How do problem-solvers approach problems? What cognitive processes are involved in problem-solving? What are the relationships between problem-solving processes and the final performance? Maoxin Zhang`s research is to explore answers to these interesting questions.
Multiple approaches and joint analysis
─Together with my supervisors and co-authors, I have used log files from problem-solving tasks in computer-based large-scale assessments, such as OECD's Programme for International Student Assessment (PISA) and Programme for the International Assessment of Adult Competencies (PIAAC). Log files record the complete human-computer interaction, i.e., time-stamped action sequences, from which we can “read” how respondents solve a given problem and infer their unobservable cognitive processes, Maoxin explains.
─ However, it is challenging to analyze log files due to their great volume, complicated structure, and noise, she continues.
To overcome these challenges, Maoxin has developed a data management procedure including data selection, cleaning, transforming, and storing, in collaboration with her supervisors and co-authors. They have applied multiple approaches, such as social network analysis, machine learning, and latent variable models, to answer their research questions. In addition, they have focused on the joint analysis of performance and process data and implemented a fast estimation algorithm based on Laplace approximations to obtain the estimates accurately and efficiently.
─ In summary, my doctorate dissertation aims to better understand problem-solving using the information from the problem-solving processes through various quantitative research methods, she says
Human beings as the drive
─ Before joining CEMO, I completed my Bachelor’s and Master’s degree in Psychology at Beijing Normal University in China. I have always been interested in human beings, and want to understand what people think, how people behave, and why people are different.
This motivated her to choose psychology as her undergraduate major. She studied many branches of psychology, such as cognitive psychology, social psychology, personality psychology, biological psychology, and counseling psychology.
The more Maoxin read and did psychological studies, the more she realized the importance of appropriate statistical methods to draw reliable conclusions. This is why she chose psychological statistics as her master’s major. She learned a variety of statistical models used in psychology, such as regression models, structural equation modeling, hierarchical linear models, and item response theory.
Maoxin tells us that she really enjoyed learning these methods and love applying them to answer interesting psychological questions, such as how air pollution impacts people’s well-being and how friendship dynamics influence adolescent aggression.
─ My seven years at Beijing Normal University ignited my passion for research. I wanted to delve deeper into interesting and valuable research questions and decided to apply for a Ph.D. position. I was lucky to get the opportunity to study and do research at CEMO. The four years at CEMO are definitely some of the best of my life.
Problem-solving is crucial in our everyday life
Maoxin`s research interests lies at the intersection of problem-solving and process data.
─ My interest in process data dates back to my master’s thesis, in which I jointly modeled response accuracy and response times. I see great potential in mining the information embedded in process data to describe different aspects of test-taking behaviors. I truly believe that process data can help open a window into the “black box” of unobservable mental processes.
She elaborates that she is also interested in problem-solving because our daily and professional lives are full of problems, and therefore problem-solving is crucial and considered as one of the 21st century skills. She is curious about how different people approach various problems, and information about the response processes can perfectly answer her questions.
When it comes to what she hopes to discover from her research and what are some of the most significant current findings, she explains:
─ The overarching goal of my dissertation is to gain a deeper understanding of problem-solving through the analysis of process data. This involves two key elements: understanding problem-solving and using process data. For the first element, we aim to discover respondents’ solution patterns and validate the process indicators for cognitive processes. In addition, we aim to address the challenges of process data analysis. This includes developing and refining methods for structuring process data, extracting valuable features, and reflecting on the latent constructs underlying the extracted process features.
Maoxin continues with explaining that their research suggests that a) respondents showed different solution patterns based on the network features from their action sequences and response times, b) it is essential to examine the internal construct validity when generalizing the process indicators across tasks, c) the relationships between cognitive processes can be highly task-dependent, which should be considered in modeling, and d) the higher-order Laplace approximations can provide a fast yet accurate solution when simultaneously analyzing performance and process data within generalized linear latent variable models.
Broadening accessibility to practitioners
And when we ask her why her research and research findings are important, she replies that the findings of her dissertation make theoretical, practical, and methodological contributions.
─ First, our findings deepen the understanding of solution behaviors and provide evidence for the validity of the process indicators for planning and non-targeted exploration. Second, our research can potentially benefit educational practice. For example, teachers can tailor their instructions according to students’ solution patterns. Third, we implemented a computationally efficient algorithm based on Laplace approximations for generalized linear latent variable models with continuous and discrete data, which can be applied to other mixed data such as game-based assessments, ecological data, and health data. The method has been implemented in an R package called lamle, making it more accessible to practitioners.
Diligence and persistence is at the core
Maoxin says that she has had the honor of meeting and working with many excellent researchers and that they are all curious and passionate about their research.
─ There are some questions that they are eager to find the answers to. This intrinsic motivation is quite important to drive researchers to push their limits and delve deeper and deeper into their field. Second, a good researcher is also a cautious problem-solver because doing research is about solving a series of novel problems, including defining a solvable problem, devising plans, conducting the strategies, monitoring the executing process, and reflecting on the solution. In addition, I truly believe that diligence and persistence are common characteristics of excellent researchers.
The road ahead
As the PhD life soon is over, what career path does she wish for, and what is her dream job? She quickly replies:
─ I want to work as a quantitative methodology consultant, helping people understand their data and find out the suitable way to harness the power of data to answer their questions. I am eager to apply what I have learned to solve challenging problems. I also believe in lifelong learning and hope to learn from my work and progress every day.
As we round up this interview, and if there are people out there curious about the Ph.D life, Maoxin have several tips:
─ First, it should be a strong intrinsic motivation that drives you to do research. Second, supervisors play an important role in the Ph.D. journey. It is important to figure out the best way to work with your supervisors. However, the Ph.D. journey requires more independence than graduate school. Third, pursuing a doctorate degree isn’t easy, so stay resilient and take care of yourself. Fourth, being a researcher or a doctoral candidate is a valuable experience, with both joys and sorrows.
Trial lecture: Maoxin Zhang - Centre for Educational Measurement
Disputation: Maoxin Zhang - Centre for Educational Measurement