Bayesian One-Sample T-Test (Stan)

Quentin F. Gronau

2017-10-20

In this vignette, we exlpain how we can compute the (log) marginal likelihood and the Bayes factor for models fitted in Stan. This approach has the advantage that the user only needs to pass the fitted stanfit object which contains all information that is necessary to compute the (log) marginal likelihood. Here we show how one can conduct a Bayesian one-sample t-test as implemented in the BayesFactor package (Morey & Rouder, 2015).

Model

The Bayesian one-sample t-test makes the assumption that the observations are normally distributed with mean \(\mu\) and variance \(\sigma^2\). The model is then reparametrized in terms of the standardized effect size \(\delta = \mu/\sigma\). For the standardized effect size, a Cauchy prior with location zero and scale \(r = 1/\sqrt{2}\) is used. For the variance \(\sigma^2\), Jeffreys’s prior is used: \(p(\sigma^2) \propto 1/\sigma^2\).

In this example, we are interested in comparing the null model \(\mathcal{H}_0\), which posits that the effect size \(\delta\) is zero, to the alternative hypothesis \(\mathcal{H}_1\), which assigns \(\delta\) the above described Cauchy prior.

Data

In this example, we will analyze the sleep data set from the t.test example. This data set shows the effect of two soporific drugs (increase in hours of sleep compared to control) on 10 patients. These data can be analyzed via a one-sample t-test by first computing the difference scores and then conducting the t-test using these difference scores as data. The difference scores are calculated as follows:

library(bridgesampling)

# Sleep data from t.test example
data(sleep)

# compute difference scores
y <- sleep$extra[sleep$group == 2] - sleep$extra[sleep$group == 1]
n <- length(y)

Specifying the Models

Next, we implement the models in Stan. Note that to compute the (log) marginal likelihood for a Stan model, we need to specify the model in a certain way. Instad of using "~" signs for specifying distributions, we need to directly use the (log) density functions. The reason for this is that when using the "~" sign, constant terms are dropped which are not needed for sampling from the posterior. However, for computing the marginal likelihood, these constants need to be retained. For instance, instead of writing y ~ normal(mu, sigma) we would need to write target += normal_lpdf(y | mu, sigma). The models can then be specified and compiled as follows (note that it is necessary to install rstan for this):

library(rstan)

# models
stancodeH0 <- '
data {
  int<lower=1> n; // number of observations
  vector[n] y; // observations
}
parameters {
  real<lower=0> sigma2; // variance parameter
}
model {
  target += log(1/sigma2); // Jeffreys prior on sigma2
  target += normal_lpdf(y | 0, sqrt(sigma2)); // likelihood
}
'
stancodeH1 <- '
data {
  int<lower=1> n; // number of observations
  vector[n] y; // observations
  real<lower=0> r; // Cauchy prior scale
}
parameters {
  real delta;
  real<lower=0> sigma2;// variance parameter
}
model {
  target += cauchy_lpdf(delta | 0, r); // Cauchy prior on delta
  target += log(1/sigma2); // Jeffreys prior on sigma2
  target += normal_lpdf(y | delta*sqrt(sigma2), sqrt(sigma2));  // likelihood
}
'
# compile models
stanmodelH0 <- stan_model(model_code = stancodeH0, model_name="stanmodel")
stanmodelH1 <- stan_model(model_code = stancodeH1, model_name="stanmodel")

Fitting the Models

Now we can fit the null and the alternative model in Stan. One usually requires a larger number of posterior samples for estimating the marginal likelihood than for simply estimating the model parameters. This is the reason for using a comparatively large number of samples for these simple models.

# fit models
stanfitH0 <- sampling(stanmodelH0, data = list(y = y, n = n),
                      iter = 20000, warmup = 1000, chains = 4, cores = 1,
                         control = list(adapt_delta = .99))
stanfitH1 <- sampling(stanmodelH1, data = list(y = y, n = n, r = 1/sqrt(2)),
                      iter = 20000, warmup = 1000, chains = 4, cores = 1,
                         control = list(adapt_delta = .99))

Computing the (Log) Marginal Likelihoods

Computing the (log) marginal likelihoods via the bridge_sampler function is now easy: we only need to pass the stanfit objects which contains all information necessary. We use silent = TRUE to suppress printing the number of iterations to the console:

set.seed(12345)
H0 <- bridge_sampler(stanfitH0, silent = TRUE)
print(H0)
## Bridge sampling estimate of the log marginal likelihood: -20.81018
## Estimate obtained in 4 iteration(s) via method "normal".
H1 <- bridge_sampler(stanfitH1, silent = TRUE)
print(H1)
## Bridge sampling estimate of the log marginal likelihood: -17.96119
## Estimate obtained in 4 iteration(s) via method "normal".

