Read
Build the mental model
Move through the guided explanation first so the central distinction and purpose are clear before you evaluate your own work.
Bayesian Probability
Bayesian Comparison of Rival Hypotheses
Connects Bayesian updating to comparative reasoning between competing hypotheses using the Bayes factor and qualitative Bayesian comparison.
Read the explanation sections first, then use the activities to test whether you can apply the idea under pressure.
Start Here
What this lesson is helping you do
Connects Bayesian updating to comparative reasoning between competing hypotheses using the Bayes factor and qualitative Bayesian comparison. The practice in this lesson depends on understanding Likelihood, Posterior Probability, and Bayes Factor and applying tools such as Respect Base Rates and Distinguish P(E | H) from P(H | E) correctly.
How to approach it
Read the explanation sections first, then use the activities to test whether you can apply the idea under pressure.
What the practice is building
You will put the explanation to work through comparison exercise, quiz, evaluation practice, diagnosis practice, analysis practice, rapid identification, and argument building activities, so the goal is not just to recognize the idea but to use it under your own control.
What success should let you do
Compare 4 rival-hypothesis cases using prior odds, qualitative Bayes factors, and posterior direction.
Reading Path
Move through the lesson in this order
The page is designed to teach before it tests. Use this sequence to keep the reading, examples, and practice in the right relationship.
Study
Watch the move in context
Use the worked examples to see how the reasoning behaves when someone else performs it carefully.
Do
Practice with a standard
Only then move into the activities, using the pause-and-check prompts as a final checkpoint before you submit.
Guided Explanation
Read this before you try the activity
These sections give the learner a usable mental model first, so the practice feels like application rather than guesswork.
Core idea
Rival hypotheses always frame the question
Bayesian updating is always, implicitly, a comparison. Even when you ask 'how likely is H given E?', the answer depends on how likely E is under not-H — that is, under the rival hypotheses. You cannot separate the question of how much evidence supports H from the question of how well the evidence is explained by alternatives.
This is why the best Bayesian practice is to make the rivals explicit. Instead of asking 'is H well-supported?', ask 'is H better supported than H'?' The comparison forces you to evaluate both hypotheses against the same evidence, which exposes hidden assumptions and prevents one-sided reasoning.
What to look for
- Always name at least one rival hypothesis.
- Evaluate the evidence against both simultaneously.
- Ask whether the evidence is also likely under the rival.
Bayesian confidence is always relative — to the rivals you put in the comparison.
Key concept
The Bayes factor
The Bayes factor is the ratio P(E | H1) / P(E | H2). It captures how much more the evidence favors H1 over H2. A Bayes factor of 1 means the evidence is equally compatible with both hypotheses — it tells you nothing to prefer one over the other. A Bayes factor of 10 means the evidence is ten times more expected under H1 than under H2, which is substantial.
The Bayes factor is a clean way to describe the weight of evidence independent of the prior. Two observers with different priors can look at the same Bayes factor and agree on how much to shift — even if they disagree on where they started. That is what makes it useful for scientific communication.
What to look for
- Compute or estimate P(E | H1) and P(E | H2).
- Take the ratio.
- Interpret: 1 means no effect, 10 is substantial, 100 is strong.
The Bayes factor isolates the evidence's weight from the starting point.
Practical skill
Qualitative Bayes factors
You can estimate Bayes factors qualitatively without any numbers. Ask: is the evidence much more expected under H1 than H2, somewhat more expected, or roughly equal? The three qualitative buckets correspond roughly to Bayes factors of 10+, 3 or so, and around 1. With practice, you can distinguish these without ever writing a formula.
This qualitative move is especially valuable when precise probabilities are not available. In forensic reasoning, medical decisions, and policy debates, you often have to say which way the evidence leans and by roughly how much. The qualitative Bayes factor gives you a language for that without pretending to false precision.
What to look for
- Classify the evidence as strong, moderate, or weak.
- Use your qualitative judgment to set the direction and rough size.
- Avoid false precision; say 'roughly 10x' or 'about even' rather than '9.7'.
