METHODOLOGY

Introduction

The research methodology used was based on the 7-step hypothetico-deductive process outlined by Sekaran (1992);
  • 1. Observation.
  • 2. Preliminary information gathering.
  • 3. Theory formulation.
  • 4. Hypothesising.
  • 5. Further scientific data collection.
  • 6. Data analysis.
  • 7. Deduction.

    The first step is the observation that there are phenomena important enough to be the subject of research. In this case, it the author's observation from his own experience prompted him to significant importance of technology transfer to many companies operating in Ireland. It was observed that difficulties could be encountered which complicate the technology transfer process. Therefore, it is a subject worthy of in-depth research.

    Secondary sources of information were used for preliminary information gathering. An extensive literature review was conducted to find out the extent of knowledge of the research area. The full review is included in Chapter 2. Based on the information gained in the review, theories were formulated and hypotheses developed. Several key questions were asked and answered during these two steps. The first was whether the study would be primarily qualitative or quantitative, what primary sources should be used and what method should be used for collection of data from primary sources. These questions are discussed below. Further reviews of the literature were made to determine the optimum primary data collection method. Hypotheses are deduced from the literature and are compared with the results of this study in Chapter 5.

    The next step is the collection of data from primary sources. The method chosen was a mail survey and the procedure followed is described below. The data are then analysed and this analysis is included in Chapter 4.

    Research Objective

    The objective of this research is to ascertain how successfully multinational pharmaceutical and medical device companies have managed the process of technology transfer into Ireland. The questions to be asked were determined by analysing secondary sources of data.

    Secondary Sources of Data

    The secondary sources of data used for this study were the many studies conducted by others in the past. A summary of this body of knowledge is included in
    Chapter 2. Lists obtained from the IDA database of pharmaceutical and medical device companies provided most of the information needed for the survey. References to other studies that are most important as secondary sources of ideas are included below in the description of the development of the questionnaire.

    Quantitative of Qualitative?

    The primary focus of research can either be quantitative or qualitative depending mainly on how well defined the objectives of the study are. A quantitative study would be used for causal and predictive studies when the dimensions of the problem are well defined. A qualitative study would be used for exploratory studies when the dimensions are less well defined.

    The study of secondary sources of information found that many studies of technology transfer between affiliates of multinational companies. For this reason the methods of conducting technology transfer and the various approaches taken are well defined. The literature review revealed many useful questions to ask and provided a number of hypotheses to be tested. It was preferred that questions related to the benefit of various strategies could be answered with certainty. For these reasons, a primarily quantitative approach was preferred.

    Typically quantitative studies are of a large sample. Small samples are used for qualitative studies. Quantitative studies are used to verify hypotheses while qualitative studies are used to discover theory. Questions are pre-set rather than open. The researcher is uninvolved. There may be a degree of researcher involvement in qualitative studies. Facts are gathered, not questioned. Statistical data analysis is used. There is focus on the content and not on action or process.

    Primary Data Collection

    The first decision to be made about data collection is the choice of research methodology. The choice of mail survey is discussed by comparison with other research methodologies and with the methodologies used by other researchers of the subject area. The mail survey procedure followed is described in detail.

    Choice of Research Method

    Typical quantitative studies require larger samples than qualitative studies. Questions are usually pre-set. In quantitative studies questions may be open-ended. In quantitative studies the researcher is usually not involved. Data collection methods usually used for quantitative studies observation, field and laboratory experiments, computerised data bases and surveys. The types of surveys used for quantitative analysis are mail surveys, administered questionnaires, and telephone interviews. Methods suitable for qualitative analysis include in-depth, structured and focus group interviews, and case studies, grounded theory and action research.

