5 Layers of Modeling - Color Code Schema_255
pdf
keyboard_arrow_up
School
McGill University *
*We aren’t endorsed by this school
Course
312
Subject
Communications
Date
Jan 9, 2024
Type
Pages
2
Uploaded by MasterBison3853
5 Layers of Modeling Color Code Pink = Theorization What is the question and why are you asking it? Who has already talked about it? What are the core ways they think the world works? What is your "theory" of the problem / issue? This section can also be thought of as a "literature review" (meaning you review existing knowledge / scholarship on this topic). Lancashire example = there is a theory that dementia and "language production" are inversely related (rise in one correlates with a decline of the other). Purple = Conceptualization How do they translate this large theoretical framework into a smaller set of key concepts that will be explored? Think of this as a reduction of complexity, a process of specification. Lancashire example = what does "language production" consist of? For them they translate this larger concept into "vocabulary richness" Blue = Implementation (Measurement) How are these concepts going to be measured? This is a further step of reduction of complexity. No measurement perfectly maps onto a concept. Try to understand what exactly is being measured and how this misses certain aspects of the concept or theory. That is where the opportunity for further research / better knowledge is located. Lancashire example = type/token ratio; repeating phrases; ambiguous words (thing/something/anything). These are the measures of "vocabulary richness". What about rich vocabularies (or language production) do they not capture? Do these matter for the purposes of the study or the goals of inference? Yellow = Selection (Data) What is the data used in the experiment? How was it chosen? How well does it map onto the population they are discussing. Remember, in almost all cases you are working with a "sample" of some larger "population." The population = everything in your category. The sample = the items you actually observed. Lancashire example = sample 1 = Agatha Christie (n=1), i.e. 1 writer of many, many writers ever. Sample 2 = 14 of her 85 novels. Sample 3 = the first 50k words. For each case, ask yourself, why might this sample be a biased representation of the larger population?
Red underlining = Validation There are two kinds of validation: - instrument validation = does your measurement capture what you say it does? - findings validation = Is there a statistically meaningful association between two (or more variables) or is there a statistically meaningful difference in the behavior of two different groups? What is "meaningful"? That is a complicated question. We use quantitative heuristics to help us (statistical tests) but we can also use our judgment by looking at the numbers. Is this decline or difference important
for some real world experience? Lancashire example 1 = they do not validate their measures. I.e. we do not know if type token ratio captures vocabulary richness. They do cite work that validates this. This is common and acceptable. Lancashire example 2 = they use linear regression to validate their judgment that we see a meaningful decline of language production over the course of Christie's career. Question: what are the limitations of using linear regression to validate this judgment? Orange dotted lines - Findings Always a good idea to underline key findings and "takeaways" from the article. These may be summarized in the abstract or the intro or the conclusion. But always be able to summarize "what did this article discover?" What have I learned about this topic from this article? Lancashire example: it appears to be the case that Christie's late novels begin to exhibit a meaningful decline in vocabulary richness that departs from her earlier behavior. While this may be due to stylistic change later in her career, we suspect it is better explained by cognitive decline.
Your preview ends here
Eager to read complete document? Join bartleby learn and gain access to the full version
- Access to all documents
- Unlimited textbook solutions
- 24/7 expert homework help