Qualitative Bayes factors are rigorous without being mathematical.
Putting it together
From comparison to posterior
The posterior is the prior multiplied by the Bayes factor (in odds form): posterior odds = prior odds × Bayes factor. This is the cleanest way to write the update. If you started 2:1 in favor of H1 and the evidence gives a Bayes factor of 5 for H1, you end up 10:1 in favor of H1, which is about 91%.
In qualitative form: a strong prior needs strong evidence to overturn; a weak prior can be decisively shifted by strong evidence; weak evidence against a strong prior produces almost no change. All of these patterns follow from the same multiplication rule, and they match the ordinary intuition that extraordinary claims require extraordinary evidence.
What to look for
- Convert the prior to odds if working quantitatively.
- Multiply by the Bayes factor.
- Read off the posterior odds and convert to probability if needed.
Posterior = prior × evidential weight. Extraordinary claims require extraordinary evidence — literally.
Core Ideas
The main concepts to keep in view
Use these as anchors while you read the example and draft your response. If the concepts blur together, the practice usually blurs too.
Likelihood
The probability of the observed evidence on the assumption that a given hypothesis is true, written P(E | H).
Why it matters: Likelihood measures how well a hypothesis predicts the evidence — not how probable the hypothesis is.
Posterior Probability
The revised degree of confidence in a hypothesis after incorporating new evidence, written P(H | E).
Why it matters: The posterior is the result of rational belief updating — what you should believe now.
Bayes Factor
The ratio of likelihoods under two rival hypotheses, P(E | H1) / P(E | H2), which captures how strongly evidence favors one over the other.
Why it matters: Bayesian reasoning often compares relative evidential support between hypotheses rather than single isolated probabilities.
Reference
Open these only when you need the extra structure
How the lesson is meant to unfold
Concept Intro
The core idea is defined and separated from nearby confusions.
Rule Or Standard
This step supports the lesson by moving from explanation toward application.
Worked Example
A complete example demonstrates what correct reasoning looks like in context.
Guided Practice
You apply the idea with scaffolding still visible.
Assessment Advice
Use these prompts to judge whether your reasoning meets the standard.
Mastery Check
The final target tells you what successful understanding should enable you to do.
Reasoning tools and formal patterns
Rules and standards
These are the criteria the unit uses to judge whether your reasoning is actually sound.
Respect Base Rates
A probabilistic judgment should not ignore background prevalence or prior probability when the context makes it relevant.
Common failures
- A striking test result is treated as if it overrides the base rate automatically.
- Rare-event contexts are assessed as though all hypotheses started equally likely.
Distinguish P(E | H) from P(H | E)
A likelihood is not the same thing as a posterior probability. Swapping them is the 'prosecutor's fallacy'.
Common failures
- The probability of evidence given a hypothesis is mistaken for the probability of the hypothesis given the evidence.
- Diagnostic accuracy is confused with posterior certainty.
Update Proportionately to Evidence
Belief revision should reflect both prior plausibility and the relative explanatory weight of the evidence — not the vividness or novelty of the evidence.
Common failures
- A small piece of evidence causes an excessive revision.
- Strong contrary evidence produces almost no change in confidence.
Compare Evidence Under All Rival Hypotheses
The weight of evidence depends not only on how well it fits the favored hypothesis, but also on how well it fits the rivals.
Common failures
- Asking only whether the evidence fits H and ignoring whether it fits not-H equally well.
- Treating evidence as strong because it 'supports' H without checking whether it also supports rival hypotheses.
Patterns
Use these when you need to turn a messy passage into a cleaner logical structure before evaluating it.
Bayesian Update Schema
Input form
evidence_assessment_problem
Output form
prior_likelihood_posterior_analysis
Steps
- State the hypothesis under evaluation.
- Identify the relevant prior probability or base rate.
- State how likely the evidence would be if the hypothesis were true (the likelihood).
- State how likely the evidence would be if the hypothesis were false (the false positive rate).
- Compute or estimate the posterior proportionately.
Watch for
- Skipping the prior entirely.
