model {

  mu ~ dunif( -2, 5 )

  for ( i in 1:k ) {

    y[ i ] ~ dnorm( mu, tau.y[ i ] )

  }

  positive.effect <- step( mu )

}