    Mail Survey

    Mail survey was chosen for the research. It is well suited for gathering a lot of data using pre-set questions. The author chose not to include his own company's experiences in the final study to maintain the uninvolvement preferred for hypothetico-deductive research. It is the most suitable method logistically to survey the entire population of 142 firms throughout the Republic of Ireland. The names of the firms were obtained from lists of Multinational Healthcare and Pharmaceutical firms manufacturing in Ireland maintained by the IDA. The information provided included the name of the managing director or equivalent. The questionnaire was posted to this person, who has the option of filling it in or passing it to another appropriate individual.

    Other advantages of mail survey are that its results can be generalised and there is a reduction in interviewer bias versus personal interviewing (Weiers, 1988). The ability of the respondent to fill in the questionnaire anonymously should also improve the objectivity of responses. The large number of respondents to the mail survey allows the bias of individual responses to be less apparent than would be the case of a more limited survey method.

    Other Methods

    Mail survey was chosen because it was considered the most suitable method to economically obtain a large number of data for conducting a quantitative analysis of technology transfer. Other methods suitable for quantitative methodologies are less attractive for this study. Experiments are not feasible because a typical international technology transfer involves many people travelling great distances and spending a lot of money over a considerable period of time. The author could not devise a method for the simulation of technology transfer. Direct and participant observation are not feasible for similar logistical reasons. Computer data base analyses are not available. Most companies are secretive with information related to sensitive subjects such as transferring proprietery technology.

    There are two other types of survey considered. An administered questionnaire was not suitable logistically because there were insufficient human resources available to administer the number of surveys needed to conduct statistically valid analysis. It would also be difficult to schedule interview times. Telephone interviews were not used because of the time required to get in touch with the correct person. Managing directors are busy and many are likely to resent the imposition on them needed to answer the questions. Telephone surveys are limited in the range of questions that can be asked. The person phoned may not be the one most suitable for answering the survey. A mailed survey could be passed on, while the necessity to make additional phone calls would increase the difficulty of administration. The person to be surveyed may not be in the office when called.

    Of the methodologies suitable for qualitative analysis, the case study was considered because of the author's experience with an earlier technology transfer. This methodology would have resulted in a different type of study. It was not chosen because there was already sufficient information in the literature to develop hypotheses for a hypothetico- deductive analysis. It is doubtful that any new information would be elucidated by a single case study. In addition to this problem, there may have been considerable difficulties for the author to produce an objective study because of his continued employment for the company. A study with a number of case studies could be effective. There were insufficient resources available to conduct such as study.

    The Methodologies used in Relevant Studies

    A mail survey was used by Zander and Kogut (1995) for their study of forty-four innovations by Swedish companies (80% response rate). Taggart (1998) used a mail survey for their survey of 500 affiliates of multinational companies in the UK to examine R&D (34.2% usable responses). Mudambi (1998) sent 226 questionnaires to multinational engineering companies in the UK West Midlands in his study of the role and duration of multinational investment strategies (31.2% usable response). Szulanski (1996) received a response of 271 questionnaires (60%) in his study of the impediments to the transfer of best practice. Ghoshal and Bartlett (1988) mailed a questionnaire to 438 American and European multinational companies in the third phase of their study of adoption and diffusion of innovations by multinational corporations. Oh these just 66 were usable in all respects (15%).

    Methods with a higher sample cost were used in other studies. Ghoshal and Barlett (1988) interviewed 184 managers in nine countries in their study of the creation, adoption and diffusion of innovations by the subsidiaries of multinational corporations. This study had both quantitative and qualitative method. Subramaniam et al. (1998) conducted multiple open-ended interviews at 13 multinational company affiliates in Japan in their study of global new product development processes. Teece (1977) also used an unspecified methodology with a high sampling cost in his survey of twenty-six technology transfer projects.

    The resources were not available to conduct similar high sample cost methodologies. However, for further more qualitative studies in this area, a methodology similar to that used by Subramaniam et al. (1998) with focus groups or multiple interviews in a number of companies could be considered.