- Using only one-sided likelihood information.
- Treating the posterior as certainty rather than a revised degree of support.
Qualitative Bayesian Comparison
Input form
competing_hypotheses_with_evidence
Output form
relative_support_judgment
Steps
- State the competing hypotheses.
- Compare their priors qualitatively (which was more plausible before the evidence?).
- Compare how strongly each predicts the evidence.
- Multiply qualitatively: a higher prior and a higher likelihood both push the posterior up.
- State the posterior ranking with appropriate caution.
Watch for
- Assuming the hypothesis with the most vivid story automatically gets the higher posterior.
- Ignoring that weaker priors can sometimes be overcome by much stronger evidence.
Worked Through
Examples that model the standard before you try it
Do not skim these. A worked example earns its place when you can point to the exact move it is modeling and the mistake it is trying to prevent.
Worked Example
Unexpected Sensor Reading
Bayesian reasoning compares how evidence shifts rival hypotheses rather than only asking which sounds appealing. A favorable prior can be overcome by a strong Bayes factor.
Candidate Hypotheses
- The instrument malfunctioned.
- The environment really changed in the reported way.
Comparison
Posterior Shift
The evidence pushes confidence toward the environmental explanation; it may end up roughly even or slightly favoring the environmental shift, depending on how much weight you give the prior.
Prior Plausibility
Instrument malfunction is somewhat more common in this setup, so the prior slightly favors malfunction.
Likelihood Of Evidence
A true environmental shift would match the specific reading pattern more tightly than a random malfunction. Bayes factor modestly favors the environmental shift.
Worked Example
The prosecutor's fallacy in forensic evidence
The prosecutor's fallacy shows why swapping conditionals can mislead juries and anyone else interpreting probabilistic evidence.
Setup
A DNA match has a random-match probability of 1 in 1 million. The prosecutor argues: 'The probability of a random match is 1 in 1 million, so there's only a 1 in 1 million chance the defendant is innocent.'
Analysis
This confuses P(E | not-H), the random-match probability, with P(not-H | E), the probability the defendant is innocent given the match. The two are different. Without the prior — how many people could plausibly have been at the scene? — the random-match probability doesn't directly tell you the posterior.
Correct Bayesian Treatment
If 100 people could plausibly have been at the scene (prior odds 1:99), and the Bayes factor from the match is 1,000,000, then posterior odds are 1,000,000:99 ≈ 10,000:1 in favor of the defendant being the source. That's still very strong, but it's computed correctly rather than by swapping the conditional.
Pause and Check
Questions to use before you move into practice
Self-check questions
- How plausible were the hypotheses before the evidence?
- Which hypothesis predicted the evidence better?
- Does the Bayes factor justify the size of the update I'm making?
Practice
Now apply the idea yourself
Move into practice only after you can name the standard you are using and the structure you are trying to preserve or evaluate.
Comparison Exercise
Best ExplanationCompare Rival Hypotheses Bayesianly
For each case, identify the two hypotheses, estimate the prior odds qualitatively, estimate the Bayes factor qualitatively (strong/moderate/weak, direction), and state the posterior direction.
Four Bayesian comparison cases
You don't need precise numbers. Use natural language: 'roughly even,' 'modestly favored,' 'strongly favored.'
Case 1 — Sensor reading
An industrial sensor reports a pressure spike. Hypothesis 1: the system is really experiencing a spike. Hypothesis 2: the sensor malfunctioned. Recent maintenance logs show no prior malfunctions this year. The pressure reading is consistent with a known failure mode of an upstream valve.
Which hypothesis better predicts the specific reading pattern, and what's the prior on sensor malfunctions?
Case 2 — Forensic match
A DNA sample from a crime scene matches the defendant. The match's random-match probability in the population is 1 in 1 million. Hypothesis 1: the defendant is the source. Hypothesis 2: someone else unrelated left the sample.
What's the prior on the defendant versus a random stranger, and how does the Bayes factor combine with that?
