financial performance. This research study examines the scope of
IP contribution to market and financial performance in service related industry in India. It is for
this reason a knowledge based service industry like information technology has been chosen for
this research study. The IT industry, because of its decades of spectacular success, is recognized
as global brand ambassador for India.

The IT industry for the last twenty years followed an essentially commoditized model of revenue
based on time and resources. Interactions with NASSCOM and FICCI representatives have
revealed that there is a recent shift in the industry to revenue based on business outcomes of
customer business. Clients increasingly want to pay for business solutions and not for the people
and the hours they have logged in. Innovative companies therefore need to move up the ‗value
chain‘ in providing high-end customized solutions for their clients across the globe. There is an
urgent need to develop systems and processes which will help build intellectual property.
Companies have to move to IP-based model. It is in this context, that management of intellectual
property becomes all the more critical for IT industry in India.

Given the scope of the problem, this research study will primarily do an empirical study on how
intellectual property contributes to company performance in the Information Technology industry
in India. This study also examines whether IT companies with higher R&D spend and patent
royalty earnings tend to have higher market valuation and additionally whether IP contribution and
market performance of Indian IT companies (BSE/NSE listed) tend to differ from international IT
companies (NYSE/NASDAQ listed).

In formulating the scope and design of this research study, a pilot study was conducted with a
sample size of ten medium and large IT companies located in Bangalore and Pune. Discussions
were held with the key stakeholders in these companies to bring clarity in the research

126
objectives, formulation of the hypothesis, defining the construct variables, the type of data to be
collected and the questionnaire design for primary data collection. Senior faculty of statistics was
consulted on appropriate statistical and analytical procedures to be adopted for this study. Most of
the large and mid size Indian and global IT companies are located in Bangalore and Pune and
therefore it represents a pan-Indian microcosm of the IT industry. The pilot study provided a deep
understanding of the scope and relevance of this thesis study from the industry and end-user
perspective and helped develop a sound architecture and framework for the overall research.

3.2 Research Questions

Based on the literature review and industry interaction, the following research questions have neem
formulated:
(a) How does intellectual property (IP) contribute to business and market performance in the
IT industry in India?
(b) Can IP be considered a strategic issue for top management?
(c) Do stock markets reward IP-centric companies?
(d) Does IP provide any competitive advantage?
(e) Is there any standardization in valuating and reporting of intangible assets?
(f) Is there a difference between Indian and International IT companies, in terms of IP intensity
and its impact on market and financial performance?

3.3 Research Objectives:

Based on the scope of the research problem and the key research questions, this study has the
following objectives:

(a) To determine whether intellectual property (IP) contributes to market & financial
performance in the service industry like IT
(b) To determine whether IP-centric IT companies have strategic advantages in operational
profitability, competitiveness and performance in the stock market
(c) To verify whether higher R&D activities and royalty earnings from patents contribute to
higher market and financial performance

127
 (d) To determine whether IP intensity and its impact on performance tend to differ between
Indian and International IT companies

3.4 Research hypothesis

Based on the scope of the problem and the research objectives, this thesis study proposes the
following hypothesis:

H0: There is no correlation between intellectual property owned by IT companies and their
business performance.
H1: There is a correlation between intellectual property owned by IT companies and their
business performance.

H0a: Operational profitability of IP-centric IT companies does not correlate with performance in
the stock markets.
H1a: Operational profitability of IP-centric IT companies does correlate with performance in the
stock markets.

H0b: IT companies with higher R&D activities and earnings from patent royalty do not have
higher market valuation and financial performance.
H1b: IT companies with higher R&D activities and earnings from patent royalty have higher
market valuation and financial performance.

H0c: There is no difference between international IT companies (NYSE/NASDAQ listed) and

Indian IT companies (BSE/NSE listed) in intellectual property intensity
H1c: Intellectual property intensity differs between International IT companies
(NYSE/NASDAQ listed) and Indian IT companies (BSE/NSE listed).

H0d: There is no difference in financial performance between international IT companies

(NYSE/NASDAQ listed) and Indian IT companies (BSE/NSE listed).
H1d: Financial performances differ between International IT companies (NYSE/NASDAQ
listed) and Indian IT companies (BSE/NSE listed).

128
H0e: There is no difference between international IT companies (NYSE/NASDAQ listed) and
Indian IT companies (BSE/NSE listed) on company‘s stock market performance.
H1e: Company‘s stock market performances differ between International IT companies
(NYSE/NASDAQ listed) and Indian IT companies (BSE/NSE listed).

