Data analysis

Preamble

Load Libraries

library(ezknitr)
library(knitr)
library(metafor)

Clear environment

remove(list=ls())

Document settings

opts_chunk$set(fig.width = 6, fig.height = 4)

Load data

load(file="data/output_data/data_cleaned.R")

Meta-analysis of rust severity

Overall raw mean difference

overall.rust <- rma.uni(yi = yi, vi = vi, data = rust.data.MD, method="ML")
summary(overall.rust)
## 
## Random-Effects Model (k = 324; tau^2 estimator: ML)
## 
##     logLik    deviance         AIC         BIC        AICc  
## -1558.2544    808.4819   3120.5088   3128.0703   3120.5462  
## 
## tau^2 (estimated amount of total heterogeneity): 690.3940 (SE = 69.8629)
## tau (square root of estimated tau^2 value):      26.2753
## I^2 (total heterogeneity / total variability):   78.16%
## H^2 (total variability / sampling variability):  4.58
## 
## Test for Heterogeneity: 
## Q(df = 323) = 1604.9995, p-val < .0001
## 
## Model Results:
## 
## estimate       se     zval     pval    ci.lb    ci.ub          
## -35.7289   1.6571 -21.5611   <.0001 -38.9768 -32.4810      *** 
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Effects of Active Ingredient on SMD

rust.active <- rma.uni(yi = yi, vi = vi, data=rust.data.SMD,
                      method = "REML", mods = ~activeIngClean-1)
summary(rust.active)
## 
## Mixed-Effects Model (k = 324; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -487.7921   975.5842   999.5842  1044.5387  1000.6242  
## 
## tau^2 (estimated amount of residual heterogeneity):     0.4793 (SE = 0.0911)
## tau (square root of estimated tau^2 value):             0.6923
## I^2 (residual heterogeneity / unaccounted variability): 42.31%
## H^2 (unaccounted variability / sampling variability):   1.73
## 
## Test for Residual Heterogeneity: 
## QE(df = 313) = 560.0651, p-val < .0001
## 
## Test of Moderators (coefficient(s) 1,2,3,4,5,6,7,8,9,10,11): 
## QM(df = 11) = 549.8626, p-val < .0001
## 
## Model Results:
## 
##                      estimate      se      zval    pval    ci.lb    ci.ub
## activeIngCleanazo     -1.1482  0.4477   -2.5648  0.0103  -2.0256  -0.2708
## activeIngCleandual    -3.1655  0.2736  -11.5715  <.0001  -3.7017  -2.6293
## activeIngCleanflus    -1.9558  0.4310   -4.5374  <.0001  -2.8006  -1.1110
## activeIngCleanflut    -1.5930  0.1685   -9.4549  <.0001  -1.9232  -1.2627
## activeIngCleangly     -0.4990  0.4053   -1.2310  0.2183  -1.2934   0.2954
## activeIngCleanmixed   -1.1772  0.0936  -12.5771  <.0001  -1.3606  -0.9937
## activeIngCleanoth     -0.8763  0.1588   -5.5190  <.0001  -1.1875  -0.5651
## activeIngCleanpyra    -1.2766  0.2155   -5.9245  <.0001  -1.6989  -0.8543
## activeIngCleantebu    -1.4328  0.1958   -7.3169  <.0001  -1.8166  -1.0490
## activeIngCleantetra   -1.2249  0.3422   -3.5794  0.0003  -1.8957  -0.5542
## activeIngCleanthio    -1.3532  0.4847   -2.7920  0.0052  -2.3031  -0.4033
##                         
## activeIngCleanazo      *
## activeIngCleandual   ***
## activeIngCleanflus   ***
## activeIngCleanflut   ***
## activeIngCleangly       
## activeIngCleanmixed  ***
## activeIngCleanoth    ***
## activeIngCleanpyra   ***
## activeIngCleantebu   ***
## activeIngCleantetra  ***
## activeIngCleanthio    **
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Effects of Class on SMD

