Evaluation of Computer Games Developed by Primary School Children to Gauge Understanding of Programming Concepts
Amanda WilsonProf. Thomas ConnollyDr Thomas [email protected]@[email protected]
Presentation structure•Introduction•Programming for Children•Games-based construction•Scratch •Methods •Results•Discussion•Further Work
Introduction•New Curriculum in Scottish Schools
•New Computing Science Strand
•GBL encouraged
•Lack of Empirical evidence of GBL in schools
Programming for children•Programming can be taught from an early
age.
•Variety of languages aimed at the novice user such as: Alice, Toontalk, Starlogo, Etoys, Scratch and Kodu.
•Initiatives such as after school clubs also run to introduce children to programming.
Games-based Construction• Game construction is one of 3 ways to introduce
games-based learning (Van Eck, 2006).• Commercial off the Shelf games (COTS)• Using games specifically created for education• Use of game construction as an approach to GBL
• Game construction is relatively unexplored within the primary classroom (Baytak and Land 2010).
• Applications such as Scratch make games construction more accessible to children.
Scratch
• A simple game creation platform using a visual programming language.
• Developed originally for use in after school computer clubhouses
Scratch
Objectives•Develop coding scheme for Scratch games
•To evaluate games made by primary school children using Scratch
Methods• 60 children aged between 8 and 11 from 3
classes (Primary 4, 5/6 and 6/7) in one school participated
• They undertook lessons in pairs (or groups of 3)
• 8 one hour lessons were delivered▫Lesson 1 – Introduction to Scratch▫Lessons 2-4 – Creating a simple maze game▫Lessons 5-8 – Extending maze game/Creating new
game
Methods • A game coding scheme was adapted from Denner,
Werner and Ortiz (2011) and refined based on the programming concepts that can be learned with Scratch
• The coding was split into 3 main categories and 22 subcategories▫ Programming Concepts▫ Code organisation▫ Designing for Usability
• Each game was coded for the presence of each element (either 0/1) or in some cases the extent to which that element was used within the categories using a range from either 0-2 or 0-3.
Methods – Game Coding • Programming Concepts category focused on the
concepts that Scratch supports that include:▫ Sequence▫ Iteration▫ Variables▫ Conditional Statements▫ Lists ▫ Event handling▫ Threads▫ Coordination and Synchronisation▫ Keyboard Input▫ Random Numbers▫ Boolean Logic▫ Dynamic Interaction▫ User Interface Design
Methods – Game Coding• Code organisation category:
▫Extraneous blocks▫Sprite names (the default is overridden).▫Variable names
• Designing for usability category:▫Functionality▫Goal▫Sprite customisation▫Stage customisation▫Instructions clear▫Game originality
Results• During the study 29 games were created by the groups
of children
• The mean game score was 48% and a Kruskal-Wallis one way analysis of variance test showed no significant difference in game scores between class groups (χ 2
= 0.072, p < 0.965) or between gender groupings (χ 2
= 0.483, p < 0.785).
Class Stick with maze game
Adapt maze game
(change background adapt game)
Adapt maze game (change background
adapt game to two player)
Create new game (come up with another idea
other than maze game)
P4 0 3 4 2
P5/6 3 1 0 6
P6/7 2 2 0 6
Results• The games varied in their complexity with over 90%
of them using the keyboard or mouse to control the game with others either not having been completed or having the user answer questions instead.
• Kruskal-Wallis tests showed no significant difference in concepts used between class groups (χ 2
= 0.176, p < 0.916) or between gender groupings (χ 2
= 0.472, p < 0.790). However, Mann-Whitney U test showed a significant difference between the concepts used in maze-based games compared to the original games made by all classes (Z=-2.535, p < 0.010).
ResultsProgramming Concepts
% of games including programming concepts
Sequence 93Event Handling 90Conditional Statements 86Threads 83Variables 72Coordination and Synchronisation 72Iteration 55Keyboard Input 7Random Numbers 3
Results•21% of the games included extraneous
blocks.
•72% of the games included meaningful variable names most games only included a timer, however some games also implemented a scoring system as well.
•Only 3% of games had changed the default sprite name.
Results•Out of the 29 games only 1 had no
functionality. The others had varying degrees with 28% of the games being fully functional.
Designing for Usability % of games Functionality 97Sprite customisation 97Stage customisation 93Clear Instructions 86Game originality 83Goal 59
Game Example 1
Adapted maze game from Primary 5/6 class
Game Example 1•The sprite in the game is controlled by the
arrow keys. Conditions are set that if the sprite touches the green walls it will bounce off them while the timer counts down how many seconds left the user has.
Game Example 2
2 Player adapted maze game from the Primary 4 class
Game Example 2• The sprites in the game are controlled by
the arrow keys for player 1 and player 2 used the w,a,s,d keys.
•There were 2 variables within the game and while the timer worked the scoring wasn’t working correctly.
Discussion•Few studies have been undertaken to look
at learning programming concepts through game construction for children.
•These studies have shown that children are able to learn programming concepts through game construction although these have been conducted mainly after school and little is known within the classroom setting.
Discussion•The 8 lessons were a basic introduction to
game making with Scratch.•Most children were successful in creating
their own game. •With those in the Primary 4 class
preferring to adapt the maze game rather than create their own.
•60% of each Primary 5/6 and 6/7 class opted to create their own game.
Discussion• The concepts most commonly used by the children in their
games were similar to those found in Maloney et al (2008)▫ Key handling▫ Iteration▫ Conditional statements
• Gender groupings did not have a significant effect on the games created, however the Primary 5/6 class which consisted of mixed gender groups did have the highest mean score as well as the most functional games of all 3 classes.
• Similar to the results of Baytak and Land (2011) this study shows that given a short timeframe children were able to make progress with Scratch show advances in their programming skills.
Future Work•Work with more classes (from Primary 4
to Primary 7) in other schools within Glasgow to attain further empirical results to produce more statistically significant evidence.
•Refinement of the instrument of evaluation through a series of pilot studies.
Thank you