Afterwards, a large replication project of social-science experiments published in Nature and Science (Camerer et al., 2018) did not reproduce the main effect of the study
If an experiment produces a clear result but a later replication fails to reproduce it, how should researchers decide whether the original finding was unreliable or whether the difference came from the conditions of the replication?
dense columns mean converging work, sparse columns mean thin evidence, empty columns mean a gap
An empty column might mean that the research is missing, but couldn't it also mean our search strategy failed to find the relevant studies? I think we should be careful about claiming a research gap based only on the papers we selected.
Every database has its own syntax for Boolean operators, wildcards and field limits, and a search log records which syntax was used where.
I noticed this when searching for my assignment. The same keywords gave very different numbers of results in ACM Digital Library and Google Scholar. It made me realise that the number of hits depends on how the database searches and indexes papers, not necessarily how much relevant research exists.
ssumption baked in: “why does TypeScript make developers happier?” assumes the effect before anyone has shown it
This connects to my research on AI-generated code. I originally assumed that AI would reduce developers' understanding, but some studies found no clear disadvantage in later maintenance.
Double-barrelled: “how do users choose and remember passwords?
Later in the chapter, there's an example about reducing code review time without reducing defect detection. Isn't that also asking about two outcomes? How do we decide when a question is double-barrelled and when it's acceptable to combine two outcomes?
You are not expected to answer your research questions in the essay — you are expected to justify how you would
in Assignment 3, which asks us to answer our research question using our own data or existing published studies. Should we actually answer the question in the essay, or only explain how we would investigate it?
Both are deductive and postpositivist; the same theory family, two samples, and partly different results — which is what replication looks like in practice.
If two studies use similar theories but find different results, how can we tell whether the differences come from the countries being studied or from how the researchers collected their data?
Deduction runs top-down from theory (typical for quantitative research); induction runs bottom-up from data (typical for qualitative research). In practice research cycles: induction proposes a theory, deduction then tests it.
Is inductive reasoning more suitable for researchers who already have access to data? As students, we often don't have the budget or resources to collect our own data. Can we use findings from existing studies to identify patterns and develop new ideas through inductive reasoning?