3.5 Sample selection & design

Sampling procedure in qualitative research is not as rigid as in quantitative research. Morse (1991)
posits that lack of clear guideline on principles of sample selection has resulted in much confusion
in qualitative research. Morse (1991) argues that if researchers used small random sampling in
qualitative studies, it would violate the quantitative principle that requires adequate sample size to
ensure representativeness and also violate the qualitative principle of appropriateness that requires
purposeful sampling of good informant. The respondent in a qualitative study should be
knowledgeable, reflective, articulate and willing to share with the interviewer (Moses, 1991). In
qualitative study research sample selection has a profound effect on the quality of the research
(Coyne, 1997). Researchers have been criticized for not describing their sampling strategies in
sufficient detail, which makes interpretation if the findings difficult (Kitson, et al. 1982).

Schatzman & Strauss (1973) state that ―selective sampling is a practical necessarily that is shaped
by the time the researcher has available to him, by his framework, by his starting and developing
interests and by any restrictions placed on his observations by his hosts‖. Patton (1990) also
propounded similar sampling criteria called the ‗purposeful sampling‘. The logic and power of
purposeful sampling lies in selection of information-rich cases for in depth study. Information-rich
cases are those from which one can learn a great deal about issues of central importance to the
purpose of the research (Patton, 1990). Sandelowski et al. (1992) suggest that for qualitative
research there exists selective and theoretical sampling. ―Selective sampling refers to a decision
made prior to beginning a study to sample subjects according to a preconceived but reasonable
initial set of criteria (Sandelowski et al. 1992).

129
This research study being primarily qualitative, does sample selection based on selective
sampling. This study has a total sample size of 30 companies (large/medium size IT companies)
which has been obtained from NASSCOM member directory of IT companies located in India.
Primary data for ‗Intellectual Property Score‘ value was collected from IPR representative
respondents of all the 30 companies. The list of 30 IT companies chosen for this study is given in
Table 3.1. The sample size of 30 companies in this research study fulfills the stringent sample
selection criteria and represents most of the leading IT companies in annual revenues.

Table 3.1
List of 30 IT companies chosen for this study
Ser.No. Company Name Ser. No. Company Name

1 Infosys Technologies 16 HP
2 WIPRO 17 IBM
3 Zensar Technologies Ltd. 18 Oracle International Corp.
4 Patni Computer Systems 19 Symantec Corporation
5 Mindtree Ltd. 20 Unisys
6 Mphasis Ltd. 21 BMC Software
7 Tata Consultancy Services 22 Accenture
8 KPIT Cummins Infosystems 23 Qualcomm Incorporated
9 Hexaware Technologies 24 Red Hat Linux
10 Geometric 25 Sybase Software
11 Onward Technologies 26 Tata Elxsi Ltd.
12 Sonata Software 27 Sasken Communication Tech.
13 Sterlite Technologies Ltd. 28 Cisco Systems Inc.
14 Persistent Systems Ltd. 29 Tech Mahindra Ltd.
15 Amdocs 30 Yahoo Incorporated

As per NASSCOM Report 2005, for the year 2003-04, only 41 Indian IT companies had annual
sales of more than US 50 million dollars, and only 17 IT companies had annual sales of more than
US 100 million dollars. Bulk of the Indian IT companies (2,653 in numbers) had sales of less than
US 2.1 million dollars. This implies that less than 1% of the IT companies had earned more than
US 100 million dollars. Though the annual revenues earned by these companies have gone up
considerably since then, the same set of companies dominate the Indian market. As per
NASSCOM, the top 5 Indian IT companies (TCS, Infosys, WIPRO, HCL Technologies and

130
Mahindra IT Business Services) have collectively earned revenues of $22.5 billion in the financial
year 2011-12. This sample size of 30 companies in this study is representative of the top revenue
earners, and is adequate for the non-parametric statistical analysis tools used for this research
study. Both primary and secondary data has been used in this research study.