    Response Rate

    The great advantage of mail survey is the ability to easily survey a large number of people over a wide geographical area. The major disadvantage is a poor response rate. Much has been written about ways to improve this and how to correct the results to allow for the non-respondents. The latter issue will be dealt with in the results analysis section. Many suggestions are made about methods of improving the response rate and some of these methods are adopted for this survey. Jobber and O'Reilly (1995) found that prior (i) telephone notification, (ii) monetary incentives, (iii) non-monetary gifts, (iv) stamped addressed return envelopes, (v) promises of anonymity, and (vi) follow-up questionnaires improved response rates. A shorter length of questionnaire is suggested to improve response rates, although Scott (1961) found the effect to be insignificant. Schneider and Johnson (1995) found that university sponsors achieved higher response rates in surveys of business professionals. One of Scott's (1961) conclusions was that it was helpful to convince respondents of the importance of the study. However, Kalafatis and Tsogas (1994) found that offering respondents the opportunity to receive summaries of research results made no difference to the response rates. The use of colour in questionnaires was found not to make a difference (Greer and Lohtia, 1994). A procedure for conducting a mail survey is suggested by Dillman (1978). Overall it is important encourage response by (i) rewarding the respondent through interest and appeal, (ii) reducing the cost in effort and money to the respondent, and (iii) establishing trust. Dillman (1978) and Murphy (1997) outline procedures for conducting the mail survey. Dillman's procedure is
  • Individual, one-page cover letter printed on headed paper,
  • Blue ballpoint signature added to each cover letter,
  • A booklet-type questionnaire with an attractive cover and no questions on the front and back,
  • Questions laid out in a vertical answer format,
  • First class post on all mailings,
  • A post-card follow-up sent after one week with a date and blue ink signature,
  • A second follow-up with a new questionnaire sent three weeks after initial mailing,
  • A third follow-up to non-respondents sent seven weeks after initial mailing.

    Murphy proposes a similar method although follow-ups are at 3, 6 and 9 weeks after initial mailing. Other key items highlighted are the use of a pilot survey to critique the questionnaire. It is important that the questionnaire form itself is as clear and user-friendly as possible by ensuring that there are no ambiguous questions and that there are the correct answer options. Murphy recommends that the length be 3-4 pages.

    Procedure

    As many of the guidelines as practical to improve response rate and clarity of response were followed. The first draft the survey was critiqued by a number of people known to the researcher and working for the same company. They responded with feedback that helped to make the questionnaire more user-friendly. The final questionnaire is included in Appendix A. The questionnaire was sent to all the firms listed in the IDA list of Healthcare firms and to the multinational pharmaceutical firms in the IDA list of pharmaceutical and chemical firms. When firms had multiple sites listed the survey was sent to each site. The total number of questionnaires sent out was 142. The author's own company was not included in the study to maintain uninvolvement which is appropriate for hypothetico-deductive research. The list of firms to which the survey was sent is listed in Appendix A.

    Cover Letter

    The cover letter that was included was individualised and sent to the managing director or equivalent from the IDA lists. Mail merge was used to produce the letters and these were printed on WIT headed notepaper. Each letter was individually signed. The first name of the addressee was used in the greeting. A first name was chosen because it was easier for the mail merge program to process and avoided the problem of addressing people by their incorrect title. This less formal address is more normal for American firms, the majority in the survey (>60%). When the first name was not available in the list the title and name as listed was used instead.

    The letter outlined the subject of study. It explained the reasons the study was being conducted and stressed the importance of the subject to the recipients. A copy of the research results was offered to respondents. This was done both as a carrot to entice responses and also as a means of promulgating the results to interested parties.

    The last page of the questionnaire asked the respondents if they would like to receive a copy of the results and asks them to provide their address. If respondents wish to remain anonymous they may do so as no identification number was included on each form. The respondents were also asked if they would be willing to participate in follow-up interviews.