Case 3 — Financial fraud
An accountant notices unusual entries in a ledger. Hypothesis 1: fraudulent activity. Hypothesis 2: legitimate but rushed bookkeeping under deadline pressure. The unusual entries happened during a known quarter-end crunch.
Which hypothesis better explains the timing as well as the style of the entries?
Case 4 — Climate attribution
A region experiences an unusually severe heat wave. Hypothesis 1: climate change increased the probability. Hypothesis 2: natural variability. Historical records show the current heat wave is 5 times more likely under a climate-change model than under natural variability alone.
What's the Bayes factor, and how does it combine with prior confidence in climate change?
Use one of the cases above, compare the competing explanations, and defend the one that best fits the evidence.
Quiz
InductiveScenario Check: Bayesian Comparison of Rival Hypotheses
Each question presents a scenario or challenge. Answer in two to four sentences. Focus on showing that you can use what you learned, not just recall it.
Scenario questions
Work through each scenario. Precise, specific answers are better than long vague ones.
Question 1 — Diagnose
A student makes the following mistake: "Choosing the most vivid hypothesis rather than the best probabilistic update." Explain specifically what is wrong with this reasoning and what the student should have done instead.
Can the student identify the flaw and articulate the correction?
Question 2 — Apply
You encounter a new argument that you have never seen before. Walk through exactly how you would compute bayes factor, starting from scratch. Be specific about each step and explain why the order matters.
Can the student transfer the skill of compute bayes factor to a genuinely new case?
Question 3 — Distinguish
Someone confuses bayes factor with likelihood. Write a short explanation that would help them see the difference, and give one example where getting them confused leads to a concrete mistake.
Does the student understand the boundary between the two concepts?
Question 4 — Transfer
The worked example "Unexpected Sensor Reading" showed one way to handle a specific case. Describe a situation where the same method would need to be adjusted, and explain what you would change and why.
Can the student adapt the demonstrated method to a variation?
Use one of the cases above, identify the evidence base, and judge how strong the conclusion is once you account for rival factors.
Evaluation Practice
InductiveStrength Ranking: Bayesian Comparison of Rival Hypotheses
Rank these inductive arguments from strongest to weakest. Explain what makes one stronger than another.
Practice scenarios
Work through each scenario carefully. Apply the concepts from this lesson.
Argument 1
In a survey of 10,000 patients across 15 hospitals, the new treatment showed a 40% improvement over the control group.
Argument 2
My three friends who tried the supplement said they felt better, so the supplement probably works.
Argument 3
In every chemistry experiment conducted over 200 years, mixing sodium and chlorine has produced table salt.
Use one of the cases above, identify the evidence base, and judge how strong the conclusion is once you account for rival factors.
Diagnosis Practice
InductiveSample Critique: Bayesian Comparison of Rival Hypotheses
Evaluate the sampling method in each scenario. Identify potential biases and suggest improvements.
Practice scenarios
Work through each scenario carefully. Apply the concepts from this lesson.
Study A
To learn about national reading habits, researchers surveyed visitors at a book festival and found that 95% read more than 10 books per year.
Study B
A tech company surveyed its own users about smartphone satisfaction and concluded that 88% of Americans are satisfied with their phones.
Study C
Researchers randomly selected 5,000 households from every state and conducted in-person interviews about dietary habits.
Use one of the cases above, identify the evidence base, and judge how strong the conclusion is once you account for rival factors.
Analysis Practice
InductiveAnalogy Builder: Bayesian Comparison of Rival Hypotheses
Assess the strength of each analogical argument. Identify relevant similarities and differences, then explain whether the analogy supports the conclusion.
Practice scenarios
Work through each scenario carefully. Apply the concepts from this lesson.
Analogy 1
The human brain is like a computer. Computers can be reprogrammed. Therefore, human habits can be reprogrammed.
Analogy 2
A company is like a ship. A ship needs a captain. Therefore, a company needs a strong CEO.
Analogy 3
Earth and Mars are both rocky planets with atmospheres. Earth supports life. Therefore, Mars might support life.
Use one of the cases above, identify the evidence base, and judge how strong the conclusion is once you account for rival factors.
