The data collection approach was multifaceted, comprising of three principal parts over the course of the following three iterations of the broader longitudinal study:
Iteration 1 – the 2001/2 survey of the top private sector firms in Australia, Canada and Sweden consisted of both qualitative (open-ended) and quantitative (dichotomous and categorical-scaled) questions.
Iteration 2 – the 2005/6 survey of Australia, Canada, and Sweden (public and private sectors), the 2007 survey of Taiwan and the USA, and the 2009 survey of the UK consisted of both qualitative (open-ended) and quantitative (categorical-scaled) questions with the addition of a multi-item Likert scale. The multi-item Likert scale measured respondent attitude towards code implementation, and was developed and based on analysis of the qualitative (open-ended) responses from Iteration 1.
37
Iteration 3 – the 2010/11 survey of Australia, Canada and Sweden, and the 2015 survey of Australia moved to exclusively quantitative questions, consisting of both categorical and multi-item Likert scales (measuring respondent attitude towards code implementation, contribution and content) developed and based on analysis of the qualitative (open-ended) responses from the previous iterations.
In summary, both quantitative and qualitative questions were employed in the first and second iterations of the study, while the third iteration was comprised entirely of quantitative questions with the inclusion of the multi-item scales developed from responses to the open-ended qualitative questions in the first two iterations.
With three exceptions, all iterations of this study consisted of a survey of the top 500 organisations, listed by revenue, operating in the private sector of each country.
Sample frames of the populations of interest were obtained from each country’s authoritative source (see Table 3.1 below). The three exceptions to this method were:
the 2002 private sector of Sweden which used a smaller sample frame of the top 100 organisations by revenue; the 2005 public sector of Sweden which used a disproportionate, stratified sample frame of the top 100 organisations; and the 2007 private sector of Taiwan which required a unique direct approach to individual respondents given the culture and the use of surveys in Corporate Taiwan.
38 Table 3.1: Survey iterations and replications
Survey year Country Sample frame* Responses
(%)
Questionnaire 1995 Australia Di Biasi and McBride
(1993), Australia’s Top 500 Companies
340/500
(68) Seminal study 2000
(Chapter 7)
Australia Shoebridge (2000), The Big List, Business Review
Canada Financial Post (2001),
Financial Post 500 140/500
(28) Iteration 1 2002
(Chapter 7)
Sweden Statistika Centralbyran (2001), Top 100 Private Sector Organisations
74/100
(74) Iteration 1 2005
(Chapter 4)
Sweden Statistika Centralbyran (2005), Top 100 Public Sector Organisations, consisting of:
Top 40/277 National Top 40/289 Municipalities &
20/20 City Councils
Sweden Statistika Centralbyran (2005), Top 500 Public Sector Organisations
Australia Business Review Weekly (2005), Top 500 of the Top
Canada Financial Post (2005),
Financial Post 500 142/500
(28.4) Iteration 2 2006
(Chapter 5)
Turkey Istanbul Sanayi Odasi (2005) 137/500
(23.4) Iteration 2 2007
(Chapter 5)
Taiwan Top 330 companies as determined by revenue and country based researcher
Australia IbisWorld (2010), Top 500 Organisations Operating in Private Sector Australia
106/500
(21.2) Iteration 3 2011 Sweden Statistika Centralbyran
(2011), Top 500 Public Sector Organisations
123/500
(24.6) Iteration 3 2011 Canada Financial Post (2011),
Financial Post 500 91/500
(18.2) Iteration 3 2015 Australia IbisWorld (2015), Top 500
Organisations Operating in Private Sector Australia
125/500
(25) Iteration 3 Note that Chapter 8 is a conceptual paper that details the development of the corporate model of sustainable business practices from an ethical perspective, and as such contains no data.
*All surveys were based on a sample frame of the top 500 organisations listed by revenue, with the exception of the 2002 private sector of Sweden which used a smaller sample frame of 100 as detailed above; the 2005 public sector of Sweden which used a disproportionate, stratified sample frame of 100 as detailed above; and the 2007 private sector of Taiwan which required a unique, more direct approach given the culture and the use of surveys in corporate Taiwan.
39 Each replication of the study involved a survey conducted via a self-administered mail questionnaire. Each recipient was sent a package consisting of: an introductory letter;
details compliant with the research ethics requirements in each country of the study; a copy of the questionnaire; a one-page overview of the background of the study including the researchers involved; and a reply-paid envelope.
The package was intentionally addressed to the public relations manager of each organisation rather than the CEO, to maximise response rates. In contrast with CEOs and other senior executives, public relations managers are more likely to find and make time to provide a considered response as representing the company and its policies (including ethics and corporate social responsibility) is an integral part of their position. Furthermore, in order to ensure that the responses received were actually completed by the intended recipient, rather than a personal assistant, secretary or other subordinate not fully qualified to provide the level of perspective required, it was determined that public relations managers were the most appropriate sample unit for the survey due to greater accessibility.
This choice of addressee was largely based on an examination of other studies where CEOs or equivalent positions were targeted, which generally resulted in lower comparable response rates. For example, Chatov (1980) achieved a 28.1% response rate from 1,000 requests made to CEOs; Davis (1988) a 17.8% response rate; 14.6%
for Hite, Bellizi and Fraser (1988) from letters addressed to CEOs; and Giallourakis et al. (1988) received 84 codes of ethics from 1,000 requests (8.4%). While Schlegelmilch and Houston (1989) managed a 49% response rate from personalised letters addressed to the top 200 of the Times 1000 Company Business Directory, possibly indicating a higher response rate by focusing on the highest achievers (which included CEO level) in the population, such access – both then and now – to personal details of this nature is prohibitive in terms of availability given the increases in protection of an individual’s privacy over time. Furthermore, Matthews (1987) achieved a response rate of 71% from 485 selected manufacturing companies;
however, this was likely a reflection of the judgement-based selection method (and personal involvement) used in determining the sample frame. Lefebvre and Singh (1992) also achieved a noteworthy 48.9% response rate from 461 companies sampled from the Financial Post’s Top 1000 listing in Canada; and Farrell and Cobbin (1994)
40 a response rate of 41.9% in their Australian study. Yet in contrast, the progenitor to this research (Wood, 1997) achieved an impressive 68% response rate from targeting public relations managers.
The higher response rate achieved by Wood (1997), in addition to the longitudinal comparative advantages from maintaining consistency with the seminal research from that study, underpinned the decision to target public relations managers of the highest achieving organisations, based on revenue, from the available sample frames for this study.
However, given the replicative and longitudinal nature of this research, maintaining comparable response rates has been an ongoing concern. The length of the questionnaire and subsequent time commitment required of respondents in considering and answering multiple, open-ended qualitative questions (in the initial iterations) was a substantial motivator in the evolution of the questionnaire to a more concise, quantitative format. This evolution was also supported by the consistency of items captured from the qualitative questions over the course of the second iteration of the survey.