HIGHLIGHTS
- who: from the (UNIVERSITY) have published the article: Online supplemental material, in the Journal: (JOURNAL)
SUMMARY
Three main patterns emerge from these results. Unsurprisingly, less favorable power results (e_g for T=25, 50) obtain if initial values are drawn from distributions characterized by relatively high variation (e_g discarding 50 presample values; setting κ=1 in the design of Madsen ). Third, while power outcomes are similar in relative terms for homoskedastic and variance break models, actual power estimates are sensitive to both location and direction of variance breaks. Unlike for the benchmark outcomes in Table . . .
If you want to have access to all the content you need to log in!
Thanks :)
If you don't have an account, you can create one here.