    Follow-up Letter

    Four weeks after the initial mail-out a follow-up letter was sent to all those who could not be identified as having responded. The letter is included in Appendix A. The letter reminded the recipients of the questionnaire and explained that a response had not been received. They were again asked to complete it or have it completed. They were thanked if they had responded anonymously and asked to ignore the letter. If they had not received or had mislaid the survey they were asked to respond by phone or email to be sent another. They were again reminded that a copy of the results would be sent to all respondents who wished it.

    Only one follow-up letter was sent and not three as suggested by Dillman and Murphy. One gentle reminder to tardy respondents was considered appropriate. It was felt that more follow-up letters or phone calls were not likely to be well received by recipients who chose not to reply. For this reason they were considered unlikely to result in many usable responses.

    Table 4. Summary of methods used to improve the response rate.
    Stamped addressed response envelopes included
    Promise of anonymity
    One follow-up letter
    Use of college headed paper for letters
    Individualised letters individually signed
    Promise of summary of results

    Research Questions

    The objective of this research is to ascertain how multinational pharmaceutical and healthcare companies manage the process of technology transfer into Ireland, and to find out what they have learned in the process that can help them do it better in future. With this information a strategy for technology transfer will be proposed. The answers will be compared to previous findings in the literature. After reviewing the literature the key issues were determined. The effects of the size of the subsidiaries, the duration of their establishment in Ireland and their degree of autonomy, on the smoothness of technology transfer, as measured by adherence to the schedule and budget is assessed. The benefits or otherwise of techniques, such as travel by Irish based employees to the head office or other company sites and travel by other employees to Ireland is assessed. The effect of quality of training and documentation is measured. The effect of the 'newness' of the technology and the application of various business improvement initiatives on the success of technology transfer is assessed. The knowledge gained by those involved in the transfer process that could help them transfer technology more efficiently in future is also assessed. The extent of Research and Development by the Irish subsidiaries is measured and compared with technology transfer capabilities.

    The Questionnaire

    The questionnaire is included in Appendix A. It has four sections numbered A-D. In section A twelve questions are asked about the Irish operation. These questions relate to the length of time established, the size in terms of employees and sales, the percentage exported, and the degree of autonomy that the company has with respect to human resources, R&D, finance, product specifications and customer relations. For these questions a five-point scale similar to the Likert scale was used. In general, most questions in the survey use a 5-point scale. Some questions use numerical scales. Others were influenced by the scale employed by Szulanski (1996) for many of his questions, i.e. 'YES!', 'yes, but', 'no opinion', 'no, not really', 'NO!'. A similar scale was employed by Ghoshal and Bartlett (1988) in their study of adoption of innovations by multinational corporations.

    The information in Section A is required for making comparisons of technology transfer and research and development by multinational firms in Ireland. Comparisons are made with respect to their size and degree of local autonomy. Many of the categories are based on those used by Taggart (1998) in his study of the complexity of R&D conducted by affiliates of multinational firms in the UK. Similar information is also used by Mudambi (1998) in his study on multinational investment strategies and by Ghoshal and Bartlett (1988) in their study of the creation, adoption and diffusion of innovations by subsidiaries of multinational firms.

    The largest part of the questionnaire relates to technology transfer. Section B has thirty-one questions. The seminal study by Teece (1977) inspired several of the questions. Teece found that the cost of technology transfer was related to the experience of conducting transfers and the degree of complexity of the transferred process. He also found that transfer costs decline with increasing application of the technology. Information related to these hypotheses is sought in several questions. This topic was further examined by others including by Zander and Kogut (1995). A question is asked to determine the role of the respondent. A question is asked about the importance of the Japanese practice of limiting the number of new problems to be dealt with for a given technology transfer. In general the Japanese only like one of the site, people and process to be new.

    Travel by key personnel from both the technology transferee to the transferor and vice versa is a key part of this survey of the technology transfer process. This was previously examined in an Irish context by Farrell (1991). Ghoshal and Barlett (1988) examined the importance of links between head office and subsidiaries. The strength of such links was examined by Hansen (1999).

    Questions are asked to measure the effectiveness of technology transfer. These questions are based on those used by Szulanski (1996) in his study of 'internal stickiness', the impediments to transfer of best practice within the firm. All answers are asked for on a 5-point scale. Questions are asked about schedule and finance, satisfaction, commitment to the process, training and skills.

    Two open questions are asked seeking qualitative information about technology transfer. The purpose of these questions is to give the respondents the opportunity to contribute additional ideas about practices and methods which help and hinder technology transfer. Answers here may be in addition to methods asked about, or may give the respondent the opportunity to emphasise aspects about their technology transfer experience.

    Questions inspired by Zander and Kogut (1995) who investigated the effect of codification and imitation on technology transfers within firms are asked. Aspects of this investigated are the quality of documentation and training, the uniqueness of the technology, and employee defections.

    Questions are asked about the application of a variety of business improvement methods to the technology transfer process. Quazi and Bartels (1998) investigated the application of TQM principles to technology transfer. Questions are also asked about business process reengineering, six sigma, benchmarking, integrated technologies and lean thinking. A question is asked about the application of the Internet to technology transfer, a subject studied by Hoetker (1997).

    The third section examines Research and Development. The surveyed are asked if there are R&D facilities in Ireland. The scale devised by Taggart (1998) is used to grade the complexity of research and development conducted in Ireland. Questions are asked concerning the global product development process investigated by Subramaniam et al (1998). Another is asked about the nature of the firm's product development process. The models are inspired by Nonaka and Takeuchi (1995) and describe the Relay, Sashimi, Rugby and American Football approaches to product development.

    In the final section of the questionnaire respondents are asked if they would like to receive a copy of the results, and whether they are interested in participating in more in- depth investigation of technology transfer. The question was asked a priori in case the quality of information gained from the mail survey was insufficient and further data gathering was required.

    Analysis of Questionnaires

    Responses were received for 69 of the questionnaires (of 142) - a 49% response rate. Of these 58 (41%) were completed, with answers to most questions given by 56 (39%). A full summary of the responses is listed in Appendix B. This is a good response for a mail survey where an initial response rate around 20% can be expected (Murphy, 1997). The response rate compares favourably with 34.2% for Taggart (1998) in his investigation of the complexity of R&D at multinational companies in the UK. It is a lower response rate than the 80% achieved by Zander and Kogut (1995) in a much smaller survey of Swedish multinationals (35 returned of 44). However, they phoned in advance to determine the most suitable person for answering the questionnaire.

    There were several reasons why questionnaires were returned incomplete. Some addresses changed and several were returned to sender without reaching the addressee. In other cases the recipients refused to complete because of a corporate policy on surveys. In others there had been no recent transfers with which people employed at present were familiar. In another case the company was involved in a different industry (e.g. cosmetics). It may be more appropriate to place a number of the companies listed on the IDA list of healthcare companies in a different industry category. Several forms were returned by indigenous Irish firms. The two firms that did not complete the majority of questions were Irish firms. Unlike the pharmaceutical company list, the medical device list did not distinguish between multinationals and Irish firms.

    More responses (49) were made before reminder letters were received than after (20). Nine anonymous responses were received, all before the reminder letters were sent. At the end of the questionnaire respondents were asked if they wished to receive copies of results (46 wish to receive results), and if they were willing to participate in follow-up interviews (20 are willing). Sufficient information was obtained from this study. Therefore this follow-up study was not conducted.

    Preparation of Data

    The data from all returned forms was input into an Excel spreadsheet. Some adjustments of the data were required. A five-point scale was applied to many of the questions with a numerical response. For instance the questions about duration of company operation in Ireland (1&3) were scored (1) <=5 year; (2) 6-10 years; (3) 11-20 years; (4) 21-30 years; (5) > 30 years. Similar scales were applied to the answers to the questions about number employed, the number of transfers and the age of R&D operations in Ireland.

