Skip to content

Wrong results from anova

5 messages · Li Li, Phil Spector, William Dunlap +1 more

Li Li
# 
Hi all,
  I have following data. When I perform an anova, the residual sum of
square returns to be zero.
But this is wrong, since we can hand calculate the RSS to be around 0.0308.
  Did anyone come across the same problem before? Any suggestions?
  Thanks.
    Hanna
Y sample.Y
1   0.477      0.5
2   0.477      0.5
3   0.478      0.5
4  27.320       27
5  27.420       27
6  27.300       27
7  29.440       29
8  29.620       29
9  29.610       29
10 35.840       35
11 35.900       35
12 35.850       35
Df Sum Sq Mean Sq F value Pr(>F)
sample.Y     3   2203   734.2  190706 <2e-16 ***
Residuals    8      0     0.0
---
Signif. codes:  0 ?***? 0.001 ?**? 0.01 ?*? 0.05 ?.? 0.1 ? ? 1
Phil Spector
# 
Li Li -
    I belive it's a rounding error -- try setting

options(digits=8)

    before displaying the output:
Df Sum Sq Mean Sq F value Pr(>F) 
sample.Y     3   2203     734  190706 <2e-16 ***
Residuals    8      0       0 
---
Signif. codes:  0 ?***? 0.001 ?**? 0.01 ?*? 0.05 ?.? 0.1 ? ? 1
Df  Sum Sq Mean Sq F value    Pr(>F) 
sample.Y     3 2202.70  734.23  190706 < 2.2e-16 ***
Residuals    8    0.03    0.00 
---
Signif. codes:  0 ?***? 0.001 ?**? 0.01 ?*? 0.05 ?.? 0.1 ? ? 1

 					- Phil Spector
 					 Statistical Computing Facility
 					 Department of Statistics
 					 UC Berkeley
 					 spector at stat.berkeley.edu
On Tue, 14 Apr 2015, li li wrote:

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      William Dunlap
    

    

      
      #
    

  


  

      
It is a printing problem - the default number of digits in print.summary.aov
is max(3L, getOption("digits") - 3L).   Set options(digits=7) instead of
your current 6 (?) or try print(summary(fit.Y), digits=7) to see more
digits.


Bill Dunlap
TIBCO Software
wdunlap tibco.com

        
On Tue, Apr 14, 2015 at 10:23 AM, li li <hannah.hlx at gmail.com> wrote:

        

            

      
      
      
    

        

      
      
    

        

  
  


  


      

    

    

      
      

        


  

        

      Michael Dewey
    

    

      
      #
    

  


  

      
See in-line

        
On 14/04/2015 18:23, li li wrote:

            

      
      
      
    

        
734.2 / 197096
[1] 0.003725088

Did you mean 0.0038 above?

            

      
      
      
    

        

      
      
    

        

  
    
  


  


      

    

    

      
      

        


  

        

      Li Li
    

    

      
      #
    

  


  

      
0.003725088 is the mean square error.





2015-04-14 13:44 GMT-04:00 Michael Dewey <lists at dewey.myzen.co.uk>: