Create a summary regression table similar to those produced for `lm`

```
# S3 method for estimateEffect
summary(object, topics = NULL, nsim = 500, ...)
```

object

an object of class `"estimateEffect"`

, usually a result of a call to
`estimateEffect`

topics

a vector containing the topic numbers for each a summary is to be calculated.
Must be contained in the original `estimateEffect`

object

nsim

the number of simulations to use per parameter set to calculate the standard error. Defaults to 500

...

further arguments passed to or from other methods

This function along with `print.summary.estimateEffect`

creates
regression tables that look like typically summaries you see in R. In general
we recommend that you use non-linearities such as splines via function like
`s`

and in those circumstances the tables are not particularly
interpretable.

Confidence intervals are calculated by using draws from the covariance matrix of each simulation to estimate the standard error. Then a t-distribution approximation is applied to calculate the various quantities of interest.