type I and II error
Posted by: happyking02 (IP Logged)
Date: May 16, 2010 11:55PM
Suppose that all of a firm’s managers are outperforming the benchmark, some by a little, some by a lot. If the confidence intervals for a quality control charts in portfolio management were widened, what would the most likely effect be?
A) Type I error would become more likely and Type II error would become more likely.
B) Type I error would become less likely and Type II error would become more likely.
C) Type I error would become more likely and Type II error would become less likely.
Can u please explain? thanks!
Question 2:
type I or II
Posted by: cfasf1 (IP Logged)
Date: May 5, 2009 12:09AM
Suppose that a portfolio management firm has decided that the costs of hiring and firing managers are excessive. Which of the following would be their most appropriate course of action? The firm should:
A) tolerate more Type I error to reduce Type II error.
B) reduce both Type I and Type II errors.
C) tolerate more Type II error to reduce Type I error.
Edited 1 time(s). Last edit at Sunday, April 17, 2011 at 01:45AM by deriv108.作者: zwjy 时间: 2011-7-11 19:06
B and C
You are not rejecting those people who are actually not performing well. Type 2 error.作者: mar350 时间: 2011-7-11 19:06
B and B
NO EXCUSES作者: bodhisattva 时间: 2011-7-11 19:06
B - If the band gets wider, it's harder to reject null so Type 2 error is more likely
A - A firm would want to tolerate some managers that provide zero (value) Type 1 error but avoid firing managers that provide positive value (Type 2 error)作者: jmh530 时间: 2011-7-11 19:06
AFers let us agree on one thing straight off the bat,
1A
2B
can be thrown out immediately. a reduction in one error type leads to an increase in the other ........................great so now we at 50% ............................a type 1 error is what happens at most firms, firms err on the side of caution and keep the bad with the good rather than losing a good manager (hiring costs and beauracracy also plays a part).................a type 2 error is when u fire anyone and everyone that shows any signs of weakness ( happened at enron i heard) so u risk throwing out good managers but get rid of most if not all bad managers.......................
so my final answer is B and A
Thanks again for posting this作者: strikethree 时间: 2011-7-11 19:06
deriv108,
Are B and A the correct answers ?作者: Windjam 时间: 2011-7-11 19:06
Yes. They can be found in AF. 作者: IAmNeil 时间: 2011-7-11 19:06
so more chance of Type I errors...(more false positives).
CP作者: skycfa 时间: 2011-7-11 19:06
Another point to memorize:
The manager could be fired if his portfolio's returns are inside [or below] of the Quality Control Chart.
Can anyone tell why the QC chart is narrowing over time? I don't understand Schweser's explanation.作者: bkballa 时间: 2011-7-11 19:06
> Can anyone tell why the QC chart is narrowing over
> time? I don't understand Schweser's explanation.
Essentially mean-reversion. For a manager to show significant alpha, year after year, and never underperform (such as Madoff reported to have done) is statistically aberrant and should draw the attention of the SEC.作者: Analti_Calte 时间: 2011-7-11 19:06
thanks, that helps.作者: ohai 时间: 2011-7-11 19:06
alta168 Wrote:
-------------------------------------------------------
> Sorry, I am confused. Refering to Question 1
> mentioned above & Exhibit 5 on P.176 of CFAI
> text,Vol6 :
>
> What is meant by "the confidence intervals for a
> quality control charts were widened" here ? Does
> it mean : the confidence interval is increased
> (e.g., from 80% to 90%) ? And in this case, will
> the funnel-shaped lines (upper/lower envelopes) of
> quality control charts be more far away from
> X-axis (horizontal line) ?
I think "the confidence intervals for a quality control charts were widened" shall mean that the confidence intervals : e.g., changed from 90% to 80%, and the funnel-shaped lines of quality control charts be more far away from X-axis (horizontal line).
Anyone else can confirm ?作者: thommo77 时间: 2011-7-11 19:06
> I think "the confidence intervals for a quality
> control charts were widened" shall mean that the
> confidence intervals : e.g., changed from 90% to
> 80%, and the funnel-shaped lines of quality
> control charts be more far away from X-axis
> (horizontal line).
>
> Anyone else can confirm ?