3.6 Sample selection criteria

The sample selection criteria used for this study is based on selective sampling:

(a) IT companies should have pan India operations with global presence
(b) Companies should be among the top 50 IT companies in terms of annual
revenues
(c) Companies should own patents in their names, besides having other intellectual
property rights, like copyrights, trademarks and proprietary technology, etc.
(d) Companies should be listed in either Indian stock exchange (BSE/NSE) or USA
stock exchange (NYSE/NASDAQ)
(e) Companies should have a designated senior manager who is responsible for
intellectual property management

The companies in this study are limited to Indian IT sector, so that a homogeneous sample can be
examined. Furthermore as the literature review revealed, and also validated by NASSCOM, no
similar study has been done for the Indian IT service sector. This research study relies on both
primary and secondary data for various statistical analysis to test the proposed hypothesis. A
detailed questionnaire was designed with inputs from stakeholders during the pilot study to collect
data on ‗IP score‘ and was administered to senior managers of the company responsible for IPR
management and other IP related activities. Sample questionnaire is shown in
Appendices Table A7.

131
3.7 Dependent variables:
3.7.1 Price-to-Book ratio (P/B ratio)
This research study uses as dependent variables, financial performance measures based on Value
Added Intellectual Coefficient (VAIC) market valuation and profitability, which are represented
by two key financial ratios, ‗Price-to-Book ratio‘ (P/B ratio) and ‗Operating Margin‘ (OM).
Value Added Intellectual Coefficient was first made public by Pulic (1998) and further developed
by Barnemann (1999). It gives a new insight to measures of value creation and monitors the value
creation efficiency in companies using basic accounting figures. ―The VAIC index examines the
association with three measures of corporate performance, profitability, productivity and market
valuation‖ (Firer & Williams, 2003). ‗Price-to-book‘ value is represented by the difference
between market value of the firm and its book value. This difference can be attributed primarily to
the market valuating the company‘s intangible assets like intellectual property, more than its
tangible physical assets. The book value depends on the national or international accounting
standard under which the accounts have been prepared. Price-to-Book value of more than one
means that the market value of the company is greater than its book value. This difference is
sometimes called the market value added or market goodwill, which would mainly comprise of
the firm‘s intangible assets.

3.7.2 Operating margin (OM)

Operating Margin indicates how much profit a company makes after paying for variable costs of
production such as wages, raw materials, etc. Operating margin shows how efficiently the
managers of a firm are using business operations to generate profit. In the IT service industry, a
higher operating margin also reflects the company‘s capability to command a higher client billing
margin mainly because of its intellectual capital and assets vis-à-vis the competition. These two
dependent variables have been drawn from data published in CRISIL Research and National Stock
Exchanges, Forbes Research Bureau (USA) and Annual Reports published by the respective
companies, which were further crosschecked with financial data analysis report by Dion Global
Solutions and Money Control Financial Research. Keeping in mind the fluctuations in the stock
market prices and financial performances of companies, a five-year average value has been
compiled for the 30 listed IT companies (April 2007 to March 2011).

132
3.8 Independent variables:
3.8.1 Intellectual Property Score (IP score)
The following independent variables have been used to measure the total IP value creation and
contribution to a company, which is termed as „IP score‟. The construct variables are: ‗IP
contribution to profitability‘, ‗IP contribution to turnover‘, ‗IP contribution to market share‘,
‗R&D spend‘, ‗Royalty form patents‘, ‗IP impact on service value chain‘, ‗IP impact on client
billing margins‘ and ‗Business model link with patent‘. The response to these variables was
obtained through a questionnaire administered to IP managers of 30 IT companies. Response to
these eight variables was given on a scale of 1 to 5 (1 being the lowest and 5 being the highest).

These values were finally added up to arrive at the ‗IP score‘. The IP score of 30 IT companies is
shown in Table 3.2.

Validity and reliability of data are two fundamental elements in the evaluation of any statistical
measurement. Validity is the extent to which an instrument measures what it is intended to measure
and reliability is consistency of that measure. A statistical tool called Cronbach‘s alpha test has
been used to check the validity and reliability of primary data obtained in response to the
questionnaire in this research study. Cronbach‘s alpha score of 0.82 has been obtained, which
denotes a high internal consistency and reliability of data. Cronbach‘s alpha SPSS reliability
analysis results are shown in Appendices Table A1.