rust.class <- rma.uni(yi = yi, vi = vi, data=rust.data.SMD,
                      method = "REML", mods = ~classClean-1)
summary(rust.class)
## 
## Mixed-Effects Model (k = 324; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -517.2490  1034.4979  1046.4979  1069.0891  1046.7672  
## 
## tau^2 (estimated amount of residual heterogeneity):     0.6891 (SE = 0.1075)
## tau (square root of estimated tau^2 value):             0.8301
## I^2 (residual heterogeneity / unaccounted variability): 51.35%
## H^2 (unaccounted variability / sampling variability):   2.06
## 
## Test for Residual Heterogeneity: 
## QE(df = 319) = 657.1024, p-val < .0001
## 
## Test of Moderators (coefficient(s) 1,2,3,4,5): 
## QM(df = 5) = 438.5411, p-val < .0001
## 
## Model Results:
## 
##                        estimate      se      zval    pval    ci.lb
## classCleanherbicide     -0.5089  0.4467   -1.1393  0.2546  -1.3844
## classCleanmma           -1.3931  0.0967  -14.4137  <.0001  -1.5825
## classCleanother         -1.0389  0.1871   -5.5526  <.0001  -1.4056
## classCleanstrobilurin   -1.0313  0.2053   -5.0227  <.0001  -1.4337
## classCleantriazole      -1.5641  0.1188  -13.1693  <.0001  -1.7969
##                          ci.ub     
## classCleanherbicide     0.3666     
## classCleanmma          -1.2037  ***
## classCleanother        -0.6722  ***
## classCleanstrobilurin  -0.6288  ***
## classCleantriazole     -1.3313  ***
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Effects of Growth stage on SMD

rust.growth.stage <- rma.uni(yi = yi, vi = vi, 
                            data=rust.data.SMD[rust.data.SMD$growthStateClean!="unknown",],
                            method = "REML", mods = ~growthStateClean-1)
summary(rust.growth.stage)
## 
## Mixed-Effects Model (k = 306; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -450.8324   901.6647   919.6647   952.9385   920.2897  
## 
## tau^2 (estimated amount of residual heterogeneity):     0.4318 (SE = 0.0890)
## tau (square root of estimated tau^2 value):             0.6571
## I^2 (residual heterogeneity / unaccounted variability): 39.90%
## H^2 (unaccounted variability / sampling variability):   1.66
## 
## Test for Residual Heterogeneity: 
## QE(df = 298) = 507.3941, p-val < .0001
## 
## Test of Moderators (coefficient(s) 1,2,3,4,5,6,7,8): 
## QM(df = 8) = 554.8558, p-val < .0001
## 
## Model Results:
## 
##                     estimate      se      zval    pval    ci.lb    ci.ub
## growthStateClean1    -0.8845  0.3240   -2.7298  0.0063  -1.5195  -0.2494
## growthStateClean1+   -1.6147  0.1527  -10.5715  <.0001  -1.9141  -1.3153
## growthStateClean2    -2.4927  0.4070   -6.1241  <.0001  -3.2905  -1.6949
## growthStateClean2+   -2.1057  0.1263  -16.6721  <.0001  -2.3532  -1.8581
## growthStateClean3    -0.9136  0.0983   -9.2900  <.0001  -1.1063  -0.7209
## growthStateClean4    -1.0194  0.2496   -4.0848  <.0001  -1.5086  -0.5303
## growthStateClean5    -0.8014  0.1991   -4.0250  <.0001  -1.1916  -0.4111
## growthStateCleanV    -0.3088  0.3097   -0.9972  0.3187  -0.9157   0.2981
##                        
## growthStateClean1    **
## growthStateClean1+  ***
## growthStateClean2   ***
## growthStateClean2+  ***
## growthStateClean3   ***
## growthStateClean4   ***
## growthStateClean5   ***
## growthStateCleanV      
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Effects of application number on SMD

