A Bayesian analyst computes a 95% credible interval for a parameter θ. What does this interval correctly claim?
AIf we repeated the experiment many times, 95% of the computed intervals would contain the true θ
BGiven the observed data and the prior, there is a 95% probability that θ lies within this interval
Cθ is guaranteed to lie in this interval with 95% certainty regardless of the prior used
DThe interval contains 95% of the observed data values
Option B is the correct Bayesian interpretation: the credible interval is a statement about the posterior distribution — given this data and prior, P(θ ∈ interval | data) = 0.95. Option A is the frequentist confidence interval interpretation, which is subtly different and is often what people mistakenly attribute to both. The frequentist interval makes no probability claim about the specific interval computed; it describes a procedure that works 95% of the time in repeated sampling. The Bayesian interval directly answers the question practitioners typically want answered.
Question 2 Multiple Choice
Two analysts apply Bayesian inference to the same dataset of 1,000 observations where evidence strongly suggests θ ≈ 0.8. Analyst A uses an informative prior strongly concentrated near θ = 0. Analyst B uses a flat (uniform) prior. What happens to their posteriors?
AThey remain dramatically different because the prior always determines the posterior
BThey converge to approximately the same posterior because the large likelihood from 1,000 data points overwhelms both priors
CAnalyst A's posterior peaks near 0, Analyst B's peaks near 0.8
DOnly Analyst B's analysis is valid; informative priors are not permitted in Bayesian inference
With 1,000 data points, the likelihood P(data|θ) is extremely concentrated and dominates both priors. Even Analyst A's prior, which favors θ = 0, is overwhelmed by the cumulative evidence from so many observations. Both posteriors will peak near 0.8. This is a key feature of Bayesian inference: when data is plentiful, the prior matters little and analysts with different priors reach similar conclusions. The prior matters most — and sensitivity analysis becomes critical — when data is sparse.
Question 3 True / False
A frequentist 95% confidence interval and a Bayesian 95% credible interval answer the same question about parameter uncertainty, just using different calculation methods.
TTrue
FFalse
Answer: False
They answer different questions. A 95% confidence interval describes a procedure: if you repeated the experiment and computed intervals each time, 95% of them would contain the true parameter. It makes no probability claim about the specific interval in hand — the true θ either is or is not in it. A 95% credible interval makes a direct posterior probability claim: given the observed data and prior, there is 95% probability that θ lies in the interval. This is the statement practitioners usually want, but it requires a prior and cannot be made under frequentist assumptions.
Question 4 True / False
In Bayesian inference, treating a parameter as a random variable allows you to make direct probability statements about it, such as 'there is a 72% probability that θ exceeds 0.5 given the data.'
TTrue
FFalse
Answer: True
This is one of the key practical advantages of the Bayesian framework. Because the posterior P(θ|data) is a full probability distribution over θ, you can compute the probability that θ falls in any region by integrating the posterior over that region. Under frequentist assumptions, θ is a fixed (if unknown) constant, so saying 'there is 72% probability that θ > 0.5' is not a valid statement — θ is either greater than 0.5 or it isn't. Bayesian inference enables exactly this kind of probabilistic statement, which is often what decision-makers need.
Question 5 Short Answer
Why is the Bayesian approach philosophically distinct from the frequentist approach in how it treats unknown parameters, and what is the key practical consequence of this difference?
Think about your answer, then reveal below.
Model answer: Bayesian inference treats parameters as random variables with probability distributions, reflecting uncertainty. Frequentist inference treats parameters as fixed but unknown constants. The key practical consequence is interpretability: Bayesian results in credible intervals that directly state the probability that a parameter lies in a range given the data, while frequentist confidence intervals describe a repeated-sampling procedure that does not make probability claims about specific intervals. Bayesian inference also allows prior knowledge to be incorporated formally.
The philosophical divide traces to the question of what 'probability' means. Frequentists define probability as long-run frequency — a property of repeated experiments. Bayesians define it as degree of belief — a property of an agent's state of knowledge. This difference has real consequences: frequentist methods cannot assign probabilities to one-time events or fixed parameters, while Bayesian methods can but must specify a prior. Neither framework is universally superior; the choice depends on the problem, available prior knowledge, and what question you want to answer.