133
 Table 3.2

Data on „IP score‟ for IT companies in India

Company Name IP_ IP_ IP_ R&D IP_ Royalty IP_ Patent_ IP
Profit Turn Market Spend Service Patents Billing Business Score
Accenture 4 4 4 4 5 3 4 5 33
Amdocs 4 4 3 4 3 4 3 3 28
BMC Software 5 4 5 4 3 4 4 5 34
Cisco Systems 4 4 4 5 4 4 4 5 34
Geometric Ltd. 3 1 1 4 1 3 3 2 18
Hexaware Technologies Ltd. 3 2 2 3 4 1 4 3 22
HP Ltd. 2 3 2 3 2 3 1 5 21
IBM 4 3 3 5 4 3 3 3 28
Infosys Technologies Ltd. 4 4 4 4 3 2 4 3 28
KPIT Cummins Infosystems 3 3 3 3 3 2 3 3 23
Mind Tree Ltd. 4 4 3 4 4 3 4 4 30
Onward Technologies Ltd. 4 3 3 4 3 4 2 3 26
Mphasis Ltd. 3 2 4 4 4 2 3 3 25
Oracle International Corp 4 5 5 4 4 5 4 4 35
Patni Computers Systems 3 3 1 3 4 1 4 4 23
Persistent Systems Ltd. 4 3 3 4 3 3 4 4 28
Qualcom Incorporated 4 3 2 3 3 5 3 5 28
Red Hat Linux 5 4 4 3 4 3 3 5 31
Sonata Software 1 2 1 3 3 1 1 4 16
Sterlite Technologies Ltd. 4 4 4 4 5 2 4 5 32
Sybase Software India 4 3 3 4 3 2 4 3 26
Symantec Corporation 4 2 3 3 3 4 3 2 24
Tata Consultancy Services 4 3 4 3 4 2 3 4 27
Tata Elxsi Ltd. 2 3 2 4 3 1 2 1 18
Tech Mahindra Ltd. 1 1 2 4 1 1 1 3 14
Unisys Global Services India 4 4 4 4 5 3 4 5 33
WIPRO Ltd. 3 3 4 3 4 1 4 3 25
Yahoo Software 4 4 4 4 4 2 4 4 30
Zensar Technologies Ltd. 2 3 2 3 4 1 2 4 21
Sasken Communications 3 1 1 4 1 3 3 3 19

(Variables: IP_profit = IP contribution to profitability, IP_turn = IP contribution to turrover,

IP_market = IP contribution to market share, R&D_spend = R&D/patent spending, IP_service =
IP impact on service value chain, Royalty_patents = Royalty from patents, IP_billing = IPimpact
on client billing margins, Patent_business = Business model link with patent)

Scale: 1- 5 (1 being the lowest, 5 being the highest)

134
3.8.2 Intangible Assets Ratio

Another independent variable, called the ‗Intangible Assets Ratio‘ (AIR) has been used to
revalidate the results using the primary data in the research study. AIR is a new financial measure
conceived by Hewitt Heiserman (2004) which is the percentage of total intangible assets to total
assets of a company. Intangible assets include all types of intellectual property owned by a
company including goodwill. AIR is comparatively new and not currently reported as a separate
ratio by Indian companies in their annual reports, though it can be calculated using two financial
data, reported by companies (intangible assets/goodwill and total assets). Intangible assets ratio
indicates how effective a company is in creating value driven intangibles utilizing all its tangible
assets. It also reflects a company‘s innovativeness and how it creates and commercializes
intangible assets like intellectual property.

A theoretical question which arises is what should be the optimum intangible assets ratio for a
company. As per Heiserman (2004), anything over 20% is worrisome "because management might
be overpaying for the acquisitions that gave rise to the goodwill‖. This could be especially relevant
for mergers and acquisitions and how much the intangibles are valued and paid for. However there
are knowledge-intensive companies which are highly successful and are likely to have intangible
assets ratio, much higher than 20 percent. This also holds true for some of the sample IT companies
in this study. Intangible assets ratio is particularly relevant for the IT companies, as intangibles
play a more prominent role is service delivery to the clients than the tangibles. In this research
study, a five-year average (April 2007 to March 2011) of the intangibles assets ratio of 30 IT
companies were obtained from the respective company published financial documents and annual
reports. Data on Intangible Assets Ratio, Price-to-Book ratio & Operating Margin of 30 IT
companies is shown in Table 3.3.

