CHAPTER THREE: METHODOLOGY
3.5 Research Methods – Purpose, Design and Implementation
As explained in Section 3.3, the deliberate choice of an abductive approach in conducting this research lends itself to a combination of various research methods in order to achieve the research objective. This section explains the selection of research methods which have been applied throughout the project, and details the purpose, design and implementation of each one of them.
3.5.1 Quantitative and Qualitative Research Methods
The selection and design of research method is greatly influenced by the research approach adopted. Although, as aforementioned, the traditional links of qualitative-inductive and quantitative-deductive are less evident in logistics research than in other disciplines, the general patterns still remain. Table 3.3 gives an
Deductive methodology in contemporary logistics research. Hence the pattern shown in Table 3.3 is an illustration of common practice rather than a rigid classification.
Table 3.3: Research methods normally applied for three research approaches
Source: compiled by author
As highlighted in the table, empirical testing on the preconstructed sustainable framework was conducted by means of quantitative surveys and semi-structured interviews. Case study was chosen as a fit-for-purpose method at the final demonstration stage. Therefore, both quantitative and qualitative data were collected for analysis in reaching the conclusion of the research.
The benefits of such a flexible combination of both data collection methods are substantial – the advantages of each method are retained; while the different types of data complement each other, minimising the disadvantages of using either one of them in isolation and providing a comprehensive, yet also in-depth, perspective on the research questions under investigation. Table 3.4 lists the strengths and weaknesses of both quantitative and qualitative research methods in a matrix,
Quantitative Methods Qualitative Methods
Strengths Weaknesses
Efficient and low cost - large amount of information in limited time scale Time-consuming; costly in terms of resources
Credibility - less bias Bias (mainly from researchers)
Anonymity for respondents Anonymity concerns (possible bias from respondents)
Greater accessibility Access to subjects and information
Generalisable results from representative samples Reflexivity; Selectivity Objectivity and Transferability Subjective and less transferable
Weaknesses Strengths
Complexity of method; Artificial theoretical framework Tailored method with flexible framework Oversimplification of the real world problem Open to incorporating new issues and critical factors Risk of misinterpretation (by either respondents or researchers) Reality; contexual, hence better comprehension Past-, not future- oriented (Snapshots) Insightful exploration
Limited depth of information Targeted, detailed and in-depth information
Less control over / communication with the respondents More control over / communication with the respondents validity subject to researcher's understanding of the question Self-correction and validation along the open process
showing the complementary characteristics of each method.
Table 3.4: Strengths and weaknesses of quantitative and qualitative research methods
Source: compiled by author
Traditionally, rigour as a critical quality criterion favours the quantitative method, leading to its dominance in logistics research over a long period. However, the arguably inherent lack of flexibility and sacrifice of relevance in the rigorous process of a quantitative method can be well complemented by the qualitative method, which is showing an increasing emergence in logistics.
In general, the quantitative method works well in measurement and analysis of a causal relationship between variables (Denzin and Lincoln, 2000), whereas qualitative research is normally applied in process examination, and phenomenon exploration and explanation. In simpler terms, quantitative research is good for testing hypotheses, and qualitative research for discovery (Bryman, 1988). The nine hypotheses developed in Section 3.4 represent a mixture of both types of question.
The quantitative method is required in deciding whether a correlation or interconnection exists between certain factors (variables), whilst the qualitative method is necessary for characterising and interpreting the existing relationships.
Therefore, it is believed that a hybrid approach is appropriate for the research.
3.5.2 Questionnaire Survey
A two-stage questionnaire survey was adopted in the early stages of the field work to collect statistical evidence in order to address the majority of the hypotheses, complemented by the in-depth interviews which were subsequently carried out. The survey aimed at a large sample of LSPs and LSUs operating in the UK. Information on sustainable logistics operations were collected from individual businesses across vertical sectors to give an overview of the main actors‟ attitudes and state of practice.
Questionnaire Design
The survey was designed to have two major phases with different data requirements.
Accordingly, two questionnaires were developed (see Appendix Two) and distributed respectively at each phase.
