1
$\begingroup$

I am studying the effects of 5 treatments (CT,CCM,CCI,F, S) on grape yield. My experimental design consists in: 5 treatments and 2 farms (SG, MT) which I use as block. I created an additional variable as interaction between farm and treatment which I called plot.

In each farm the treated vine rows are divided in 3 ps.replicates Rep which reflect the altitude (coded as 1,2,3; 1= uphill,2= middle, 3=bottom of the field). I had problem with the soil variability so I carried out an extensive survey and found 2 soil typessoil_cluster in each farm (coded as : 1,2,3,4)

This is the model I have implemented

  m5=glmer(yield~ Farm+Treat+(1|plot/soil_cluster)+(1|Farm:Rep),family=gaussian(link="log"),data=YIELB

However:

  1. I am not sure about structure of the error since farms and treatments have a different number of soil cluster.

  2. I get singular fit.

What do you think? How would you proceed?

EDIT: Thanks for your answer. As requested, I am reporting summary(YIELB) and str(YIELB) below. Soil cluster refers to the cluster of soils obtained through k-mediods clustering. The YIELB dataset also presents other dependent variable such as ncluster which is the no. of bunch of grapes per plant and cluster_weight which is the average bunch weight per plant. 

    > summary(YIELB)
     Farm              Treat           Rep       Plant code      ncluster         
yield        cluster_weight   soil_cluster      plot    
 Length:300         Length:300         1:100   Min.   : 1.0   Min.   : 2.00   
Min.   : 206.0   Min.   : 22.89   1: 40        MT.CCI : 30  
 Class :character   Class :character   2:100   1st Qu.: 3.0   1st Qu.: 5.00   
1st Qu.: 810.8   1st Qu.:142.62   2:110        SG.CCI : 30  
 Mode  :character   Mode  :character   3:100   Median : 5.5   Median : 8.00   
Median :1463.5   Median :186.72   5:110        MT.CCM : 30  
                                           Mean   : 5.5   Mean   : 8.15   
Mean   :1646.7   Mean   :193.23   6: 40        SG.CCM : 30  
                                           3rd Qu.: 8.0   3rd Qu.:11.00   
3rd Qu.:2249.5   3rd Qu.:231.25                MT.CT  : 30  
                                           Max.   :10.0   Max.   :14.00   
Max.   :5285.0   Max.   :468.00                SG.CT  : 30  

(Other):120  

> str(YIELB)
Classes ‘tbl_df’, ‘tbl’ and 'data.frame':   300 obs. of  9 variables:
 $ Farm          : chr  "MT" "MT" "MT" "MT" ...
 $ Treat         : chr  "CCI" "CCI" "CCI" "CCI" ...
 $ Rep           : Factor w/ 3 levels "1","2","3": 1 1 1 1 1 1 1 1 1 1 ...
 $ Plant code    : num  1 2 3 4 5 6 7 8 9 10 ...
 $ ncluster      : num  3 3 9 6 9 5 7 5 7 6 ...
 $ yield         : num  330 379 1230 1182 1416 ...
 $ cluster_weight: num  110 126 137 197 157 ...
 $ soil_cluster  : Factor w/ 4 levels "1","2","5","6": 2 2 2 2 2 2 2 2 2 2 ...
 $ plot          : Factor w/ 10 levels "MT.CCI","SG.CCI",..: 1 1 1 1 1 1 1 1 1 1 ...
Improve this question
$\endgroup$
4
  • $\begingroup$ What is the variable cluster ? Please include the output of summary(yield) and str(yield) $\endgroup$
    – Robert Long
    May 7, 2019 at 8:13
  • $\begingroup$ Please check the edits ;)thank you! $\endgroup$
    – William WIne
    May 8, 2019 at 16:52
  • $\begingroup$ According to your original post the dataset is named yield although now you have changed it to YIELB....that is what I asked to see the summary() and str() for, not the fitted model object. $\endgroup$
    – Robert Long
    May 8, 2019 at 19:29
  • $\begingroup$ Ops sorry. Ok fixed. Thank you! $\endgroup$
    – William WIne
    May 9, 2019 at 8:38

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.