Adjusting standard errors by sqrt c-hat

questions concerning analysis/theory using program PRESENCE

Adjusting standard errors by sqrt c-hat

Postby vegemite » Fri Feb 21, 2020 1:36 pm

Hi all,

My most global model outputs a c-hat value of 1.62. As such, I have altered c-hat via the Tools menu. I believe that I also need to inflate my standard errors by the square root of the c-hat.

Is there a way for Presence inflate SEs automatically, or do I need to go through each model and calculate the adjusted SEs manually?

I ask this because the SEs do not presently change when I alter c-hat, but it is implied on p92 of Donovan and Hines (2007) Exercise 4 that they should:

If c-hat was greater than 1, you want to adjust c-hat because many analysts feel that any c-hat estimate greater than 1 potentially biases the standard error estimates. To do this, return to the main menu and go to Tools | Change c-hat.
Then enter the new c-hat value, and click OK. You should see that the standard errors for the different parameter estimates have increased as a result.


Thank you,

Abi
vegemite
 
Posts: 12
Joined: Fri Jan 31, 2020 10:57 am
Location: UK

Re: Adjusting standard errors by sqrt c-hat

Postby jhines » Fri Feb 21, 2020 3:01 pm

At the moment, Presence only uses c-hat in the model summary AIC table, not for the standard errors in each model. So, it will need to be done manually. I suggest cut/paste into a spreadsheet so the revised std. errors can be computed via formula, or use RPresence to re-run models and compute new std. errors.
jhines
 
Posts: 599
Joined: Fri May 16, 2003 9:24 am
Location: Laurel, MD, USA

Re: Adjusting standard errors by sqrt c-hat

Postby vegemite » Fri Feb 21, 2020 3:16 pm

Great, thank you for the speedy response.
vegemite
 
Posts: 12
Joined: Fri Jan 31, 2020 10:57 am
Location: UK


Return to analysis help

Who is online

Users browsing this forum: No registered users and 9 guests