We can use the error_measures function to compute an approximate percentage error of the estimates:

# compute percentage errors
print(error_measures(H0)$percentage)
## [1] "0.0485%"
print(error_measures(H1)$percentage)
## [1] "0.0657%"

Computing the Bayes Factor

To compare the null model and the alternative model, we can compute the Bayes factor by using the bf function. In our case, we compute \(\text{BF}_{10}\), that is, the Bayes factor which quantifies how much more likely the data are under the alternative versus the null hypothesis:

# compute Bayes factor
BF10 <- bf(H1, H0)
print(BF10)
## The estimated Bayes factor in favor of x1 over x2 is equal to: 17.27033

We can compare the bridge sampling result to the BayesFactor package result:

library(BayesFactor)
print(ttestBF(y))
## Bayes factor analysis
## --------------
## [1] Alt., r=0.707 : 17.25888 ±0%
## 
## Against denominator:
##   Null, mu = 0 
## ---
## Bayes factor type: BFoneSample, JZS

One-sided Test

We can also conduct one-sided tests. For instance, we could test the hypothesis that the effect size is positive versus the null hypothesis. Since we already fitted the null model and computed its marginal likelihood, we only need to slightly adjust the alternative model to reflect the directed hypothesis. To achieve this, we need to truncate the Cauchy prior distribution for \(\delta\) at zero and then renormalize the (log) density. This is easily achieved via the Stan function cauchy_lccdf which corresponds to the log of the complementary cumulative distribution function of the Cauchy distribution. Thus, cauchy_lccdf(0 | 0, r) gives us the log of the area greater than zero which is required for renormalizing the truncated Cauchy prior. The model can then be specified and fitted as follows:

stancodeHplus <- '
data {
  int<lower=1> n; // number of observations
  vector[n] y; // observations
  real<lower=0> r; // Cauchy prior scale
}
parameters {
  real<lower=0> delta; // constrained to be positive
  real<lower=0> sigma2;// variance parameter
}
model {
  target += cauchy_lpdf(delta | 0, r) - cauchy_lccdf(0 | 0, r); // Cauchy prior on delta
  target += log(1/sigma2); // Jeffreys prior on sigma2
  target += normal_lpdf(y | delta*sqrt(sigma2), sqrt(sigma2));  // likelihood
}
'
# compile models
stanmodelHplus <- stan_model(model_code = stancodeHplus, model_name="stanmodel")
stanfitHplus <- sampling(stanmodelHplus, data = list(y = y, n = n, r = 1/sqrt(2)),
                         iter = 30000, warmup = 1000, chains = 4,
                         control = list(adapt_delta = .99))

The (log) marginal likelihood is then obtained as follows:

Hplus <- bridge_sampler(stanfitHplus, silent = TRUE)
print(Hplus)
## Bridge sampling estimate of the log marginal likelihood: -17.27045
## Estimate obtained in 4 iteration(s) via method "normal".
print(error_measures(Hplus)$percentage)
## [1] "0.136%"

The one-sided Bayes factor in favor of a positive effect versus the null hypothesis can be computed as follows:

# compute Bayes factor
BFplus0 <- bf(Hplus, H0)
print(BFplus0)
## The estimated Bayes factor in favor of x1 over x2 is equal to: 34.45738

We can compare the bridge sampling result to the BayesFactor package result:

print(ttestBF(y, nullInterval = c(0, Inf)))
## Bayes factor analysis
## --------------
## [1] Alt., r=0.707 0<d<Inf    : 34.41694  ±0%
## [2] Alt., r=0.707 !(0<d<Inf) : 0.1008246 ±0.06%
## 
## Against denominator:
##   Null, mu = 0 
## ---
## Bayes factor type: BFoneSample, JZS

References

Richard D. Morey and Jeffrey N. Rouder (2015). BayesFactor: Computation of Bayes Factors for Common Designs. R package version 0.9.12-2.