This week we discussed “Hypothesis Testing Methods at Scale” and “Causal Inference” based on the reading of “Counterfactuals and Causal Inference” by Stephen Morgan and Christopher Winship (Cambridge Press).
Sorry to say but the class was extremely boring… looking in more detail what we can do with Machine Learning and textual analysis…
This week we kept on advancing on the Machine Learning, more specifically about textual analysis. We went over the “bag of words” technique of trying to make predictions about the type of annotation created with Laguna Stories.
Seems a little counterintuitive that by simply analyzing the word count in the comments would generate any viable conclusion, yet it seems like it could. Not to say that this is magic, but very detailed work in coding the information, training the algorithm with enough data, and then iterating through the process of adjusting the parameters, groupings, and sometimes even going back to coding the data.
Great talk today about how machine learning – the second of a series. We looked at collecting, cleaning, and coding data to create the training data set in a supervised learning model.
Emily Schneider, PhD candidate in Learning Sciences and Technology Design at Stanford University’s Graduate School of Education.
She is doing research on Lacuna Stories – a very cool document annotation tool that teachers can use to observe reading comprehension and critical interpretation. It was created by the Poetic Media Lab – yet another very interesting group here at Stanford that I didn’t know about… FoMO!
Had a great presentation today on Machine Learning – the first of a series of 3 sessions -by Iris Howley. We went over the basic types of machine learning such as supervised and unsupervised learning as well as styles:
- Numeric Prediction
- Neural Networks
- Deep Learning
- Decision Trees
Good class. Went over research terminology and talked in pairs about each term to clarify any missconceptions. My pair was Anita Tseng, who is doing some very intersting research on common scientific misconception on the internet and social media.
Quick exercise we did in class… write down our own research goal and discuss with our partners based upon it.
My Research Goal
- To understand what are the best practices an instructor should use to create online courses.
(to do… go through Candace’s slides to evaluate my research question)
The Research Methods Knowledge Base: 3rd edition by William M.K. Trochim and James P. Donnelly. Chapter 1
- 1-1 The Language of Research
- 1-1a Types of Studies (Cumulative)
- Relational: two variables
- Causal: most demanding
- 1-1b Time in Research
- Cross-sectional – single point in time
- Longitudinal studies – over time
- Repeated measures
- Two or a few waves of measurement
- Time series
- Over 20 waves of measurement
- 1-1c Type of Relationships
- Simple correlational relationship – act in synchrony
- Causal relationship – one explains another
- Third-Variable problem – something else which affects both variables
- No relationship
- Positive relationship
- Negative relationship
- 1-1d Variables
- Dependent / Independent
- Exhaustive: holds all possible values
- Mutually exclusive: employed, not employed + 2nd job
- 1-1e Hypotheses
- Prediction, in concrete terms
- Not all research has a hypothesis – exploratory
- Alternative hypothesis
- Null hypothesis
- One-tailed hypothesis
- 1-1f Types of Data
- Quantitative & Qualitative
- “All quantitative data is based upon qualitative judgements; and all qualitative data can be described and manipulated numerically.” (Trochim & Donnely, 2006)
- 1-1f The Unit of Analysis
- Individuals, Groups, Artifacts, Geo, Social
- 1-1h Research Fallacies
- Ecological fallacy – group stereotyping
- Exception fallacy – one does not represent the group
- 1-2 Philosophy of Research
- 1-2a Structure of research
- Hourglass shape – broad area of interest to measurement and observation to generalize back to question
- Components of a study
- Research problem
- Research question
- Program (cause)
- Outcomes (effects)
- 1-2b Deduction and Induction
- More general to more specific
- Top-down approach
- From specific to general
- Bottom-up approach
- 1-2c Positivism and Post-Positivism
- Epistemology – philosophy of knowledge
- Methodology – how
- Empiricism – Knowledge as only what could be observed and measured
- Deductive reasoning
- Theoretical reasoning & experience-based evidence
- Critical realism
- External reality
- Never accurate (critical)
- World is solely a creation of your mind
- Reality is a conceptual construction
- Evolutionary epistemology or natural selection theory of knowledge
- Ideal have survival value
- knowledge evolves through a process of variation, selections, and retention (evolution)
- 1-2d Validity
- Best available approximation to the truth of a give proposition, inference, or conclusion
- “Validity of what?”
- Four types
- Is there a relationship between cause and effect?
- Is the relationship causal?
- Can we generalize to the constructs?
- Cause Construct
- Your theory about the cause in a cause-effect relationship
- Effect Construct
- Your theory what the outcome is in a cause-effect relationship
- Can we generalize to other persons, places, or times?
- Threat to Validity
- Reasons why your conclusion or inference might be wrong
- Conclusion Validity
- The degree to which your conclusions about relationships in your data are reasonable
- 1-3 Ethics in Research
- 1-3a The Language of Ethics
- Voluntary participation
- Make sure there is no coercion to participate
- Informed consent
- Information about procedures and risks involved
- Their personal/individual information or identity will not be released beyond the scope of the study
- Institutional Review Board (IRB)
- Panel who reviews research proposals with respect to ethical implications
- 1-4 Conceptualizating
- Concept Mapping
- 2D graphs of a group’s ideas – used to develop conceptual framework for a research project
- 1-4a Problem formulation
- Comes from practical problems in the field
- Request for Proposals (RFP)
- Literature Review
- 1-4b Concept mapping
- Pictorial representation of ideas
- 1-4c Logic models
Nathan, M., & Alibali, M. (2010). Learning sciences. WIREs Cognitive Science. DOI: 10.1002/wcs.54
Engineering Education – Week 2.1 – Reading Notes
Great initial class going over what the Lytics Lab does: looks at data provided by online education and contributes to the emerging field of Learning Sciences. Sounds like it will be a perfect match for the the Engineering Education course which looks at the ‘front-end’ of education while this class looks at the ‘back-end’.
Terrible picture – got into class early – class is packed actually!