Multi-Agent Psychometric Item Generator

MAPIG, which stands for Multi-Agent Psychometric Item Generator, is a system built to help create questions for surveys and tests. It uses smart computer programs called a multi-agent system and Large Language Models, which are like advanced AI, to develop these questions. This system guides users through the whole process, from deciding what to measure to getting the final questions ready for testing.

Benefits

MAPIG makes developing assessment questions easier and more reliable. It follows proven methods for creating good questions and uses AI to help draft, check, and improve them. The system keeps a record of every step taken, so you can see how each question was developed. It also allows for human feedback to make the questions even better. This ensures the process is clear, can be repeated, and the questions are ready for real-world testing.

Use Cases

This tool is useful for anyone needing to create surveys, questionnaires, or tests. This includes researchers, educators, and businesses that want to measure opinions, knowledge, or traits accurately. It can be used to develop scales for measuring anything from customer satisfaction to employee engagement or academic understanding. The system can handle both simple and complex measurement goals.

Vibes

The system is designed to be transparent and reproducible, meaning users can trust the process and results. It aims to produce psychometrically sound items, which are questions that accurately measure what they are intended to measure. The integration of human feedback helps ensure the final items are relevant and well-written.

Additional Information

MAPIG uses 13 different AI agents, each with a specific job. Some agents find information from research papers and the web, others figure out the structure of what needs to be measured, and some write the actual questions. There are also agents that review the questions for clarity, bias, and content accuracy. Finally, agents help decide if questions are good enough or need revision and then make the necessary changes. The system also provides tools to analyze the quality of the generated questions before they are used in a real study.