Transcript Presentation slides as a file

Evaluating the
usability of the BlueJ
environment.
Marilyn Barallon
Supervisor: Dr.Linda McIver
Outline.








What is BlueJ.
What is usability.
How we measure usability.
What has been done in the past.
Studying BlueJ.
What we found- a ‘picture’ of what students are
doing with BlueJ
Conclusion
Further Work
What is BlueJ?

A visual introductory programming environment to help
teach JAVA and OO.

Based on the Blue System developed at Sydney and
Monash University.

Currently being maintained by Deakin University,
Australia and the University of Kent, UK.

Currently being used by 412 institutions worldwide.

Encourages an ‘objects first’ approach to teaching Object
Orientation.
What is BlueJ? (cont’d).

Designed to be simple to use:
 A simple main window:
 UML picture of the program
 ‘Workbench’ area that displays objects that can be interacted
with.
 Simple Debugger:
 Stack window
 Instance, local and static variable windows
 State of the program window (running, finished)
 A simple
syntax directed editor
What is usability?

Subjective

There are three main views from which usability
can be observed and assessed (N Bevan and J
Kirakowski et al):
 Product-oriented view: features of the product
 User-oriented view: mental effort and attitude of
the
user.
 User-performance view: how the user interacts with
the product.

Our research focuses on an assessment of
BlueJ’s usability based on the last two views.
Measuring Usability?

Evaluation frameworks (examples):

Price et al. framework includes 6 broad and general categories:







Scope
Content
Form
Method
Interaction
Effectiveness
Phillips and Mehandjiska et al. framework included 5 categories,
each of which was further broken down into specific
subcategories:





Learnability
Efficiency
Flexibility
Robustness
Feedback
What has been done?

Only formal evaluations so far: surveys and data
loggers.

Problems with survey data:
 Requires students to reflect on past interactions
 Rank thoughts and/or feelings according to unique
judgement of what is ‘very useful’
 Never really used anything else, so how would they
know? (How well did it help you learn JAVA?)

Problems with program logger data:
 Fails
to tell us what was happening between
compilations.
 Or what were the ‘substantial edits’
Understanding the user.

Usability assessment requires understanding of
users.

Surveys and data loggers do not provide a
mechanism for understanding users.

Video-taped ‘think-out-loud’ observational
experiments
 Directly
observe and analyse ‘what’ students are
doing, natural behaviour.
 And ask ‘why’ at the time.
 ‘Discuss’ problems they encounter, at the time. (JAVA
or interface problem)
 Create an overall ‘picture’ of BlueJ user behaviour.
Problems with Observational
Experiments.

Researcher Bias

Reactive behaviour

Limited degree to which we can generalise
behaviour to other people.

In order to mitigate these risks, long term
observational studies and training could be
done.
Studying BlueJ

Observations of students using BlueJ.

Better understanding of users.

A first step towards a comprehensive
usability study.

Detailed picture of student understanding
of BlueJ and JAVA.
Studying BlueJ (cont’d)

Observations involved debugging a program
using BlueJ.
video taped ‘think-out-loud’ sessions
including a short questionnaire
 Two-hour
 Conducted
4 pilot study experiments, with final year
students (1 from software engineering, 2 from digital
systems, 1 computer science).
5 final experiments with 2nd years
(Computing degrees) whom have completed
cse1203- Programming with JAVA 2.
 Conducted
The Program.

Program draws pictures using graphical
output, making it easy to ‘see’ what is an
error.

Students were first shown what the correct
output of the program should look like.

5 semantic bugs.
Problem Solving Abilities.

Did not use BlueJ to explore the program.
 Minimal
exploratory behaviour of the program with the
workbench.
 Jumped straight into code.

Lacked independent exploration and testing
abilities.
 Sought

guidance whenever they could.
As a result, poor hypothesis establishment and
refinement of the behaviours of errors.
Can students use the debugger?

One student.

Common misconceptions and difficulties:



Setting breakpoints on non-executable lines of code.
Instantiation of the debugger window- (which method to call)
Did not understand how the debugger traces the program.





Should I set one on each line?
Will it go from here to there?
Can I set a breakpoint on each line in this class and use the
debugger to step through this class?
Can I move backwards?
These problems were seen in both final and 2nd years.
Common Debugging Strategies.

All students used code comparison at least once.

Students appeared to edit ‘Suspicious’ statements for no
apparent reason.


We hypothesis that students were selecting particular
statements to change based in the unfamiliar ways in
which they were being used


For example in a loop condition, change ‘<‘ to a ‘<=‘
For example in a loop structure, count+=2 instead of count +=1
Interestingly, most students expressed print statements
as their preferred testing strategy.
Problems with BlueJ.

Compiler and JAVA exception error messages.
 Failed
to take notice of them.
 Failed to understand what they were telling them.

Status bar:
 Students
failed to use it to distinguish between when
the machine is busy and when it is free.

‘Remove’ option on the workbench
 Problem
for ‘new’ users learning OO, language syntax
and BlueJ.
Conclusion.

Final and 2nd years do not possess
independent debugging and problem
solving ability.

Students cannot use the debugger.
 Shown
the correct workings of the program
beforehand.
 Graphical output makes it easy to understand
‘what’ is happening.
 Could ask questions at any time.
Conclusion (cont’d)

Lack of understanding of JAVA and OO.

Find ways of teaching and developing
debugging and problem solving ability.
 Enforcing
use of the debugger.

Need to make aware to students how it
facilitates exploration.

The way BlueJ handles its error messages,
displays its status bar and object workbench
need to be investigated further for possible
usability issues and redesign.
Further Work.

We can now move forward and construct a usability
framework specific to BlueJ.

Redesign the way error messages are displayed:



Relocate the error message area to another section of the screen.
Remove the error message area and use pop-up windows
Keep the error message area and better direct the users attention to it.

Further investigation to determine what ‘new’ BlueJ
users think the workbench ‘remove’ option does.

Re-locate the status bar so that it encourages students
to use it to determine when they should perform their
next action.