These are reading notes on Mixed Methods: A Short Guide to Applied Mixed Methods Research by Sam Ladner (2019). They also draw on Christian Madsbjerg, Roger Martin, Rita McGrath and Cedric Chin. They are adapted from David Drysdale's Moby Diction notes.
1The thesis
Ladner sets out the strengths and weaknesses of qualitative and quantitative research and shows how to use each for what it does best. She argues the difference is deeper than data types. It comes from different epistemological and ontological lenses. These notes follow the thread through three modes of reasoning and the test of falsifiability.
2Three modes of reasoning
2.1Inductive reasoning
Inductive research builds a new theory or hypothesis from observing limited data. Its purpose is not to state a hard truth but to reach a cogent position, one that is likely true given the data. It adds to human knowledge. It creates frameworks that let us predict future events or unobserved phenomena, accurately or not.
Cedric Chin links induction to creativity. Deduction applies an existing mental model to data. Induction takes raw observations and processes them into a new mental model, joining separate threads into a fresh view. We should be ready to destroy those models if they turn out to be incoherent or out of line with reality.
2.2Deductive reasoning
Deductive research tests a theory or hypothesis against observed data. It starts with the general theory and gathers data to confirm or reject it. It always ends in a concrete conclusion: the proposition is true or false.
- It cannot add to human knowledge. It can only check whether existing knowledge is accurate. It cannot predict future events or explain unobserved phenomena.
- A deductive study starts from a falsifiable hypothesis. An inductive study explores open questions to learn what matters and what counts. Combining the two is complicated.
- Roger Martin argues that Western education favors deductive (and inductive) reasoning over abductive reasoning. It accepts a given framework and applies it to a problem, which limits our ability to imagine new approaches. Deduction focuses on what we think "should be." That narrows our view of what is but does not fit the model.
2.3Abductive reasoning
Sherlock Holmes is famous for deduction, but his method is usually abductive. Abductive reasoning infers the most likely explanation from new observed details that fit no existing model.
- It is non-linear. It starts with no supposition or hypothesis, only openness to all kinds of data, which are then synthesized into one or more theories. The theories grow into an explanatory insight.
- It is powerful and also fallible. The expert relies on experience to steer toward insights that are viable and worthwhile.
- Medical professionals use it most. They must make their best guess from incomplete information.
- C. S. Peirce held that abduction is the only mode suited to complex, wicked problems, because it best takes in new insight or knowledge.
- Martin names the capacity for abductive reasoning as a key quality of integrative thought. See the Opposable Mind notes.
- The OODA loop relies heavily on abductive reasoning in its observe and orient phases.
3Falsifiability and data exhaust
Falsifiability is the capacity of a statement to be proven false. A single black swan falsifies "All swans are white." Karl Popper suggested that focusing on falsifiability can move us faster from observation to scientific law. We may never verify that all swans are white, but one non-white swan settles it. So we should look for evidence that disproves our ideas, not just proof that confirms them.
Ladner says to be skeptical of data designed and used without falsifiability in mind. We collect plenty of data "exhaust" from internet-of-things devices and app analytics. Collecting it does not produce knowledge. That takes analysis and interpretation. In her words:
"But I often tell people to tamp down their excitement about data exhaust because none of these data are actually designed for falsifiability in mind, it's simply the detritus of our digital lives. Just because we have more data doesn't mean we are doing better research."
4What to take from it
- Name the mode you are in: building a theory (induction), testing one (deduction), or explaining a surprise (abduction).
- Write hypotheses that a single observation could kill, and go looking for that observation.
- Do not mistake analytics data for research. It needs analysis and a question designed in advance.
- Be ready to discard a mental model that stops matching reality.
5Sources
- Ladner, Sam. Mixed Methods: A Short Guide to Applied Mixed Methods Research. Sam Ladner, 2019.
- Madsbjerg, Christian. Sensemaking: The Power of the Humanities in the Age of the Algorithm. Hachette Books, 2017.
- Martin, Roger L. The Opposable Mind. Harvard Business Review Press, 2007.
- McGrath, Rita Gunther. Seeing around Corners. Houghton Mifflin Harcourt, 2019.
- Chin, Cedric. "Good synthesis is the start of good sensemaking."
Adapted from David Drysdale's Moby Diction notes.