Evaluation Practice
InductiveDeep Practice: Bayesian Comparison of Rival Hypotheses
Evaluate the inductive strength of each argument. Consider sample size, representativeness, and alternative explanations.
Complex inductive arguments
Rate each argument's strength on a scale of 1-5 and justify your rating with specific criteria.
Argument 1
A pharmaceutical company tested its new pain reliever on 200 adults aged 18-65 and found 78% reported reduced pain. They conclude the drug is effective for all adults.
Argument 2
Over 30 years of weather data from 50 stations show that average temperatures in the region have increased by 1.5 degrees Celsius. Scientists project this trend will continue.
Argument 3
A survey of 5,000 randomly selected voters across all states found 52% favor the policy. The margin of error is 1.4%. Political analysts predict the referendum will pass.
Argument 4
Every iPhone model released in the past 10 years has been more expensive than the last. Therefore, the next iPhone will be even more expensive.
Use one of the cases above, identify the evidence base, and judge how strong the conclusion is once you account for rival factors.
Evaluation Practice
InductiveReal-World Transfer: Bayesian Comparison of Rival Hypotheses
Evaluate real-world inductive arguments from media, science, and daily life. Apply the criteria you have learned to assess their strength.
Induction in practice
Evaluate each real-world argument. Identify the type of induction and assess its strength.
News claim
A news article reports: 'Based on polling data from 1,200 likely voters in swing states, the candidate leads by 3 points with a margin of error of 2.8 points.' How strong is the inductive basis for predicting the election outcome?
Consumer reasoning
A product has 4.8 stars from 15,000 reviews on Amazon. A friend says: 'With that many positive reviews, the product must be excellent.' Evaluate this reasoning, considering potential biases in online reviews.
Scientific claim
A nutrition study followed 50,000 people for 20 years and found that those who ate fish twice weekly had 25% fewer heart attacks. The researchers conclude fish consumption reduces heart attack risk. What would strengthen or weaken this conclusion?
Use one of the cases above, identify the evidence base, and judge how strong the conclusion is once you account for rival factors.
Rapid Identification
InductiveTimed Drill: Bayesian Comparison of Rival Hypotheses
Quickly classify each argument's inductive type (enumerative, analogical, statistical, causal) and rate its strength on a 1-5 scale. Speed and accuracy both matter.
Rapid inductive classification
Classify the inductive type and rate the strength (1-5) for each item. Target: under 45 seconds per item.
Item 1
The last 20 volcanic eruptions on this island occurred between March and June. The next eruption will likely occur between March and June.
Item 2
A clinical trial with 8,000 participants found the drug reduced symptoms by 35% compared to placebo, with p < 0.001.
Item 3
My neighbor's golden retriever is friendly. My cousin's golden retriever is friendly. Therefore, the golden retriever I meet at the park will probably be friendly.
Item 4
Every time the factory increased shifts, accident rates went up within two weeks. Adding a third shift will likely increase accidents.
Item 5
In a poll of 150 college students at one university, 73% supported the policy. Therefore, most college students nationwide support it.
Item 6
Countries that invested heavily in renewable energy in the 2010s now have lower energy costs. Investing in renewables lowers energy costs.
Use one of the cases above, identify the evidence base, and judge how strong the conclusion is once you account for rival factors.
Evaluation Practice
InductivePeer Review: Bayesian Comparison of Rival Hypotheses
Below are sample student evaluations of inductive arguments. Assess each student's analysis: Did they correctly identify the argument type? Did they properly evaluate its strength? What did they miss?
Evaluate student analyses
Each student evaluated an inductive argument. Assess their work and identify what they got right and wrong.
Student A's analysis
Original argument: 'A survey of 200 Twitter users found 80% support the policy.' Student A wrote: 'This is a strong statistical argument because the sample size of 200 is large enough for reliable results.'
Student B's analysis
Original argument: 'The sun has risen every day for billions of years, so it will rise tomorrow.' Student B wrote: 'This is a weak inductive argument because past observations cannot guarantee future events. The sample is biased toward observed sunrises.'
