Merge pull request #827 from stan-dev/issue-826-typo

src/stan-users-guide/measurement-error.qmd

@@ -184,7 +184,7 @@ exp(b))` in numerically more stable way.
 
 Alternatively, the model may be defined with latent parameters for the
 unrounded measurements $z_n$.  The Stan code in this case uses a
-distribution statement for $z_n$ directly while respecting the 
+distribution statement for $z_n$ directly while respecting the
 constraint $z_n \in (y_n - 0.5, y_n + 0.5)$.
 
 ```stan
@@ -270,7 +270,7 @@ $$
 $$
 
 The log odds and standard errors can be defined in a transformed
-parameter block, though care must be taken not to use integer
+data block, though care must be taken not to use integer
 division.^[When dividing two integers, the result type is an integer and rounding will ensue if the result is not exact.  See the discussion of primitive arithmetic types in the reference manual for more information.]
 
 ```stan
@@ -366,4 +366,3 @@ transforming the binomial data.
 If trial-specific predictors are available, these can be included
 directly in a regression model for the per-trial treatment effects
 $\theta_j$.
-