    A detailed summary of the answers is included in Appendix B. The answers to most questions merely needed to be input into the data sheet. However there were some inconsistencies noted while inputting which have had some influence on interpretations of the answers. The question asking the number of transfers conducted was rated 1-5 (1) 0-1 transfer; (2) 2-5 transfers; (3) 6-10 transfers; (4) 11-20 transfers; (5) >20 transfers. The mean number of transfers was 11.7. In some cases answers such as >20 were given. These cases they were interpreted as the number itself, e.g. 20.

    The answers given to the question about the number of 'news' were in some cases not consistent with the question about the number of times the technology had previously been transferred. For example some respondents answered that it was not a new process but said it was the first implementation of the technology, and vice versa. There may have been differing interpretations of the words process and technology. No attempt was made to modify the answers. The 'news' question was scored as 0, 1, 2, or 3. If there was neither a new site, new people or a new process 0 was scored and 3 if all were new. The question about the number of previous transfers of the technology was only answered as either 1, 2, or 3. Five choices were not necessary.

    The majority of answers to the question about the number of transfers which had occurred prior to the subject technology transfer was four or more transfers. This shows that the incorrect range was used for the five choices and this is consistent with the answers given to the question of the total number of technology transfers. The questions about the numbers of people travelling to assist with the technology transfer could have been worded more clearly and were answered in different ways by different respondents. Twenty-two respondents entered a number in each of the boxes showing the number of employees who made 1, 2, 3, 4, or 5 or more trips. However, the majority of respondents just ticked one box for each question. This was interpreted as being the number of employees who travelled and the other 22 answers were interpreted to simulate such a response.

    The answers to the questions pertaining to compliance with project schedules and budgets were used as the major metric of technology transfer success (Did the project start on time?; Did the project end on time?; Was the project delivered on budget?). The answers to about budgets were reversed to facilitate comparison. An answer of 1 for much more than budget was changed to 5. A mean of the 3 answers was used to examine all other response to see whether there were significant trends. The value is called TTE for Technology Transfer Effectiveness and is the major metric used for analysis in the next Chapter.

    There were 48 answers to the open-ended question asking to list specific practices and methods which aided technology transfer. There were 28 answers to the open-ended question asking to list specific practices and methods, or their absence, which hindered technology transfer.

    The questions asking about business improvement initiatives have five parts, one for each initiative. If the answer was blank, No or Don't Know, the response was scored as no. The rational for this was that if the application of the process improvement method was not known to the respondent, it had either not been applied or its application had been ineffective. The question about Internet use was scored as 1 for extensive use, 2 for moderate use and 3 for no use.

    The question about duration of R&D facilities was scored (0) for no R&D facilities and (1) for one year; (2) 2-5 years; (3) 6-10 years; (4) 11-20 years; (5) >20 years. The question about complexity of R&D was scored as 0 for none up to 5 for competence to develop new forms of technology. It should be noted that the number of firms with no R&D facilities (31) exceeds the number of firms doing no R&D (18). This indicates that many firms carry out R&D in Ireland without dedicated R&D facilities. The questions about product and process development have yes/no responses. The question about Nonaka and Takeuchi's (1995) product development methods is scored 1-4 for each of the four choices.

    Data Analysis

    The numerical responses were analysed using Excel. The responses from each questionnaire were entered on a row of a spreadsheet. The answers to each question were all in the columns. The number of answers in each column were counted and averaged. The number of each response was counted. Using Excel array functions the effectiveness of technology transfer for each response to each question was compared with the answers given by the respondents to the two questions about adherence to schedule and the question about budgets and their mean. This mean value is called 'TTE' and is the major effect variable of the study. Use of the array functions resulted in data with a better 'feel' than the use of correlation coefficients.