Actuarial Outpost
 
Go Back   Actuarial Outpost > Exams - Please Limit Discussion to Exam-Related Topics > SoA/CAS Preliminary Exams > VEE > Statistics
FlashChat Actuarial Discussion Preliminary Exams CAS/SOA Exams Cyberchat Around the World Suggestions

Meet the Employees of DW Simpson
Patty Jacobsen Simpson, Bob Morand, Kristyn Sakelaris, Sean Loboda, KC Cho, Maureen Matous, Ellen Page
Aaron Benton, Becki Tobia, Kimberly Skora, Margit Vogele, Barclay Burns, Jason Blundy, Dan Karrow, Tom Troceen
Valorie Mulder, Marianne Westphal, Carol Lee, Jennifer Retford, Kieran Welsh-Phillips, Lindsey Nelson, Emily Paxton
Angie Wachholz, Derek Mulder, Julie Garwood, Caitlin Cunningham, David Benton, Dave Retford, Sarah Cleveland, Rhonda Glick
Genevieve Shannon, Meghan Bautista, Carol Datu, Barb Rave, Jesus Perez, Dan Kane, Chris Zdenek, Scott Simon, Kriss Wells


Reply
 
Thread Tools Display Modes
  #1  
Old 01-25-2006, 09:56 AM
Calliou Calliou is offline
Member
 
Join Date: Feb 2004
Posts: 92
Default p-value and level of significance

Can someone please explain to me how they are related and how they are different? All i understood was level of significance = P( reject H0 / H0 is true)= P(type 1 error)

and p-value is the likelihood of type 1 error.
But i am still not sure i understood p-value clearly.
How do we calculate p-value?
Reply With Quote
  #2  
Old 01-25-2006, 11:24 AM
NewTubaBoy's Avatar
NewTubaBoy NewTubaBoy is offline
Member
CAS
 
Join Date: Jul 2004
Location: Boston
Studying for CAS 8
Favorite beer: Blue Moon
Posts: 1,391
Default

They're the same thing.
Reply With Quote
  #3  
Old 01-25-2006, 11:18 PM
Howard Mahler Howard Mahler is offline
Member
CAS
 
Join Date: May 2002
Location: Boston, MA
Posts: 1,660
Default

To a first approximation they are the same.

However, here is a typical example.
One would choose to test a hypotheis at for example the 10% significance level. (In practical applications and some exam questions, the significance level is chosen in advance, prior to performing the test or seeing the data.)

If the p-value is greater than 10%, for example 14%, then we would not reject H0 at the 10% significance level.
If the p-value is less than 10%, for example 8%, then we would reject H0 at the 10% significance level.

I hope this helps a little.

Howard Mahler
Reply With Quote
  #4  
Old 01-26-2006, 12:42 AM
2cek's Avatar
2cek 2cek is offline
 
Join Date: May 2005
Posts: 16
Default p-value vs level of significance

I like to think of the significance level as a benchmark as to what the p-value must be in order to accept/reject the hypothesis test. However, the significance level also plays a part in constructing confidence intervals and/or prediction intervals. When constructing a confidence interval you again are using a significance level or benchmark which determines how wide the interval will be and with how much confidence you are sure the parameter estimate is within the interval. Here is a thought for confidence intervals: The higher the significance level the wider the confidence interval is. i.e. If you are 99% sure the parameter estimate falls within the confidence interval it had better be quite wide. Where as if you use a 70% interval you are not that sure and the interval will be smaller.

When I think of p-value, I think actual probability. In a hypothesis test you are testing an observed sample and calculating a probability of Ho being true. Of course once you have this actual probability you will compare it to the benchmark(significance level) to see whether the hypothesis test is accepted or rejected.
Reply With Quote
  #5  
Old 01-26-2006, 08:51 AM
Calliou Calliou is offline
Member
 
Join Date: Feb 2004
Posts: 92
Default

Thanks.I think i now get it
Reply With Quote
  #6  
Old 01-26-2006, 11:03 AM
NewTubaBoy's Avatar
NewTubaBoy NewTubaBoy is offline
Member
CAS
 
Join Date: Jul 2004
Location: Boston
Studying for CAS 8
Favorite beer: Blue Moon
Posts: 1,391
Default

Quote:
Originally Posted by 2cek
I like to think of the significance level as a benchmark as to what the p-value must be in order to accept/reject the hypothesis test. However, the significance level also plays a part in constructing confidence intervals and/or prediction intervals. When constructing a confidence interval you again are using a significance level or benchmark which determines how wide the interval will be and with how much confidence you are sure the parameter estimate is within the interval. Here is a thought for confidence intervals: The higher the significance level the wider the confidence interval is. i.e. If you are 99% sure the parameter estimate falls within the confidence interval it had better be quite wide. Where as if you use a 70% interval you are not that sure and the interval will be smaller.

When I think of p-value, I think actual probability. In a hypothesis test you are testing an observed sample and calculating a probability of Ho being true. Of course once you have this actual probability you will compare it to the benchmark(significance level) to see whether the hypothesis test is accepted or rejected.
I think this is a good explaination... just dont get confused... becuase they are the exact same number.
Reply With Quote
  #7  
Old 01-26-2006, 11:17 AM
Bühlmann's Avatar
Bühlmann Bühlmann is offline
Member
SOA AAA
 
Join Date: Nov 2005
Studying for nothing ever again
Favorite beer: Corona Light
Posts: 1,844
Default

yes it is. thanks for posting that.
Reply With Quote
  #8  
Old 01-26-2006, 12:21 PM
2cek's Avatar
2cek 2cek is offline
 
Join Date: May 2005
Posts: 16
Default The Same?

The probability of accepting Ho when Ho is true and the likelihood of the sample being of the same form as Ho are not the same thing.
They are quite different. Depending on the profession in which the statistical analysis is being performed, the significance level will vary. For example, in the medical profession someone may want to use a 1% significance level where as a statistician may use a 10% significance level. The significance level is a matter of how precise you ought to be or how much fluctuation you can tolerate. However, given the same problem, the p-value calculation will be the same in both fields. The significance level is an assumption. The p-value is an actual calculation or probability.
Reply With Quote
  #9  
Old 01-31-2006, 01:08 PM
Frenchie's Avatar
Frenchie Frenchie is offline
Member
SOA
 
Join Date: May 2003
Location: Mississippi
Studying for Exam C
College: Georgia State University - Alumni
Favorite beer: Stella!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Posts: 8,596
Default

Quote:
Originally Posted by Howard Mahler
To a first approximation they are the same.

However, here is a typical example.
One would choose to test a hypotheis at for example the 10% significance level. (In practical applications and some exam questions, the significance level is chosen in advance, prior to performing the test or seeing the data.)

If the p-value is greater than 10%, for example 14%, then we would not reject H0 at the 10% significance level.
If the p-value is less than 10%, for example 8%, then we would reject H0 at the 10% significance level.

I hope this helps a little.

Howard Mahler
I am having trouble following some examples of obtaining the p-values, specifically for an F-statistic. I'm just not seeing it. Can you perhaps come up with an example and show me how to get the p-value using the tables?
__________________
"Never underestimate the strength of a woman. Never f@#k with one who runs 26.2 miles for fun."
Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off


All times are GMT -4. The time now is 04:44 AM.


Powered by vBulletin®
Copyright ©2000 - 2014, Jelsoft Enterprises Ltd.
*PLEASE NOTE: Posts are not checked for accuracy, and do not
represent the views of the Actuarial Outpost or its sponsors.
Page generated in 0.24466 seconds with 7 queries