Student C's analysis
Original argument: 'Rats given the chemical developed tumors. Therefore, the chemical likely causes cancer in humans.' Student C wrote: 'This is a strong analogical argument. Rats and humans share 85% of their genes, so results should transfer directly.'
Student D's analysis
Original argument: 'Five out of five mechanics I consulted said the transmission needs replacing.' Student D wrote: 'Strong inductive argument. Five independent experts agree, and mechanics have domain expertise. The only weakness is the small number of mechanics consulted.'
Use one of the cases above, identify the evidence base, and judge how strong the conclusion is once you account for rival factors.
Argument Building
InductiveConstruction Challenge: Bayesian Comparison of Rival Hypotheses
Build strong inductive arguments from scratch. You are given a conclusion to support. Construct the best evidence, explain your sampling, and address potential weaknesses.
Build inductive arguments
For each conclusion, construct the strongest possible inductive support. Specify your evidence and methodology.
Task 1
Build an inductive argument supporting: 'Bilingual children develop stronger executive function skills.' Describe what study you would design, your sample, and why your evidence would be convincing.
Task 2
Construct an analogical argument that compares managing a sports team to managing a software development team. Make the analogy as strong as possible by identifying at least four relevant similarities.
Task 3
Build a causal inductive argument supporting: 'Reducing class sizes improves student performance.' Specify what data you would need and how you would rule out confounding variables.
Task 4
Create a strong statistical argument about voter turnout among young adults. Describe your sampling method, sample size, and why your approach avoids common biases.
Use one of the cases above, identify the evidence base, and judge how strong the conclusion is once you account for rival factors.
Diagnosis Practice
InductiveCounterexample Challenge: Bayesian Comparison of Rival Hypotheses
For each inductive generalization, find or construct a counterexample that weakens the argument. Explain how your counterexample undermines the conclusion and what it reveals about the argument's limits.
Counterexamples to inductive generalizations
Each generalization seems reasonable. Find cases that challenge or refute it.
Generalization 1
Every tech startup that received Series A funding has gone on to achieve profitability. Therefore, receiving Series A funding leads to profitability.
Generalization 2
In every observed case, countries with higher education spending have higher GDP per capita. Therefore, increasing education spending will raise GDP per capita.
Generalization 3
All mammals observed so far give live birth. Therefore, all mammals give live birth.
Generalization 4
Every patient in the trial who received the drug recovered within a week. Therefore, the drug is an effective treatment.
Use one of the cases above, identify the evidence base, and judge how strong the conclusion is once you account for rival factors.
Analysis Practice
InductiveIntegration Exercise: Bayesian Comparison of Rival Hypotheses
These exercises combine inductive reasoning with deductive logic, explanation assessment, or problem-solving. Apply multiple reasoning tools to reach well-supported conclusions.
Cross-topic inductive exercises
Each scenario requires inductive reasoning plus at least one other reasoning type.
Scenario 1
A study of 10,000 workers found that those who take regular breaks are 23% more productive. A company policy states: 'If a practice is shown to increase productivity by more than 15%, it shall be adopted.' Evaluate the inductive strength of the study, then apply the deductive rule to determine what the policy requires.
Scenario 2
Historical data shows that all five previous product launches in Q4 outperformed Q1 launches. Marketing proposes launching the next product in Q4. However, the market conditions have changed significantly due to new competitors. Evaluate the inductive argument and explain (abductively) why past patterns might not hold.
Scenario 3
A nutrition study found that people who eat breakfast perform better on cognitive tests. A school is considering a mandatory breakfast program. Evaluate the causal inference, identify confounders, and design a problem-solving approach to determine whether the program would work.
Use one of the cases above, identify the evidence base, and judge how strong the conclusion is once you account for rival factors.
Diagnosis Practice
InductiveMisconception Clinic: Bayesian Comparison of Rival Hypotheses
Each item presents a common misconception about inductive reasoning or statistics. Identify the error, explain why it is wrong, and describe how the reasoning should actually work.
Common inductive misconceptions
Diagnose and correct each misconception about inductive reasoning.
