Date of Award

8-2016

Document Type

Thesis

Degree Name

Master of Science (MS)

Legacy Department

Mechanical Engineering

Committee Member

Dr. Gregory Mocko, Committee Chair

Committee Member

Dr. Joshua Summers

Committee Member

Dr. Camerson Turner

Abstract

The overall objective of this research is to address the need for using similar conceptual design problems in experiments in engineering design creativity. This is accomplished by addressing three sub-objectives i) to identify the pattern of design problem usage, ii) to enable comparison between two conceptual design problems based on their natural language representations and iii) to analyze the impact of design problems on effectiveness of example interventions used in user studies in engineering design creativity. Design problems are an essential component of experiments in creativity research.The requirements of experiment’s design sometimes limit problem sharing between researchers or studies conducted by them. For understanding and identifying the design problem usage pattern, two network representations of design problems, connected to each other by authors and papers using them has been used. Both networks indicate that several problems have been used for creativity experiments and suggest the need for using same or ‘similar’ design problems to reduce between-study differences in design problem usage.This addresses the first objective of identifying pattern of design problem usage in creativity research. Problem similarity is assessed using two methods. The first method is based on identification of five structural elements of a design problem namely goals of a problem, functional requirements, non – functional requirements, reference to an existing product and information about end user. The protocol for identifying these elements in problem statement and then comparing design problems is illustrated through two examples. The second method for similarity assessment is based on Latent Semantic Analysis (LSA) of problem statements. LSA provides an objective method to compare semantic similarity of problem statements. Both methods help address the research objective of comparing problems based on their representation but fail to evaluate problem solvability. For understanding whether design problems influence the effectiveness of examples used as interventions, a meta-regression model between effect size and problem size has been used. Regression models suggest that problem size might have a linear relationship with effectiveness of examples for quantity of ideas produced by treatment group participants but enough evidence did not exist to suggest similar relationship for metrics quality and novelty. This addresses the sub-objective of design problems affecting the effectiveness of methods tested in experiments and overall objective of the need for using similar problems in creativity research.

Share

COinS