135
Table 3.3

Data on Intangible Assets Ratio, Price-to-Book ratio & Operating Margin

of IT companies in India
5-Year Avg. 5-Year
Ser.No. Company Name Intangible Average 5-Year Avg.
Assets Ratio P/B ratio OM
1 Infosys Technologies 19.96 6.76 33.34
2 WIPRO 15.71 4.09 22.44
3 Zensar Technologies Ltd. 15.75 1.04 17.75
4 Patni Computer Systems 11.13 1.71 32.49
5 Mindtree Ltd. 6.54 3.10 18.18
6 Mphasis Ltd. 8.62 4.42 25.26
7 Tata Consultancy Services 12.28 6.38 28.45
8 KPIT Cummins Infosystems 4.67 2.62 22.92
9 Hexaware Technologies 10.60 0.93 19.90
10 Geometric 1.93 1.50 13.47
11 Onward Technologies 10.15 1.17 9.23
12 Sonata Software 13.77 1.82 25.74
13 Sterlite Technologies Ltd. 0.21 0.76 11.72
14 Persistent Systems Ltd. 7.20 2.07 27.91
15 Amdocs 41.69 1.60 38.22
16 HP 42.81 1.34 23.78
17 IBM 25.42 9.91 51.09
18 Oracle International Corp. 42.15 3.58 78.99
19 Symantec Corporation 55.01 2.49 90.42
20 Unisys 7.37 2.60 32.51
21 BMC Software 45.69 3.71 26.60
22 Accenture 7.19 9.63 31.41
23 Qualcom Incorporated 17.93 3.34 69.49
24 Red Hat Linux 26.08 5.65 88.69
25 Sybase Software 37.19 5.17 78.00
26 Tata Elxsi Ltd. 16.87 4.39 21.70
27 Sasken Communication Tech. 29.35 1.01 18.13
28 Cisco Systems Inc. 20.65 2.16 65.80
29 Tech Mahindra Ltd. 2.13 4.26 24.33
30 Yahoo Incorporated 28.10 1.43 82.20

136
3.9 Statistical Analysis Tools
3.9.1 Rationality for using Non-Parametric Tests
In statistical investigation, hypothesis testing plays a major role. Statistical analysis attempts on
examining the truth or otherwise of hypothesis about some features of one or more populations.
Almost all large and small sample tests are based on the assumptions that the parent population,
from which the sample is drawn, has a specific distribution, such as normal distribution. Non-
parametric tests do not require such assumptions and are known as distribution free tests. The term
non-parametric means there are no parameters involved in the traditional sense of the term.
―Non-parametric tests utilize some simple aspects of sample data such as signs of measurement,
order relationships or category frequencies‖ (Neave and Worthington, 1992).

The statistical inference drawn from parametric tests may be seriously affected when the parent
population distribution is not normal, especially if the sample size is small. Thus if there is a doubt
about the distribution of the parent population, a non-parametric method should be used. Moreover
in social and behavioral sciences, observations are difficult or sometimes impossible to quantify
on numerical scales and hence are suitable for non-parametric testing. The statistical tools used for
this study are Cronbach‘s alpha, Factor analysis, Spearman‘s rho rank correlation coefficient,
Kendall tau rank correlation coefficient and Mann-Whitney U test. The rationale for using these
statistical analysis tools stems from the fact, that the data captured in the study for independent
variables are qualitative in nature and best suited for non-parametric tests. The inherent strengths,
advantage and relevance of these statistical analysis tools used in this study is hereby explained:

3.9.2 Cronbach‟s alpha

Validity and reliability of data are two fundamental elements in the evaluation of any statistical
measurement. ―Validity is concerned with the extent to which an instrument measures what it is
intended to measure. Reliability is concerned with the ability of an instrument to measure
consistently. It should be noted that the reliability of an instrument is closely associated with its
validity. An instrument cannot be valid unless it is reliable‖ (Tavakol and Dennick, 2011).
Cronbach‘s alpha, developed by Lee Cronbach in 1951 is a test to measure the reliability of a set
of questions in a survey instrument. It is a measure of the internal consistency of a test or scale. It

137
is expressed as a number between 0 and 1. Alpha is an important concept in the evaluation of
assessments and questionnaires. It is important that assessors and researchers should estimate this
quantity to add validity and accuracy to the interpretation of their data. The closer
Cronbach‘s alpha is to 1, denotes a high internal consistency and reliability of data. Internal
consistency describes the extent to which all the items in a test measure the same concept or
construct and hence it is connected to the inter-relatedness of the items within the test. Cronbach's
alpha mathematical formula is represented as:

Alpha = rk / [1 + (k -1) r]

Where k is the number of items considered and r is the mean of the inter-item correlations the size
of alpha is determined by both the number of items in the scale and the mean inter-item
correlations. Gliem, G. and Gliem, R. (2003) provide the following rules of thumb for interpreting
Cronbach‘s alpha ―> 0.9 = Excellent, _ > 0.8 = Good, _ > 0.7 = Acceptable, _ > 0.6 =
Questionable, _ > 0.5 = Poor, and _ < 0.5 = Unacceptable‖. The increase in value of alpha is
partially dependent upon the number of items in the scale, but this has diminishing returns. It
should be noted that an alpha of 0.8, achieved in this study is good and a reasonable goal, especially
using Likert-type scales for internal consistency and reliability of data. SPSS
Cronbach‘s alpha test results are shown in Appendices Table A1.