Misconception 1
A student says: 'A larger sample size always makes an inductive argument stronger, regardless of how the sample was collected.'
Misconception 2
A student claims: 'Correlation proves causation as long as the correlation is strong enough. A 0.95 correlation coefficient means X definitely causes Y.'
Misconception 3
A student writes: 'An inductive argument with true premises and a true conclusion is a strong argument.'
Misconception 4
A student argues: 'Since inductive arguments can never be certain, they are all equally unreliable. You might as well flip a coin.'
Misconception 5
A student says: 'A single counterexample completely destroys an inductive generalization, just as it destroys a deductive argument.'
Use one of the cases above, identify the evidence base, and judge how strong the conclusion is once you account for rival factors.
Argument Building
InductiveScaffolded Argument: Bayesian Comparison of Rival Hypotheses
Build inductive arguments in stages. Each task provides some evidence and a partial analysis. Complete the analysis, identify gaps, and strengthen the argument step by step.
Step-by-step argument strengthening
Complete each partial analysis and improve the argument at each stage.
Scaffold 1
Claim: Mediterranean diets reduce heart disease risk. Stage 1: You have observational data from 5 countries. Describe what this evidence establishes. Stage 2: You add a randomized trial with 7,000 participants. How does this change the argument? Stage 3: A meta-analysis combines 15 studies. What does the full evidence base now support?
Scaffold 2
Claim: Later school start times improve teen academic performance. Stage 1: One school district changed start times and saw GPA increase by 0.2 points. Evaluate this evidence alone. Stage 2: Three more districts replicated the result. How does this change your assessment? Stage 3: A nationwide study with controls for socioeconomic factors confirms the pattern. What is the argument strength now?
Scaffold 3
Claim: Urban green spaces reduce crime rates. Stage 1: You have a correlation between park density and lower crime in 10 cities. What can and cannot be concluded? Stage 2: A natural experiment -- a city builds parks in high-crime areas and crime drops. How much stronger is the argument? Stage 3: Multiple cities replicate with randomized neighborhood selection. Evaluate the full argument.
Use one of the cases above, identify the evidence base, and judge how strong the conclusion is once you account for rival factors.
Evaluation Practice
InductiveSynthesis Review: Bayesian Comparison of Rival Hypotheses
These exercises combine all aspects of inductive reasoning: sampling, generalization, analogy, causal reasoning, and statistical evaluation. Each task requires integrating multiple skills.
Comprehensive inductive review
Apply all your inductive reasoning skills together.
Comprehensive 1
A government report claims: 'Based on a longitudinal study of 25,000 households across 50 cities over 10 years, households that adopted solar panels reduced their energy costs by an average of 40% and increased their property values by 8%.' Evaluate: (a) the sampling methodology, (b) the causal claim about cost reduction, (c) the causal claim about property values, (d) whether an analogical argument from these households to commercial buildings would be strong.
Comprehensive 2
Design a study to test whether flexible work hours improve employee well-being. Specify: (a) your sampling method and why it avoids bias, (b) what you would measure, (c) how you would control for confounders, (d) what conclusion different results would support, and (e) the limits of your study's generalizability.
Use one of the cases above, identify the evidence base, and judge how strong the conclusion is once you account for rival factors.
Animated Explainers
Step-by-step visual walkthroughs of key concepts. Click to start.
Read the explanation carefully before jumping to activities!
Further Support
Open these only if you need extra help or context
Mistakes to avoid before submitting
- Treating Bayesian comparison as if it required certainty rather than differential support.
- Ignoring the prior because the evidence feels overwhelming.
Where students usually go wrong
Choosing the most vivid hypothesis rather than the best probabilistic update.
Ignoring the alternative hypothesis when judging evidential force.
Treating a likelihood as a Bayes factor without computing the ratio.
Historical context for this way of reasoning
Harold Jeffreys
Jeffreys developed interpretive categories for Bayes factors (barely worth mentioning, substantial, strong, decisive) that still influence how scientists talk about evidential weight in modern Bayesian analysis.