Phase One: A single page, double-sided questionnaire (for the mail survey) containing nine questions was distributed to 600 LSPs and 700 LSUs. It was designed to be completed by all respondents, with the main purpose being to collect general information about the company, particularly in terms of its logistics operations, and its general attitude towards the 14 listed sustainable solutions.
In the questionnaire, data were gathered regarding basic attributes of the business in standardised formats, such as operational scale (measured by fleet size in the road haulage sector), geographic scope, freight transport mode in use, and vertical sectors which the business mainly serves or to which it belongs. The respondents were then requested to select and rank five out of the 14 sustainable solutions which have the greatest impacts on their logistics operations. These data were subsequently coded into SPSS (Statistical Package for the Social Sciences) to enable statistical testing for
any potential relationships between these attributes and the companies‟
attitude/behaviour pattern in sustainability.
The questionnaire designed for this preliminary phase was kept as simple and straightforward as possible in order to obtain a desired response rate of at least 10%, which was deemed as achievable after examination of previous logistics research in the British industry. It mainly consisted of multiple-choice questions and should normally be completed within five minutes. It included the option for respondents to express their willingness to attend an interview in the near future, so that the accessibility problem of approaching suitable interviewees was largely tackled. There was also a question helping to screen out the potential candidates for a valuable case-study demonstration, which addressed the same issue of identification and availability. For further contact, demographic information such as name, job title, telephone number and email address were requested from respondents as well.
Phase Two: The questionnaire for this phase was designed to reveal the detailed attitude/behaviour patterns of the business regarding the five solutions they previously selected as the most influential ones in the phase one survey. The complete questionnaire is split into 14 sections, as can be seen in Appendix Two, each exploring one of the 14 sustainable solutions respectively. However, to avoid deterring potential respondents by the whole 16-page questionnaire, with many pages not requiring an answer, extra efforts were made to only include the relevant sections of the respondent‟s previous choices in the phase one survey. By doing so, the length of the questionnaire was much reduced, and the response rate was greatly boosted.
Although the questions for each section are not identical due to distinctive natures of each sustainable solution, all the sections were designed to cover the following major issues where applicable:
- the current sustainable practice in the logistics operation, including the specific measures adopted, to what extent and in which forms they are adopted;
- the main drivers or incentives for the business to employ the sustainable solutions;
- the main concerns about or disincentives to adopting the sustainable solutions;
- the strategies of the business in response to those solutions whose implementation are beyond its control, but have impacts on its operation and require its adaptation;
- respondent companies‟ evaluation of the overall cost-effectiveness and efficiency of the sustainable solutions;
- the interactions between key stakeholders and between certain solutions.
For the ease of completion, most questions were again multiple-choice questions, in combination with a small number of ranking and rating questions. The questionnaire also presented a mixture of closed- and open-ended questions, with the former making up the great majority. To keep the subjectivity from the author to a minimal level, a comment field was added to most closed-ended questions for respondents to give any other opinions if they were not covered by the options given. This also assures that the new issues which arose during the research were built in to the following analysis, rather than being eliminated simply because they lay outside the pre-determined research structure.
Survey Implementation
Prior to the full-scale distribution of the questionnaires to the target companies, a two-stage piloting was completed, providing indicative feedback that helped to guide and adjust future research activities.
In the first stage, the draft questionnaire was presented to seven experienced researchers and two consultants, asking for their opinions on the design of the
questionnaire. Four researchers and both consultants made their comments on a range of issues such as the phrasing of certain questions, realigning the structure of the questionnaire, and the analysis method of the results to be collected. All issues raised were taken into account, and through continuous refinement, the questionnaire was finalised and an equivalent online survey was established on SurveyMonkey (a web-based tool that enables the creation of online surveys) to offer a fast and environmentally friendly way of completing it.
Based on the refined questionnaire design, a second-round pilot survey was then conducted, with 100 questionnaires distributed to LSPs, and nine completed and returned within a month of receipt. This result signified that a 10% response rate could be expected for the following full-scale survey. It was planned that in the event of a lower response rate, either a second-round phone call/email/mail reminder could be sent out to the targets, or the sample size could be expanded to ensure a minimum 120 valid responses for analysis.
Since there were no apparent problems with the way in which the returned pilot questionnaires had been completed, the full-scale survey was carried out after minor amendments. The sample frame is composed of two major interest groups in the industry: LSPs and LSUs. Samples from both groups were identified through FAME, an online database of company information in the UK and Ireland, and Applications
& Decisions (A&Ds) published by the Vehicle and Operator Services Agency (VOSA). The number of companies to be contacted was initially set at 1200 in total, with 600 for each group. However, due to the lower response rate received from the LSUs group, an additional 100 questionnaires were circulated applying the same sampling techniques as with the first 600 LSUs.
For the phase one survey, stratified and cluster sampling techniques were adopted for
each group respectively:
Stratification by logistics operation scale was applied for LSPs group. The population of this group consists of all the logistics companies operating in the UK, with the sample frame being the companies/operations listed in the FAME database and the recently published A&Ds (those from the last quarter of 2008 and the first quarter of 2009), respectively representing large LSPs (FAME, with 93% companies exceeding £1m annual turnover) and small operators (VOSA, with the vast majority operating less than ten vehicles). From each source, 300 samples were randomly drawn. This sampling frame and stratification is not only a real reflection of the highly segmented logistics industry structure, but also enables further comparative analysis by operation scale at a later stage.
For LSUs, cluster sampling by vertical sector was applied, identifying food, drink & tobacco; other FMCG (manufacture); industrial; retail; chemical and automotive as six focus sectors actively involved in logistics activities. This identification was based on the compilation of the industry sectors that the top 100 logistics companies in 2006 are mainly serving, which broadly indicates the sectors with the largest shares in the logistics market (see Appendix Four). The population for this group consists of all the large businesses trading in the UK from the six focus sectors; the companies listed in the FAME database within the six target sectors constitute the sample frame for the survey, from which a sample of 100 was drawn from each sector. As mentioned, most companies listed in FAME are large (LA) or very large (VL) companies. Through sorting the companies within each of the six sectors by annual turnover, the first 100 were targeted as leading companies of the sector, and in combination the 600 largest companies representing the majority of market share in the focus sectors constitute the initial set of target samples for LSUs.
As a result of this sampling approach, the survey findings will be applicable to the
large service users in the focus sectors who collectively have a dominant market share and greater influence in logistics practice.
The stratified or cluster sampling itself, no matter how rigorous the process it follows, does not guarantee that the desired responses will be received. However, the design of the questionnaire has incorporated elements to ensure the self-classification of the respondents (e.g. information on their operation scale and vertical sectors). The survey was also devised in such a way that it would identify any issues with sampling methods. For instance, in the responses to the phase one survey, if any other sectors were discovered with substantial influence on the logistics industry, the cluster sampling by sector could be rectified to include further investigation into these emerging sectors.
The distribution of the first questionnaires was primarily carried out via post to the registered addresses of the sample businesses. The mail was always addressed to the name of the senior logistics/distribution managers wherever possible, in order to ensure a good response rate and convenience of tracking. Freepost reply envelopes were enclosed with the questionnaire to reduce the time and effort required on the part of the respondents to complete and return their questionnaires. As an alternative, the recipients were invited in the mail to follow the link and complete the questionnaire online if they preferred.
During the second phase of the survey, the established online survey webpage played a greater part with the respondents‟ email addresses obtained from the first phase.
Most recipients were sent an email with an embedded link which would lead them to the survey page with only the sections of relevance to them, based on their choices as given in the previous questionnaire.
Survey Results Overview and Analysis Method
After the pilot survey with 100 LSPs, the remaining 1100 questionnaires (500 to LSPs and 600 to LSUs) for the first phase of the survey were distributed between February and April 2009, with a request for recipients to complete and return them within one month of receipt. The return of completed questionnaires was reasonably quick, with most coming back within the first two weeks. However, there were many returned by Royal Mail due to the addressees no longer being in existence. This is mainly due to the outdated or incomplete information provided in the data sources used, especially from FAME.
By the end of May 2009, 57 completed questionnaires had been returned from LSPs group, including the nine responses from the pilot survey, giving a fairly satisfactory 9.5% response rate within the group. In contrast, however, the responses received from LSUs were much lower, with only 36 questionnaires completed and returned (representing a 6% response rate). In an attempt to boost the response rate, reminder letters with questionnaires or email invitations including the online survey link were sent to some of the companies which had not responded yet. Given the particularly low response rate from LSUs, an extra 100 LSU companies from the focus sectors were incorporated to the original sample in order to obtain the desired number of responses for analysis. These efforts achieved considerable success and generated another 11 responses from LSPs and 23 from LSUs, bringing the final response rates of the phase one survey to 11.3% and 8.3% for LSPs and LSUs group respectively.
The second-phase questionnaire was only distributed to the companies which responded to the first one, which came to 68 LSPs and 62 LSUs (including three logistics consultants). Out of the 130 questionnaires distributed during the second phase, 51 were completed by LSPs and 43 returned from LSUs. Table 3.5
summarises the responses received by logistics group at both survey stages.
Table 3.5: Response rates of the survey by logistics group (two phases)
(* as a percentage of stage 1 responses)
Source: author’s questionnaire survey, phase 1 & 2
The completed questionnaires were analysed by using SPSS. As the vast majority of questions were closed-ended ones with predetermined sets of options, it was possible for the large quantity of categorical data (both nominal and ordinal) collected in standard format to be easily coded and input into SPSS for further processing.
Non-parametric test was the main method used for probing some of the hypotheses, while qualitative data acquired from interviews and case researches were also incorporated into the analysis. The comprehensive analysis of the questionnaire survey will be reported in detail in Chapter Four.
3.5.3 In-Depth Interviews
It was shown in Section 3.5.1 that a combined method encompassing both quantitative and qualitative analysis will facilitate the exploring of the research questions, providing a comprehensive coverage and an in-depth insight at the same time. Semi-structured interviews, as the method for collecting qualitative, insightful data in support of the survey findings, were carried out with logistics practitioners.
In-Depth Interview Purpose and Design
Phase 1 Reponses Response Rate Phase 2 Responses Response Rate*
LSP 600 68 11.3% 51 75.0%
LSU 700 59 8.4% 40 67.8%
Total 1300 127 9.8% 91 71.7%
the interviewees had completed both parts of the questionnaire survey before the subsequent interview was arranged, depending on their availability and suitability.
This arrangement benefited both interviewer and interviewee as regards gaining appreciation of the interview‟s agenda. For interviewer, particular behaviour and attitude patterns emerging during the survey from a specific respondent can be worked into the interview structure for further exploration. From the interviewees‟
perspective, participating in the questionnaire survey in advance enabled them to better comprehend the topics to be discussed in the interview and reduce the chance of confusion or misunderstanding during the course.
Bearing in mind the research objective and the consequent requirements on data collection, the sample of interviewees should ideally not only sufficiently represent both LSPs and LSUs groups, but also comprises an adequate number of companies from the major vertical sectors with a large market share and thus a major influence on logistics market. Also, to incorporate the non-standard, specialised logistics operations into the scope of the research, so that the analysis of and conclusions drawn from the evidence are effectively applicable to the whole industry, the interview sample should also include representatives for these segments which had to be left out of the questionnaire survey due to their distinct operational characteristics;
examples include air freight, waterborne transport, mail and express operators.
However, the desired comprehensiveness of the research has to be compromised given the limited resources and the accessibility of the targeted companies. Given that 52% of questionnaire respondents agreed to a further in-depth interview on the subject, as shown in Table 3.6, approximately ten interviews on average could be achieved for each sector, with roughly half of them being with LSPs and another half with LSUs. As a result, a realistic target in the range of 70 to 80 interviews was
However, the desired comprehensiveness of the research has to be compromised given the limited resources and the accessibility of the targeted companies. Given that 52% of questionnaire respondents agreed to a further in-depth interview on the subject, as shown in Table 3.6, approximately ten interviews on average could be achieved for each sector, with roughly half of them being with LSPs and another half with LSUs. As a result, a realistic target in the range of 70 to 80 interviews was