6. FACTORS INFLUENCING PORT PERFORMANCE: PORT
6.1 Data screening, cleaning and sample characteristics
6.1.2 Sample characteristics
As presented in Chapter 4, the questionnaires were distributed to specialists in the Humber estuary (UK) and Xiamen (China) respectively. In the Humber, 200 questionnaires were distributed and 92 out of the 96 responses were found to be valid (valid response rate=46%), while in Xiamen 300 questionnaires were distributed and 162 out of the 167 responses were identified to be valid (valid response rate=54%). The total valid response rate was 50.8%. Table 6.1 presents the response rate by region. The questionnaires were distributed to five types of companies, including consignors/consignees, PSPs, shipping lines, port managers and other port stakeholders.
Table 6. 1 Questionnaire distribution and response
Region Number distributed Responses received Valid responses received Valid response rate
Humber 200 96 92 46%
Xiamen 300 167 162 54%
Total 500 263 254 50.80%
Table 4.7 presents an overview of response rate by company type and region. In total, 51 responses (20.08%) were from consignors/consignees, 62 responses (24.41%) were from PSPs, 32 responses (12.6%) were from shipping lines, 49 responses (19.29%) were from port managers and port operators, and 10 responses (3.94%) were from other port stakeholders. Another 50 responses (19.69%) did not disclose their company name so they could not be simply included into any company type but were treated as missing data. The composition of the sample by type of organisation shows that company selection was devoid of demographic bias and the response rate in the two regions was very good, with approximately equal response.
Figure 6.2 presents the response profile by company type frequency and percentage with combined data of the Humber and Xiamen. The bar chart gives the number of respondents while the pie chart shows the percentage of respondents in each group.
Figures 6.3 and 6.4 present details of the response frequency and percentage by the Humber and Xiamen separately.
Figure 6. 3 Respondents profile by company type and region (frequency)
Figure 6. 4 Humber/Xiamen respondents profile by company type and region (%)
Table 6.2 presents the respondents‘ job positions. Most of them held senior positions in their organisations. 49 (19.3%) of them were directors, 125 (49.2%) of them were managers, including branch manager, general manager, shipping manager, terminal manager, transport manager, operations manager and the remaining 26 (10.2%) held other titles, such as principal consultant, master, and other port experts. 54 (21.3%) respondents did not disclose their positions. Among the 200 respondents who disclosed their job positions, 87% of them were directors and managers. This reflects that the respondents were in the right position to complete the questionnaire and provide useful, valid and insight responses.
Table 6. 2 Respondents by job roles (combined samples) Job title Frequency Percent Cumulative Percent
Directors 49 19.3 19.3
Managers 125 49.2 68.5
Others 26 10.2 78.7
Missing 54 21.3 100.0
Total 254 100.0
Figure 6.5 presents the frequency and percentage details of respondents‘ job position.
Figure 6. 5 Frequency/% of respondents with different positions (combined samples)
In terms of questionnaire response method, Table 6.3 and Figure 6.6 show that the majority of participants (172 and 67.7% of the total 254 respondents) responded with self-completion, mainly via email, supplemented by post or fax. The remaining respondents (82 with 32.3% of the total 254 respondents) responded by face-to-face survey.
Table 6. 3 Respondents by response method
Response method Frequency Valid %
face to face survey 82 32.3
self-completion by email, post and fax 172 67.7
Total 254 100
The self-completion respondents preferred to complete the questionnaire without disturbance in their own time, as they felt it more comfortable to do so. The face-to-face survey respondents thought it easier and more efficient to complete the questionnaire in the researcher‘s presence, so that they could clarify the questions with the researcher if they had some concerns or if they were not clear about the questions. Some respondents explained their willingness to offer comprehensive data for this research if needed.
Figure 6. 6 Response method (combined samples)
Non-response bias
Non-response bias may arise when the characteristics of the respondents vary significantly from those of the non-respondents. It can be a problem when response rate is lower than 40% (Lambert and Harrington 1990). The bias may occur even when the response rate is high (Carter and Jennings 2004). This is why it is necessary to test the non-response bias, even though the response rate of the current study was over 40%.
Armstrong and Overton (1977) consider that later respondents have similar views to the respondents, as they respond due to additional stimulus. They assume that non-response bias does not exist if no significant differences exist on the survey factors between the early responses and late responses. Thus, this research tested the difference to examine the potential non-response bias problem by following the recommendation of Armstrong and Overton (1977) and Rada (2005).
The non-response bias was checked by the Mann-Whitney U test and the Kolmogorov-Smirnov (K-S) Z test of the SPSS software, as they are most popularly used to test whether two independent samples come from the same underlying population (Pallant 2007). In this study, the first and last 40 respondents were compared to assess the potential non-response bias for both the Humber and Xiamen separately. The results revealed no significant differences between the early and late responses, as all p-values were greater than 0.05, meaning that the means of the groups were not significantly different. Therefore, the tests confirmed that the results emerging from the data would be valid and devoid of chance.
This research also compared the number of respondents with the number of sampling frame by different respondent groups from different types of companies. The response rates were not equal. The general response rate of the Humber (46%) was lower than
that of Xiamen (54%), and the response rates for different groups were not the same
C C Section C, performance of other ports
1 1-shipping services shipping services
2 2-shipping prices shipping prices
3 3-portcharge port charges
4 4-feeders Feeders
5 5-overall cost overall cheapest cost of logistics services
6 6-handlingspeed speed of cargo handling
7 7-risks port risks
8 8-safety port safety
9 9-techinfras port technical infrastructure, e.g. equipment and ICT
10 10-proximity port location to the customer and supplier
11 11-skills logistics skills for those working in port performance
12 12-landlinks landside links, including air, rail and road
13 13-logservices logistics services, e.g. Warehousing.
14 14-govs.upport government support
15 15-navigation depth of navigation
16 16-portservc port services (A-factor)
17 17-logsupt logistics support (A-factor)
18 18-cost logistics cost (A-factor)
19 19-shipservc shipping services (A-factor)
20 20-others Other factors (A-factor)
A1-A15 A1-shipservices to A15-navig. importance of factor 1to factor 15 B1-B15 B1-shipservices to B15-navig. performance of factor 1 to factor 15
C1-C15 C1-shipservices to C15-navig. other ports‘ performance of factor 1 to factor 15
∆(C-B) Perf. Diff. (C – B) performance difference between other ports and selected port
XM Xiamen Humber port stakeholders (40%). This is because the sample of other port stakeholders consisted of just five important government agencies, academics and consultants. This applies to a sample of 60 PSPs, of which 29 responded. Similarly, response rates of Xiamen are generally lower than those of the Humber. 16.7% of the Xiamen respondents did not disclose their identity.
Prior to presenting the analysis, to help with the understanding of the questionnaire and simplify the analysis presentation, Table 6.4 is given to refer to the terminology and abbreviations for the data analysis.