Keep or not keep unrealized tokens on a continuous scale

[ekbrown77]

What reasons are there for keeping unrealized tokens on a continuous scale in a mixed effects linear model? For example, in a study of voiceless fricative duration, what authors or arguments are you aware of for keeping the unrealized tokens of the sound (marked as 0 milliseconds, for example)? I can't seem to find the refs I know I've read on both sides of the discussion about keeping or discarding unrealized tokens of sounds, when sounds are measured on a continuous scale (rather than categorically as "present" or "absent").

This question is theoretical in nature, rather than R-code-specific, but is relevant to fitting linear models, specifically in my case, fitting mixed effects linear models with the lme4 package.

[Matías Guzmán Naranjo]

do you mean as an independent or dependent variable?

2017-04-26 20:07 GMT+02:00 ekbrown77 <ekbr...@gmail.com>:

What reasons are there for keeping unrealized tokens on a continuous scale in a mixed effects linear model? For example, in a study of voiceless fricative duration, what authors or arguments are you aware of for keeping the unrealized tokens of the sound (marked as 0 milliseconds, for example)? I can't seem to find the refs I know I've read on both sides of the discussion about keeping or discarding unrealized tokens of sounds, when sounds are measured on a continuous scale (rather than categorically as "present" or "absent").

This question is theoretical in nature, rather than R-code-specific, but is relevant to fitting linear models, specifically in my case, fitting mixed effects linear models with the lme4 package.

--
You received this message because you are subscribed to the Google Groups "StatForLing with R" group.
To unsubscribe from this group and stop receiving emails from it, send an email to statforling-with-r+unsub...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[ekbrown77]

As a dependent variable.