Develop a defensible question, method and independent evidence portfolio with a reproducible revision trail. Specify a research question, operational definitions, sampling frame, causal comparison, qualitative codebook and ethical reporting plan using supplied records.
An implemented sequence of 6 case-based units with additional methods practice and an independent project. These bounded cases introduce disciplinary mechanisms; they are not a complete semester survey, accreditation or verified learning effectiveness. Additional sustained readings, broader topic coverage and human review are needed for institutional credit.
Preparation and direct entry
Interpret supplied records and cite their provenance.
Explain a comparison, basic rates and why association need not establish causation.
No programming required; document analysis decisions so another reader can reproduce them.
Open any lesson. Recaps explain assumed concepts without locking access.
0/14 lessons completed · 0 marked skipped. Skips do not reduce this course's completion denominator or award understanding. Shared lesson work is reused across paths; project evidence belongs to this course.
Lesson sequence
Each linked investigation opens on its own page. Your selected course stays with you as you move between lessons.
A joint display aligns evidence around a shared question. Agreement, complementarity and disagreement are all informative. Interview accounts can suggest mechanisms but do not automatically estimate their population prevalence.
A sample estimate varies across repeated samples. A confidence procedure is evaluated by how often its intervals cover the population value in repeated use under its assumptions. More precision does not repair a biased frame or an ambiguous question.
Purposeful sampling selects cases for a question, such as variation, unusual experience or a proposed mechanism. A run of repeated themes in a narrow recruitment network may reveal similarity in that network rather than adequate coverage of the question.
Reliability concerns consistency; validity concerns evidence supporting an intended interpretation and use. A question can produce stable answers while consistently measuring the wrong thing. Validation examines comprehension, recall, judgment and response options.
Cumulative projectSaved workspace · human review remains separate
CUMULATIVE PROJECT
A reproducible mixed-evidence protocol — Introductory college
Specify a research question, operational definitions, sampling frame, causal comparison, qualitative codebook and ethical reporting plan using supplied records. Add a justified independent question, at least one real institutional or scholarly source, a methods appendix and a response to a competing explanation.
Evidence and review criteria
Context and definitions: accurately identify the relevant institutions, mechanisms and population.
Evidence: connect specific records to claims without presenting fiction as observed history.
Analysis: show calculations or interpretation steps and consider a credible alternative.
Limits: separate facts, assumptions, causal claims and value judgments.
Transfer: adapt to the new case rather than repeating the worked response.
Communication and revision: provide an inspectable artifact and revision trail. Record guidance separately from quality; a submitted artifact is not automatically demonstrated understanding.
Use your lesson captures and citations. This saved workspace supports a reviewer; it does not automatically grade your writing or certify proficiency.
Review examples and quality criteria
These are authored teaching examples for practicing review, not calibrated scoring standards. Apply your course’s specific rubric above. Different conclusions can be defensible when evidence supports them; factual and numerical errors still require correction.
Claim needing revision
The sample average is higher, so the intervention caused the improvement for everyone.
More defensible reasoning
The supplied sample shows a higher average under the stated measurement rule. Compare recruitment, missingness, baseline composition and the distribution of changes. Identify a plausible rival explanation and a new observation that would distinguish it from the proposed mechanism.
Why the revision helps
A descriptive difference, a causal claim and a universal claim require different support. The revision needs actual source IDs, units and design details before it is a finished argument.
Assess the artifact, with assistance recorded separately
Criterion
Developing evidence
Supported evidence
Well-substantiated evidence
Question and context
A broad topic or undefined population.
A focused question, time/place and relevant definitions.
A justified scope with context that changes the interpretation.
Sources and claims
Names a source but does not connect its contents to a claim.
Cites specific records and explains provenance and relevance.
Weighs limitations, corroboration and meaningful disagreement.
Mechanism and analysis
Gives a conclusion with missing steps or unclear units.
Shows a valid causal chain, calculation or interpretation.
Tests a credible alternative and explains sensitivity or uncertainty.
Transfer and revision
Repeats the worked answer despite changed facts.
Adapts the reasoning to the unfamiliar case.
Explains why the change matters and what evidence would reverse the conclusion.
Use “missing” when no relevant evidence is supplied. Do not average these descriptions into an automatic pass score. Record whether the learner followed an example, used guidance or worked independently, and review quality separately.
Locate one specific claim and its supporting source or calculation.
Check the context, units and intermediate reasoning.
Identify one plausible alternative and the evidence that distinguishes it.
Ask for a revision or an unfamiliar application; explain the reason for the request.
The saved project remains a draft or a local submission. This page does not record instructor certification or infer understanding from response length.
Separate application casesVersioned evidence · earlier work remains intact
Apply your reasoning to a separate case
These versioned cases add distinct evidence to the portfolio. Earlier drafts and local submissions remain intact. Write your explanation before revealing review notes when you want to retain a pre-feedback attempt.
Use this packet for a fresh written application. All records below are original fictional teaching materials. They test reasoning under stated conditions and do not establish actual historical or population facts. Your original course-project draft remains separate.
Record scores
Citation ID: social:independent:methods:v1-scores
Invented service-rating data on a 1–5 scale: group A has 4,4,5; group B has 1,2,missing. Missing means no response, not zero. These six records are a convenience sample.
Invented interview excerpts: A1 'The hours fit my work'; B1 'The opening hours exclude my shift'; B2 'The hours are fine but I cannot travel there.' A2/A3/B3 have no interview.
Record coding
Citation ID: social:independent:methods:v1-coding
A coder labels every negative rating 'schedule problem'. Another proposes separate scheduling and transport codes. The coding unit is each excerpt; multiple codes are permitted if justified by its text.
This case is shared across paths that use it. Revisiting it in another course retains drafts and known review-note exposure; it is not a new unseen assessment.
Loading application evidence…
NEW APPLICATION · VERSION 1
Does a training offer establish the effect of training?
Use this packet for a fresh written application. All records below are original fictional teaching materials. They test reasoning under stated conditions and do not establish actual historical or population facts. Your original course-project draft remains separate.
Fictional trial: 200 eligible volunteers are randomly assigned, 100 to an offer and 100 to no offer. The offer includes access to training plus a 20-unit travel payment paid immediately, whether or not the person attends. No-offer participants receive neither from this program.
60 offered participants attend training. In the no-offer group, 20 obtain equivalent training elsewhere. Attendance is observed for everyone. Participants choose whether to attend after assignment.
Complete follow-up finds mean job-search spending of 54 units in the offer group and 50 in the no-offer group. Credential completion is observed for 80 offer-group participants, of whom 24 complete, and all 100 no-offer participants, of whom 30 complete. The remaining 20 offered participants have missing credential outcomes.
An invented evaluator says the program proves training raises job-search spending by 10 units for everyone and increases credential completion because 24/80 is as high as 30/100. The evaluator treats an unobserved completion as a failure.
This case is shared across paths that use it. Revisiting it in another course retains drafts and known review-note exposure; it is not a new unseen assessment.
Record scores
Citation ID:
social:independent:methods:v1-scoresInvented service-rating data on a 1–5 scale: group A has 4,4,5; group B has 1,2,missing. Missing means no response, not zero. These six records are a convenience sample.