question on mode_check

[ilcosmin]

Hi,

I have the following question on the use of "mode_check" option in estimation.
When looking at the plots coming out of it, there seems to be 21 values on the x axis.
Q: How are these values chosen?

Presumably there are 10 values to the left and right of the mode of that estimate. So, the 11th value should be the mode. In most cases it is, but in some it is not.

Similarly, the value on the Y axis should be the likelihood evaluated at those points. So, for the 11th value, it should be the likelihood value obtained at the mode. This seems to be true when the middle point in the X axis is the mode.

If on the X axis we don't have the mode but some other points around it then what can we conclude from the mode_check?
I guess we can see in what interval on the X axis the mode actually is, and compare the value of the likelihoods for the interval with the one obtained under the mode.

Thank you,

Cosmin