3.9.3 Factor Analysis

Factor analysis is a statistical technique used to analyze interrelationships among a large number
of variables in order to explain these variables in terms of their common underlying dimensions
(factors). The approach involves condensing the information contained in a number of original
variables into a smaller set of dimensions (factors) with a minimum loss of information. ―Factor
analysis tests for interdependence, in which all variables are simultaneously considered, each
related to all others, therefore reducing number of variables to a smaller number of factors for
modeling purposes. Also it selects a subset of variables from a larger set, based on which original
variables have the highest correlations with the principal component factors‖ (Hair et al., 1998).

138
―Factor analysis can be used for exploratory or confirmatory purposes. Exploratory factor
analysis seeks to uncover the underlying structure of a relatively large set of variables. The
researcher's initial assumption being that any indicator may be associated with any factor. This is
the most common form of factor analysis. Confirmatory factor analysis seeks to determine if the
number of factors and the loadings of measured (indicator) variables on them conform to what is
expected on the basis of pre-established theory‖ (Hair et al., 1998).

The starting point in factor analysis, as with any other statistical techniques, is the research
problem. Factor analysis can identify the structure of relationships among either variables or
respondents by examining either the correlations between the variables or the correlations between
the respondents. For example, in this research study, we have data on 30 respondent companies in
terms of 8 key characteristics. Therefore to summarize the characteristics, the factor analysis is
applied to a correlation matrix of the variables. ―This is the most common type of factor analysis,
and is referred to as R factor analysis. R factor analysis analyzes a set of variables to identify the
underlying dimensions. Alternatively factor analysis also may be applied to a correlation matrix
of the individual respondents based on their characteristics. This is referred to as Q factor analysis,
a method of combining or condensing large numbers of people into distinctly different groups
within a larger population. The Q factor analysis approach is not generally used‖ (Hair et at., 1998).

In this study based on the Factor analysis, two factors were extracted:

(with Eigen values larger than 1)

Factor 1 – IP impact on Market Performance
Factor 2 – R&D & Royalty impact on Market Performance
The SPSS Factor analysis report is shown in Appendices Table A2.

3.9.4 Spearman‟s rank correlation coefficient (R)

The objective of rank correlation methods is to assess the degree of monotonicity between two or
more series of paired data. ―Monotonicity means a tendency for the values in the series to increase
or decrease together (positive correlation) or for one to increase as the other decreases
(negative correlation)‖ (Da Costa and Soares, 2005). They are applicable to paired data, where
there is some connection between corresponding members of the samples. ―Rank correlation

139
coefficients are less restrictive than other methods (e.g. Pearson‘s correlation coefficient), because
they do not try to fit one particular kind of relationship, linear or otherwise, to the data.
They achieve this by using the ranks of the sample values rather than the values themselves‖ (Da
Costa and Soares, 2005). A common rank correlation coefficient test is Spearman‘s rank
correlation coefficient, which is a non-parametric test. The main strength of Spearman‘s rho is
that, contrary to other correlation methods, it can be used not only on numerical data but on any
data that can be ranked. ―An example of the use of such methods is the analysis of sales data
where the aim is to assess whether there is correlation between marketing activities (i.e. visits to
clients) and the number of sales‖ (Neave and Worthington, 1992) Another significant advantage of
Spearman‘s Rho is that it can be used to derive a correlation coefficient even for smaller sample
size (between 10 and 30 pairs of measurement). Spearman‘s rho mathematical formula
(R) is represented by:

Where d = the difference between X and Y rank order value and n = number of paired observations.
The use of ranks allows us to measure correlation using characteristics that cannot be expressed
quantitatively, but they can be ranked. The interpretation of Spearman‘s Rho is similar to Pearson‘s
correlation coefficient (R), which will take values between -1 and +1 (Davis and Pecar, 2010). For
pairs of data to be considered a strong relationship, one needs to confirm that spearman‘s
coefficient is significant. For this one needs to determine the critical value at the significance level,
alpha = 5% (0.05). The critical value of R may be found either from a table of critical R values or
by calculation depending upon the sample size.

