Return to essays page
by Mark Hampton
(Thesis submitted towards PhD)
*** under construction***
I begin, then, by examining in some detail the section upon "Research Methodology" undertaken by Hampton.
He begins with the remark that "The secret nature of offshore finance clearly makes data collection extremely difficult, yet the topic was under-researched and had the potential to address the interesting issue of SIEs' economic development.." This is stated as an assumption. It is not clear from this precisely what is deemed "secret"? Is it client confidentialty? Is it the impossibility of getting suitable figures of client numbers, funds held etc? One matter which cannot be deemed secret, because it needs to be public as a selling point, are the various financial instruments in use in Jersey.
He goes on to state that "in much of the literature on offshore finance, the emergence of OFCs is described and discussed in terms of key factors. These key factors such as taxation, regulation, secrecy, political stability, location et cetera seemed to be a reasonable place to start from when I was at the beginning of the research process. However, as the work progressed over the three year period, the key factors approach seemed to be somewhat lacking in conceptual depth. At one level, these aspects could be arranged as key factors in Jersey's 'story', but I became increasingly interested in diving beneath the surface explanations to the more intellectually challenging depths below."
It appears from this, if I understand him correctly, that rather than producing a theoretical model, which is therefore a simplification, to try and understand the dynamics of offshore centres, there is an assumption that there exists a "real" or "essential" explanation which has more "depth" than mere description.
He states that he realised this gradually, using a "reflective methodology" so that his ideas "grew in a more organic (or perhaps serendipitous) manner." He then states this methodology in some detail, and it is clear that it is sharply divergent from an empirical model.
"Using this methodology in this project, the 'shape', as it were, of the subject being studied only emerged gradually. This is the opposite of the linear hypothesis-experiment-results-discussion type model. The process of refining the emerging construction, rather than setting up the hypothesis first, has been described as falling within the naturalistic or "constructionalist" paradigm (Guba, 1978; Guba and Lincoln, 1981). This alternative to the Positivist-Empiricist paradigm is broadly holistic; it allows for a "mutual simultaneous shaping" of interacting factors rather than simplistic linear cause and effect; and the problematic of generalisations. Eriandsonet al, 1993, argues that it may be the next "paradigm shift" (Kuhn, 1970) in the social sciences. However, to pursue this fascinating area any further lies outside the remit of this thesis. As a overarching paradigm, the naturalistic method of enquiry seemed to suit my purposes better than the Positivist-Empiricist paradigm."
I would like to draw attention to the following adjectives. The normal method of science (creating hypothesis and testing these) is described as " positivist-empiricist". If this is a remarked aimed at a "logical positivism" approach to science, it can be remarked that no scientist would subcribe to that philosophy nowadays. It seems, however, that it is used to simply cast doubt upon the viability of the normal experimental methods, as a kind pseudo-intellectual term to disparage the methods of proper scientific rigour.
Now we move to the actual method by which data was collected and sifted. Hampton remarks:
"The difficulties of researching OFCs and my choice of methodology (particularly the naturalistic method of enquiry) resulted in the need to carefully consider the specific techniques to match these criteria. Given constraints of time, and my particular approach, clearly the entire population of the OFC could not be interviewed. Thus, non-random sampling seemed to be the most promising technique to produce the 'best fit'."
"In addition, there were three other reasons for using non-random sampling methods, first, as noted earlier, a technique such as interviewing might provide the best entrance into such a sensitive topic and be a launch pad for my research. Secondly, I wanted to extract specific technical information that was not in the literature, and thirdly, to get a 'feel' for the area that could not be conveyed by the secondary sources. My own family contacts in the island became an extremely useful starting point. Through family friends who were employed in the OFC, I was able to be introduced to others who might then be willing to be interviewed. This has been called the 'snowballing' technique and has been used successfully to investigate other groups that may be hard to enter (see McCall, 1980, studying artists and Hoffman, 1980, studying the directors of a hospital board). This use-of'snowballing' enabled me to gain access to the networks, as initial personal contacts led to other interview respondents in the OFC. "
So the sampling method used by Hampton was not that of stratified sampling, which is what a proper demographic study of the field might be expected to use, but insread "snowball sampling", which as statisticians are well aware has two major deficiencies: (a) it is biased because sampling units not independent; (b) the sample is not representative of the full population, so that projecting data beyond the sample is not justified., because the sampling error is unknown and cannot be measured (see http://www.statpac.com/surveys/sampling.htm.) for a guide for non-mathematicians).
In fainess, I should comment that there is a legitmate use of "snowball sampling" in data gathering when the population being sampled is secretive and unwilling to divulge information freely. However, as a non-probabilitistic sampling technique, it cleary requires any hypotheses generated from the data to be tested against that derived from other sampling methods, or future studies starting from a different sampling point.