Chapter 2. Performance and Approaches
2.5 Processing Levels Approach
In this section, the extent to which students engage or process information when
solving tasks is discussed and represents the first approach. Processing levels in this
thesis are defined as either being surface or deep as identified by Marton and Säljö
2.5.1 Processing Levels: A Brief Background
Marton and Säljö determined these processing levels from a study involving 30
university students. In this study, each student was asked to read a passage on
‘curriculum reform’. Afterwards, the students were audio-recorded as they answered
questions relevant to the passage. In particular, the students were asked to explain the
passage (Marton, 1975). By analysing the transcripts, Marton and Säljö (1976)
suggested that there were generally two types of processing levels, deep and surface. At
a deep processing level, students based their understanding on the meaning of the
learning materials, whilst in the surface processing level, students depended on
memorising materials for reproduction (Richardson, 2005b). Marton (1975) further
qualitatively assessed that the processing levels influenced the students’ performance:
where a deep processing level was associated with a positive learning outcome, whilst a
surface processing level led to a negative learning outcome. Laurillard (1979) contended
that the use of deep or surface level processing was not an inherent quality of the
student, but rather, changed depending on the learning context.
Based on the previous research, Ramsden and Entwistle (1981) extended the
processing levels concept by suggesting that some students may have a preference for
either the deep or surface processing level when reading for a course. They proposed
that students may adopt a ‘deep approach’ or ‘surface approach’ when studying for a
course. Using their terminology, research has developed into the area of ‘approaches to
study’. The term ‘approaches to study’ is most often used when referring to the deep
and surface processing levels in the literature. Using the ‘approaches to study’ term may
prove confusing when referring to the terminology in this literature, that is, the three
approaches. Therefore, to avoid confusion with the three approaches in this thesis and
processing levels will be used whenever referring to ‘the approaches to study’ literature
where avoidable.
Through interviewing students on how they study, Ramsden and Entwistle
developed a 64-item questionnaire which included items representing the processing
levels. These items were statements which required students to use a Likert scale to
indicate their extent of agreement. They named their questionnaire the ‘Approaches to
Studying Inventory’ (ASI) which they administered to students at the end of a course.
Biggs (1987) also extended the work of Marton and Säljö (1976) and developed a 42-
item inventory which had items that measured the surface and deep processing levels of
students reading a course. He called this the ‘Study Process Questionnaire’ (SPQ).
A score for each of the processing levels is found by summing its associated
inventory items. Therefore, a student obtains scores for each processing level through
these questionnaires. The processing level scores cannot be compared against each other
since their final processing level scores only demonstrates whether a student tended to
have a high surface or deep processing level.
2.5.2 Processing Levels: Mathematics, Tasks and Boxes
Both the ASI and SPQ measured processing levels generally for all courses. In
some cases, these questionnaires have been modified to be subject specific. For
example, Crawford, Gordon, Nicholas and Prosser (1998a; 1998b) modified the SPQ to
measure processing levels in mathematics courses. This inventory was called the
‘Approach to Learning Mathematics Questionnaire’ (ALMQ). To test whether the
processing levels impacted on performance in mathematics, Crawford et al. (1998a)
administered the ALMQ to 127 students reading a first year undergraduate mathematics
course. They found through a cluster analysis of the students’ performance scores
(measured by their final examination mark) and the scores from the ALMQ that
similar to the results obtained by Marton (1975) who qualitatively assessed that there
was a relationship between positive learning outcomes and the deep processing level for
students studying the newspaper passage on curriculum reform. Therefore, Crawford et
al. (1998a) were able to show empirically that processing levels were related to
performance in mathematics. Hence, there is an expectation that students using a deep
processing level should perform well on the three task types (i.e. mechanical,
interpretive and constructive).
Crawford et al. (1998a; 1998b) found that students’ conception of mathematics
was related to their processing levels. Through analysing the open-ended responses on
what students thought about mathematics, Crawford et al. developed the Conception of
Mathematics Questionnaire (CMQ) to measure students’ conceptions of mathematics.
They considered that there were two types of conceptions, fragmented and cohesive. In
the fragmented conception, students consider mathematics as numbers, rules and
formulas which are applied to tasks. On the other hand, in the cohesive conception,
mathematics is considered as a way of thinking for carrying out complex problem
solving and for providing new insights into the understanding of the world. Students
with a cohesive concept of mathematics may thus have a better conceptual
understanding of mathematics.
Through factor analysis, the results from the CMQ and ALMQ in the study by
Crawford et al. showed that there was a relationship between the two questionnaires’
scales. That is, the surface processing level was linked to the fragmented conception of
mathematics, whilst the deep processing level was linked to the cohesive conception of
mathematics. Therefore students’ processing levels provide an indication of students’
mathematics conception and shed some insight on how the students will perform on
If students are asked to solve the three tasks, then it is expected that those with a
deep processing level will perform better. However, how well students perform, when
using the three software boxes for solving the three tasks, is uncertain. The reason for
this uncertainty is that some students may have to process more information depending
on which software box they are using and this may affect their performance.
For example, let us take a group of students who use mainly a deep processing
level. If the students are assigned to either the glass-box or the open-box software, then
they will use a deep processing level to understand both the mathematical steps and the
task. However, students assigned to the black-box software only use a deep processing
level for understanding the task, since there are no steps in the black-box software.
Students assigned to the glass-box and the open-box software may become more
cognitively or mentally fatigued than students using the black-box software. The
tendency towards cognitive fatigue in this thesis is based on the definition by Trejo et
al. (2007) and it is understood to be where students who are alert and motivated become
more unwilling to continue undertaking mental work. Therefore, those who have
cognitive fatigue are not likely to expend their cognitive effort in understanding the
information presented to them.
The reason for this possible cognitive fatigue is that students using the glass-box
and open-box software will have to find meaningful learning from the software steps
shown, whereas the students using the black-box software would not have to do so.
Therefore, this cognitive fatigue probably can affect performance using the glass-box
and open-box software.
This analysis suggests that students using the black-box software will have an
advantage and should perform better. However, there is also the possibility that using a
deep processing level to understand the steps in the glass-box and open-box software
surface processing level when looking at the steps in both of these software boxes and
possibly then be able to solve these tasks with almost the same cognitive alertness as the
students using the black-box software. Possibly, one way of knowing whether the
students’ cognitive fatigue is affecting performance is to investigate the explanations
that students are generating for themselves and to ascertain whether these explanations
incorporate cohesive mathematical concepts.