A knowledge stack I have handy on my machine is comprised of high-effect size instructional strategies. That is, lists of evidence-based instructional strategies that can turbo-charge or accelerate student learning. They are pulled annually from the Corwin Press Visible Learning Meta X database (which keeps its list in a table with hyperlinks inside of a frame, oh the horror). I don’t know why they don’t just slap an AI interface to make it more useful.

Takeaway: If you’re wondering what the short version is? The paid, web-based models work best for complex work. My lowly computer just can’t cut it except when working on simple tasks like organizing alpha-sorted lists of info, stuff like that. Also, Gemini 2.5 Pro Experimental yields the best results, even when run via API.

Backstory

I recently upgraded my Google Gemini to Plus so that it is giving me access to some new features in their models. I have been playing with Ollama-based LLMs, including Google’s open source Gemma 3, on my home machine. For fun, I optimized my knowledge stack (a.ka. Strategy Coach for short) to be composed of markdown files.

I was curious how Ollama’s models with Page Assist would compare against Google Gemini’s 2.0 Flash or 2.5 Pro in a Google Gem (which is a knowledge stack). You can see after each phase of learning (e.g. Surface, Deep, Transfer Learning), a list of five strategies with effect size. Following each in square brackets is the markdown file that I prepared. You can access those yourself online via this Google Drive in case you want to give it a go.

A point to keep in mind: A Google Gem created with 2.0 Flash Model allows markdown files, but one created with 2.5 Pro Model does NOT allow markdown. I would also combine markdown file or text documents into ONE file so as not to run afoul of the 10 file upload limit Gem imposes. I don’t know if that’s a good thing or not, but experimentation will reveal that.

Result #1: Google Gem using 2.0 Flash Model

The result is quite good and perfect for my needs.

Here are five instructional strategies for each phase of learning, with “The Jigsaw Method” included in Deep Learning, “Cognitive Task Analysis” removed, and “Outlining and Organizing” added to Deep Learning.

Surface Learning

1. The Jigsaw Method (Effect Size: 0.92): A cooperative learning strategy where students work in groups to master subtopics and teach them to their peers. [cite: CheckEffectSizes.md, 27]
2. Feedback: Corrective, Reinforcement, and Cues (Effect Size: 0.92): Providing learners with information about their performance to guide improvement, reinforce success, and clarify next steps. [cite: CheckEffectSizes.md, HESIS.md]
3. Repeated Reading (Effect Size: 0.80): Reading the same text multiple times to improve reading fluency, accuracy, and comprehension. [cite: HESIS.md]
4. Mnemonics (Effect Size: 0.65): Memory aids that help students retain and recall information more effectively. [cite: HESIS.md]
5. Direct Instruction (Effect Size: 0.56): Explicitly teaching word meanings, especially high-frequency and academic words. [cite: CheckEffectSizes.md, 27, HESIS.md]

Deep Learning

1. The Jigsaw Method (Effect Size: 0.92): A cooperative learning strategy where students work in groups to master subtopics and teach them to their peers. [cite: CheckEffectSizes.md, 27]
2. Argumentation (Effect Size: 0.86): Engaging students in constructing arguments and supporting them with evidence. [cite: CheckEffectSizes.md]
3. Outlining and Organizing (Effect Size: 0.84): Structuring information in a logical, hierarchical format to help students identify relationships, prioritize key ideas, and retain content more effectively. [cite: CheckEffectSizes.md, HESIS.md]
4. Self-Judgement and Reflection (Effect Size: 0.81): Students assessing their own performance, identifying strengths and areas for improvement, and adjusting their learning strategies. [cite: HESIS.md]
5. Elaboration and Organization (Effect Size: 0.72): Deepening understanding by explaining ideas in detail and structuring information for better recall. [cite: HESIS.md]

Transfer Learning

1. Collective Teacher Efficacy (Effect Size: 1.01): The shared belief among teachers in their collective capacity to influence student outcomes. [cite: CheckEffectSizes.md]
2. Integrating with Prior Knowledge (Effect Size: 0.93): Connecting new information to what students already know to facilitate deeper understanding. [cite: HESIS.md]
3. Transfer Strategies (Effect Size: 0.75): Teaching students how to apply learned knowledge and skills to new situations. [cite: CheckEffectSizes.md, HESIS.md]
4. Problem-Solving Teaching (Effect Size: 0.61): Focusing on teaching students effective problem-solving processes. [cite: CheckEffectSizes.md, 27, HESIS.md]
5. Self-Efficacy (Effect Size: 0.71): Students’ belief in their ability to succeed in specific situations or accomplish tasks.

Result #2: Ollama Gemma 3/Phi4 via Page Assist

The result from Gemma 3 and Phi4 models via Ollama was not really relevant or usable. Well, it offered something but it wasn’t on point like the Gem results above.

Result #3: Gemini 2.5 Pro Experimental via Page Assist

This was run through API and worked super fast:

Based on the provided context, here are five instructional strategies effective for each learning phase, meeting your specified requirements:

Surface Learning (Building foundational knowledge, recall)

1. Activating Prior Knowledge: Using pre-assessments, KWL charts, or discussions to surface and connect students’ existing knowledge before introducing new material (Docs 3, 15, 61, 68).
2. Mnemonics: Employing memory aids like acronyms, visual imagery, rhymes, or chunking to help students retain and recall facts and terms (Docs 28, 71, 75).
3. Retrieval Practice (Effects of Testing): Using quizzes, flashcards, or self-testing to encourage active recall and strengthen memory retention (Docs 6, 22, 65).
4. Feedback (Corrective, Reinforcement, Cues): Providing specific information about errors, reinforcing successes, and offering hints to guide students toward correct understanding and foundational skills (Docs 22, 43).
5. Summarizing: Teaching students to identify key points and condense information in their own words using tools like graphic organizers or structured notes (Docs 2, 74).

Deep Learning (Understanding concepts, making connections, critical thinking)

1. The Jigsaw Method: Students become experts on one subtopic and teach it to peers in mixed groups, fostering collaboration and deeper understanding (Docs 4, 9, 13, 20, 25, 36, 37).
2. Outlining and Organizing: Structuring information hierarchically using outlines or graphic organizers to identify relationships and connections between concepts (Docs 4, 5, 54, 59).
3. Critical Thinking: Posing thought-provoking questions, using real-world problems, and encouraging debate to analyze, synthesize, and evaluate information (Docs 0, 53).
4. Constructivist Teaching: Designing lessons around open-ended exploration, hands-on activities, and collaborative projects where students actively build their understanding (Docs 1, 4, 55).
5. Reciprocal Teaching: Students lead group discussions using predicting, questioning, clarifying, and summarizing strategies to enhance comprehension (Docs 23, 52).

Transfer Learning (Applying knowledge in new contexts)

1. Critical Thinking: Applying analysis, synthesis, and evaluation skills to solve novel problems or navigate unfamiliar situations (Docs 0, 53).
2. Constructivist Teaching: Applying conceptual understanding developed through active exploration to new problems or contexts (Docs 1, 55).
3. Evaluation and Reflection (Self-Assessment): Using tools like rubrics and reflective prompts for students to assess their understanding and apply learning to new goals or situations (Docs 8, 14).
4. Feedback (Timing / Corrective, Reinforcement, Cues): Using appropriately timed and specific feedback to help students refine their application of knowledge in varied contexts (Docs 43, 44).
5. The Jigsaw Method: Applying knowledge gained through expert and home group work to solve larger problems or integrate concepts in new ways (Docs 16, 18).