Skip to content

ANOVA in R

4 messages · Thanh Tran, Paul Johnson, Jim Lemon

Thanh Tran
# 
Hi eveyone,
I'm studying about variance (ANOVA) in R and have some questions to share.
I read an article investigating the effect of factors (temperature, Asphalt
content, Air voids, and sample thickness) on the hardness of asphalt
concrete in the tensile test (abbreviated as Kic). Each condition was
repeated four times (4 samples). In the paper, the authors used MINITAB to
analyze Anova. The authors use "adjusted sums of squares" calculate the
p-value I try to use ANOVA in R to analyze this data and get the result as
shown in Figure 4. The results are different from the results in the
article. Some papers say that in R, the default for ANOVA analysis is to
use "sequential sums of squares" to calculate the p-value.
So please help the following two questions: 1 / Introduction to code in R
for anova analysis uses "adjusted sums of squares". The main part of the
command in R / myself is as follows: > Tem = as.factor (temperature) > Ac =
as.factor (AC) > Av = as.factor (AV) > Thick = as.factor (Thickness) >
Twoway = lm (KIC ~ Tem + Ac + Av + Thick + Stamp + Ac + Stamp + Av + Stamp
+ Thick + Ac * Av + Ac * Thick + Av * Thick) > anova (twoway) 2/ When to
use "sequential sums of squares" and when to use "adjusted sums of
squares". Some papers recommend using the "oa.design
<https://www.youtube.com/redirect?q=http%3A%2F%2Foa.design%2F&redir_token=AaSAPDY-5UAsoHxN6BdwfyIJ7R98MTUzOTIxNDg2OUAxNTM5MTI4NDY5&event=comments>"
function in R to check for "orthogonal" designs. If not, use "adjusted sums
of squares". I am still vague about this command, so look forward to
everyone's suggestion. If you could answer all two of my questions, I would
be most grateful. Ps: I have added a CSV file and the paper for practicing
R. http://www.mediafire.com/file/e5oe54p2c2wd4bc/Saha+research.csv
http://www.mediafire.com/file/39jlf9h539y9mdz/Homothetic+behaviour+investigation+on+fracture+toughness+of+asphalt+mixtures+using+semicircular+bending+test.pdf
Paul Johnson
# 
We cannot read your message. Should post pure text, not html. Hm, my phone
now may post html, must try to stop. Your R code not legible. It seems to
be output? Lines all run together. I tried find articles you mention, but
"not found" resulted.

You should use aov() for fitting, then get post hoc comparisons.

In car package, Anova function will help. I may teach Anova soon, we'll see
if I have better answer then.

Paul Johnson
University of Kansas
On Wed, Oct 10, 2018, 1:14 AM Thanh Tran <masternhattt at gmail.com> wrote:

            

      
      
      
    

        

      
      
    

        

  
  


  


      

    

    

      
      

        


  

        

      Jim Lemon
    

    

      
      #
    

  


  

      
Hi Tranh,
I'm not sure why you are converting your variables to factors, and I
think the model you want is:

lm(KIC~temperature+AC+AV+Thickness+temperature:AC+
 temperature:AV+temperature:thickness+AC:AV+
 AC:thickness+AV:thickness,SR)

Note the colons (:) rather than asterisks (*) for the interactions.

Jim

        
On Wed, Oct 10, 2018 at 5:14 PM Thanh Tran <masternhattt at gmail.com> wrote:

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Jim Lemon
    

    

      
      #
    

  


  

      
Hi again,
Two things, I named the data frame SR as shown in the model.
The other is for those who may wish to answer the OP. The mediafire
website is loaded with intrusive ads and perhaps malware.

Jim

        
On Thu, Oct 11, 2018 at 9:02 AM Jim Lemon <drjimlemon at gmail.com> wrote: