6. Quantitative Data Analysis and Triangulation with Qualitative Outputs
6.5 Classification And Regression Tree (CART) Analysis
Classification And Regression Tree (CART) analysis was carried out as a data reduction technique. The analysis as shown below condenses the most significant relationships into a small number of groups.
The CART analysis produces a tree as shown in Figure 6.11 in which the parental node (students’ age) is split into two child nodes based on ‘Relevancy of curriculum to the
work role’.
Node 0 – Consists of students’ age
Node 1 – Agreeing for the ‘Relevancy of curriculum to the work role’ Node 2 - Disagreeing for the ‘Relevancy of curriculum to the work role’
As shown in Figure 6.12 below, 43% of postgraduate students have taken up WBL at the age range of 21-30 years. This is an unexpected high percentage which may be an indication of the current difficult economic climate where FT commitment for education is not affordable for many just after school at young age. 52% are in the age range of 31-50 years which is anticipated for WBL but interestingly there are 2 students in the age group of 51-60 years and one in the age range of 61-70 years as well. The rapid adoption of technology in delivering WBL programmes is another factor which motivates young employees to take up WBL.
Dependent Variable: What is your age?
Figure 6.11 Association between age, and relevancy of curriculum to work role
Figure 6.12 Age distribution of postgraduate student
Surprisingly, only 8% of u/g work-based learners were aged between 21-30 (see Fig. 6.13 below) and 92% were aged between 31-50 as opposed to 43% and 52% of postgraduate WBL students respectively. This explains the fact that among the workforce, there are many employees without higher academic qualifications who are still willing to obtain a degree qualification even during the later stages of their career. Also, people may
reach a work situation where they cannot progress any further without additional academic qualifications which will inevitably come later in their career, hence the older age groups.
Figure 6.13 Age distribution of u/g students
The qualitative data shows that u/g degrees are not tailored to the students’ workplace activities as much as they are in the p/g degrees. This is supported by the CART analysis (Figure 6.11). U/gs who are mainly between the age gap of 31-50 either disagree (D) or strongly disagree (SD) with the fact that their curriculum is very relevant to the work role and responsibilities whereas p/gs who are mainly in the age gap of 21-30 either agree (A) or strongly agree (SA) for the same fact. Therefore, quantitative and qualitative triangulation is proven here.
Dependent Variable: What is your gender?
The CART analysis produces a tree as shown in Figure 6.14 in which the parental node (students’ gender) is split into two child nodes based on ‘number of hours students
spend on their main employment’.
Node 0 – consists of students’ Gender Node 1 – Less than or equal to 31-40 hours Node 2 – Greater than or equal to 31-40 hours
75.9% of females work < 31-40 hours due to their maternal, family and other commitments whereas 78.6% males work > 31-40 hours. This could have an impact on the number of hours both males and females can devote to their studies. However, there are other external dependencies for individual students. Irrespective of students’ gender, these dependencies include their civil status, academic capabilities/skills, width and depth of involvement on family commitments, the time spent on travel, how much support they get from others like employer, university, peers, and family.
At the interviews, it was noted that the number of working hours depend on the type of profession and in general high calibre jobs (executive level) require employees to work extra hours due to the level of responsibility whilst other jobs entertain over-time payments which makes employees encouraged to work extra hours. Also, there are gender specific learning characteristics exist which can have an impact on their leaning hours. The differences between men's and women's learning styles were researched by Perry in 1968 (Lieb, 1991) in his study of u/g New England college students (male). From this study he determined that young men pass through a developmental sequence in their thinking modes.
Perry's "Developmental Process"
1. Male students see the world as black/white, right/wrong--they are convinced there
IS one right answer
2. Male students see there is diversity of opinion, but feel that authorities that describe
diversity are poorly qualified, or just "exercising students" so students will be forced to find the "right answer" themselves
3. Male students begin to feel that diversity is temporary. They feel that maybe the
"right" answer just hasn't been found yet
4. Male students understand that diversity is a legitimate state, but they would still
prefer to know what is "right"
5. Male students see that everyone has a right to his or her own opinion
Nearly 20 years later, Belenky et al. (1986) wondered how women fit into this "male" scale (if at all). In their study they discovered that women indeed do have different
"ways of knowing." Unlike Perry’s developmental stages, Belenky et al. chose not to
describe the way women think in a staged sequence, although women do move from one style of thinking to others as they mature and gain life experience. In outline, Belenky et al. found that women have the following possible "ways of knowing."
Belenky et al.’s "Women's Ways of Knowing"
1. Silence: women students feel mindless and voiceless, subject to whims of external
authority
2. Received knowledge: women students feel they can receive knowledge, but not
create it
3. Subjective knowledge: truth and knowledge are private and subjectively known or
intuited
4. Procedural knowledge: women students are invested in learning and applying
objective procedures for obtaining and communicating knowledge
5. Constructed knowledge: women students view knowledge as contextual and can
create knowledge found objectively or subjectively
All in all, learners’ individual motivation, dedication and commitment decide their performance leaving gender and number of hours of work per week as secondary factors.
The CART analysis produces a multi level tree as shown in Figure 6.15 in which the parental node (number of hours of work per week) is split into two child nodes based
on ‘students’ main employment’ which again splits into two nodes based on ‘employers’ support’.
Node 0 – consists of number of hours students spend on their main employment Node 1 – consists of professions in engineering and IT fields
Node 2 – consists of professions in all fields being considered in the case study
Node 3 – consists of employer support in all possible ways except for fully sponsoring Node 4 - consists of employer support in fully sponsoring
Dependant Variable: On average, how many hours a week do you spend on your main employment?
Figure 6.15 Association between hours of work per week, main employment and employer's support
Out of all employees, only 19.5% work > 40 hours per week. The first classification shows that some professions have more working hours, i.e. mechanical and electronic/electrical engineers, managers and IT specialists (75%) work > 40 hours per week (16.7% of whole cohort) whereas design engineers, IT managers, database administrators and all library and records related professionals (91.7%) work < 40 hours a
week (83.3% of whole cohort). Although this relationship cannot be explained in a straight forward way due to the small numbers of data in the engineering and IT programmes, the fact still remains that high profile jobs and some areas of work demand more work than others.
The quasi-public environment that library staffs often work in is normally highly structured and regulated and therefore there are often safeguards against working long hours plus the patterns of library openings often lend themselves to part-time working. This differs from engineering in private environments, with fewer controls over working hours. Also, many ILM students are currently in non-professional or semi-professional roles and are not chartered which is why they are doing the course. This means their usual working hours are likely to increase once they reach an appropriate level. Further classification between the majority of those who work < 40 hours per week (83.3%) and their employer support illustrates an interesting relationship. Around 20% of all employees who are fully sponsored by employers work only 31-40 hours per week whilst around 50% of students with lesser support from the employers also need to work above 31 hours per week.
It was also noted during interviews that 100% of engineering students are fully sponsored by employers through the Graduate Development Program (GDP) and they also have to work extra hours due to the project nature of their jobs. The important fact here is all the engineers are from private sector companies whereas most of the library related public sector employees are being supported by their employers in various ways if not with sponsorships due to government budgetary constraints.
As noted in the single node tree below on sponsorships for students (Figure 6.16), only 26.3% of employees have been fully/partly sponsored by employers. This needs to be increased if the WBL is to prosper. Either the government should intervene into this to assist potential WBL employees or the universities should get together with employers to find ways and means to support employees.
Dependent Variable: Who is sponsoring your studies?
The interesting fact that emerged at interviews was it is very difficult to convince donors to give student grants for WBL programmes compared to conventional qualifications. This is shown in Figure 6.16 (only 3.5%). Also, a high number of the self financed learners (70.2% self and family sponsored) make up the same high % of dropouts indicating dropouts often occur due to financial difficulties.
The CART analysis produces a multi level tree as shown in Figure 6.17 below in
which the students’ residence parental node is split into two child nodes based on ‘students’ main employment’.
Node 0 – consists of students’ residence
Node 1 – consists of professions in engineering, library and IT fields
Node 2 – consists of professions in all fields being considered in the case study
The above classification (Figure 6.17 below) shows that 72.2% of WBL students are based in the UK. Another factor shown in the classification is in Library, Information Science and Records Management related professions where a significant portion of students are based outside the UK (34.7% of 68.1% in those categories of students) whereas all the engineering students are based in the North East of the UK. These statistics support the evidence provided by the interview data that engineering WBL still needs some on campus face-to-face meetings with academics. Engineering students use their close proximity to the university to visit the programme leader and module tutors to discuss learning contracts and module submissions. The hands-on nature of the engineering discipline tends to require this.
At the same time, 48.6% of students from rest of the UK (outside North East) and 27.8% of students from rest of the world justify that WBL can be still delivered effectively using online learning technology. It was apparent from both student interviews and questionnaires that the majority of students were satisfied about what they receive online from the university. This is surprising given the interview data also suggests there are areas for improvement in the provision of these online programmes. However first time learners may not have had anything to compare with in terms of quality of online WBL and learners’ time constraints and motivation to achieve only the mere qualification may also drive their perceptions in this regard.
Dependant Variable: Where is your residence?
Figure 6.17 Association between residence and main employment
The dominance of accessing online WBL programmes from home (91.7%) (Figure
6.18 below) clarifies the students’ preferred learning styles, venue, time and pace of study.
Other contributing factors may be that there is no time to study at the workplace due to their work responsibilities, non-supportive employers who may not allow learners to study during working hours and even company IT policies which restrict or stop access to the online university systems from within the organisation.
6.6
Conclusion
The results obtained from the statistical analyses of data using different quantitative analysis techniques have been used to compare, support and justify the qualitative outputs that emerged from the interview data. Considering the unequal stratified distribution of data from the selected programmes in the representative samples, it is impractical to make general conclusions from this data. The ‘internal’ triangulation between qualitative and quantitative outputs as well as triangulation among the different quantitative techniques has produced various insights into the delivery of WBL within the FEE of Northumbria University. In addition, the collected quantitative data proved to be comprehensive as parametric and non-parametric, data reduction and exploratory quantitative analytical techniques produced triangulation among themselves which were cross validated in the interviews with the stakeholders included in the study. The use of classification and regression tree analysis indicates that these analytical techniques are well suited for the data set with heterogeneous variables and are effective in combination with the other data analysis techniques applied in this study.