The first lecture is a mish-mash of many things, but mostly focuses on setting the context rather than causal modeling as such. Introduction to the class. Examples of causal modeling. Measuring air pressure. Is a medicine effective? Does usability affect consumer preferences? Is a “rich interaction camera” better than menu-controlled camera? Why causal modeling in design and practice-based research? Alternatives briefly (lecture 10 is devoted to these). Working methods and expectations of the class.
This lecture looks at the basic elements of causal modeling, starting from specifying a research question from theory through concepts to indicators. Second, it explains the basis of causal modeling, including the types of variables and how they are specified. Key concepts of independent, dependent, and intervening variables.
The main theme of this lecture is elaboration: how one can clarify spurious relationships between variables. The second theme of the lecture is how to represent one’s theory, starring from intuitive verbalizations ending up to simultaneous equations. It ends with a critical note about the need to state one’s problem in a mathematical form.
A quick look at measurement, including the types of information, indicators and indexes, data matrix, the notion of covariation with examples of its key measures, and the role of randomization.
The main types of experimental research design, including classic experimental research (incl. double blind experiments), simple factorial designs, latin squares, quasi-experimental research. Laboratory. Main threats to validity. Reactivity.
This lecture looks at the main types of alternative research designs, including case studies, comparative studies, evaluation research, and field experiments. Is a government policy an experiment? “In the wild.”
This lecture looks at some of the main techniques for making decisions about associations between independent and dependent variables. It main covers elementary statistics and some qualitative means for analyzing associations between variables, but also mentions briefly more complicated analytic techniques and their use situations. The lecture ends with the main differences between statistical and experimental controls. It also looks at analytic strategies and the main tactics generalization.
((After Antti's talk, I went on to introduce analysis)).
This lecture looks at some problems and extensions of causal modeling, including the notion of “other things being equal” (ceteris paribus) and its justifications, the notion or error term and its role in explanation, recursive forms of causation, the problem of small N. Other things are issues that come up during the class.
The final lecture gets back to the basic question of the class: when and how can one do causal modeling, and what kinds of uses does this methodology have in design research and practice-based research. It also looks at the limits of the methodology through an example from ergonomics. This lecture gets back to some ethical issues typical to causal modeling, esp. experimental research what often involved deception of some kind.