Creating Frameworks and Models from Expert Insights. | Dr. A.J. Marsden | 463

What is a model? How can you make one? How do you know if it is useful and what is a validated model? Statistician George P. Box once said, "All models are wrong, but some of them are useful." Models give us ways to depict how our insights relate to one another, and create structure that can be supported by statistical data – validating and confirming your ideas. Today, we take a deep dive into the process that turns insights into validated models. We will explore the topic with Dr. A.J. Marsden, Associate Professor at Boston College's Psychology and Human Services Department; she's also Thought Leadership Leverage's resident Organizational Psychologist. A.J. clearly defines models, and describes where they come from and how they function. We learn the basic components of a model, and why models need to be clean, simple, and free from anything that isn't essential. A.J goes on to explain why a good model needs structure and well defined dimensions, and breaks down how insights interact with each other on a fundamental level. In addition, you need to define observable behaviors in order to measure dimensions, providing valuable data for proving the model's validity and reliability. Creating a validated model means finding any weak spots in your content, and correcting them. A.J. describes the process, and goes into the statistical techniques used to create predictive validity. From there, a model can be written up in a technical report that provides functional numerical data to back up the insight and provide proof of your content's validity. If you are a stats nerd, data curious, or just struggling to identify what makes your insights "tick," this is the perfect episode for you. Three Key Takeaways: * When gathering data it is important to use participants that know and understand the concepts being used in the model. * Part of building a good model is understanding how the dimensions interact, and how the ideas support one another. * While you can create a model on your own, it can only be confirmed by a statistics professional who can develop a technical report. Without that, a model isn't validated.