3.9.5 Kendall‟s tau rank correlation coefficient

Kendall's Tau is also a measure of correlation, and measures the strength of the relationship
between two variables. Kendall's tau, like Spearman's rank correlation is carried out on the ranks
of the data. Kendall‘s rank correlation is a good substitute for Spearman‘s rank correlation if there
are a lot of ties in the rankings (for example several pairs of samples having the same ranks). As
with other measures of correlation, Kendall's tau will take values between -1 and +1,

140
with a positive correlation indicating that the ranks of both variables increase together whilst a
negative correlation indicates that as the rank of one variable increases the other one decreases.

Spearman's rank correlation is a more widely used measure of rank correlation because it is much
easier to compute than Kendall's tau. In most cases Spearman‘s rho R value is larger than Kendall‘s
tau value, except when the deviations are larger. ―The main advantages of using
Kendall's tau are that the distribution of this statistic has slightly better statistical properties and
there is a direct interpretation of Kendall's tau in terms of probabilities of observing concordant
and discordant pairs‖ (Conover, 1980). This research study uses both Spearman‘s rho and
Kendall tau for better statistical analysis and validation.

The Kendall coefficient is denoted with the Greek letter tau (τ) and is mathematically represented
as:

τ = (4P / (n * (n - 1))) - 1

Where P is the number of concordant pairs and is calculated as the sum over all the
items, of items ranked after the given item by both rankings and n is the number of
observations.

3.9.6 Mann-Whitney U test

Mann-Whitney U test (also known as Wilcoxon sum of ranks test) is a non-parametric statistical
test for assessing whether a difference exists between two ‗groups‘, in whatever way a ‗group‘ is
defined. Mann-Whitney U test is often viewed as non-parametric equivalent of student‘s t-test. It
is ideally dependent on random selection of subjects into their respective groups. Normal
distribution of data is not necessary for this test. This test has good probabilities of providing
statistically significant results and can be used for very small sample size (between 5 and 20
observations). ―Mann-Whitney test has approximately 95% of the Student‘s t-test statistical
power‖ (Landers, 1981). ―By comparison with the t-test, Mann-Whitney test is less at risk to give
wrongfully significant result in presence of one or two extreme values in the sample under
investigation‖ (Siegel and Castellan, 1988). In addition, the Mann-Whitney U test is more

141
powerful than the t-test in exceptional circumstances. ―It is more powerful in the detection of a
difference on the extent of the possible differences between population averages than the t-test
when a small manpower is associated with a small variance‖ (Zimmerman, 1987).

Mann-Whitney U in SPSS, statistically also tests the difference in two groups in terms of mean
and medium. In statistics and mathematics, the three most commonly used measures of central
tendency (or average) are mean (or arithmetic mean), median and mode. Mean is the sum of a list
of numbers divided by the number of numbers. There could be some extreme values in the dataset,
called outliers, which have a tendency to skew the data distribution. In this case we would be use
an alternative method to calculate the average, which is called the median.

The median is literally the middle number, if we list the numbers in order of size. The median is
not susceptible to extreme values like mean, but is not suitable for continuous data variables. The
mode is defined as the number that occurs most frequently in the dataset and can be used for both
numerical and nominal data variables. A major problem with the mode is that it is possible to have
more than one modal value representing the average for numerical data variables (Davis and Pecar,
2010). In this research study, the mean and the medium measures of central tendency have been
used because of the nature of the data and its inherent advantages.

Mathematically, the Mann-Whitney U statistics are defined by the following, for each
group:

Ux=nxny+((nx(nx+1))/2)−Rx (1) Uy=nxny+((ny(ny+1))/2)−Ry (2)

Where nx is the number of observations or participants in the first group, ny is the number of
observations or participants in the second group, Rx is the sum of the ranks assigned to the first
group and Ry is the sum of the ranks assigned to the second group.

142
3.10 Concluding Remarks

In Chapter 3, the research methodology for the main study has been discussed. Also explained is
the research study design, sample selection criteria, dependant and independent variables, the
various non-parametric statistical analysis tools used to test the entire hypothesis. In Chapter 4, the
statistical analysis and testing of the entire hypothesis along with research study findings have
been discussed in length.

143