table(rust.data.SMD$applicationsNumb)
## 
##   1   2   3   5 
## 133 170  15   6
rust.applications <- rma.uni(yi = yi, vi = vi, 
                            data=rust.data.SMD,
                            method = "REML", mods = ~applicationsNumb)
summary(rust.applications)
## 
## Mixed-Effects Model (k = 324; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -525.4152  1050.8304  1056.8304  1068.1541  1056.9059  
## 
## tau^2 (estimated amount of residual heterogeneity):     0.7201 (SE = 0.1095)
## tau (square root of estimated tau^2 value):             0.8486
## I^2 (residual heterogeneity / unaccounted variability): 52.48%
## H^2 (unaccounted variability / sampling variability):   2.10
## R^2 (amount of heterogeneity accounted for):            0.24%
## 
## Test for Residual Heterogeneity: 
## QE(df = 322) = 677.7382, p-val < .0001
## 
## Test of Moderators (coefficient(s) 2): 
## QM(df = 1) = 2.1017, p-val = 0.1471
## 
## Model Results:
## 
##                   estimate      se     zval    pval    ci.lb    ci.ub     
## intrcpt            -1.1351  0.1622  -6.9990  <.0001  -1.4530  -0.8173  ***
## applicationsNumb   -0.1276  0.0880  -1.4497  0.1471  -0.3000   0.0449     
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
# Categorical
rust.data.SMD$applicationsCat <- as.character(rust.data.SMD$applicationsNumb)
rust.applications.cat <- rma.uni(yi = yi, vi = vi, 
                            data=rust.data.SMD,
                            method = "REML", mods = ~applicationsCat-1)
summary(rust.applications.cat)
## 
## Mixed-Effects Model (k = 324; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -506.4060  1012.8121  1022.8121  1041.6537  1023.0031  
## 
## tau^2 (estimated amount of residual heterogeneity):     0.5800 (SE = 0.0983)
## tau (square root of estimated tau^2 value):             0.7616
## I^2 (residual heterogeneity / unaccounted variability): 47.06%
## H^2 (unaccounted variability / sampling variability):   1.89
## 
## Test for Residual Heterogeneity: 
## QE(df = 320) = 612.5926, p-val < .0001
## 
## Test of Moderators (coefficient(s) 1,2,3,4): 
## QM(df = 4) = 492.5922, p-val < .0001
## 
## Model Results:
## 
##                   estimate      se      zval    pval    ci.lb    ci.ub
## applicationsCat1   -0.9808  0.0957  -10.2483  <.0001  -1.1683  -0.7932
## applicationsCat2   -1.6748  0.0877  -19.1016  <.0001  -1.8466  -1.5029
## applicationsCat3   -1.4124  0.2966   -4.7618  <.0001  -1.9937  -0.8310
## applicationsCat5   -0.0577  0.4245   -0.1359  0.8919  -0.8896   0.7742
##                      
## applicationsCat1  ***
## applicationsCat2  ***
## applicationsCat3  ***
## applicationsCat5     
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Meta-analysis of yield

Overall mean difference

overall.yield <- rma.uni(yi = yi, vi = vi, data = yield.data.MD, method = "REML")
summary(overall.yield)
## 
## Random-Effects Model (k = 513; tau^2 estimator: REML)
## 
##     logLik    deviance         AIC         BIC        AICc  
## -3963.2069   7926.4138   7930.4138   7938.8904   7930.4374  
## 
## tau^2 (estimated amount of total heterogeneity): 119824.4635 (SE = 18147.5135)
## tau (square root of estimated tau^2 value):      346.1567
## I^2 (total heterogeneity / total variability):   43.49%
## H^2 (total variability / sampling variability):  1.77
## 
## Test for Heterogeneity: 
## Q(df = 512) = 887.3935, p-val < .0001
## 
## Model Results:
## 
## estimate       se     zval     pval    ci.lb    ci.ub          
## 525.0076  24.6017  21.3403   <.0001 476.7892 573.2260      *** 
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Effects of Active Ingredient on SMD

yield.active <- rma.uni(yi = yi, vi = vi, data=yield.data.SMD,
                      method = "REML", mods = ~activeIngClean-1)
summary(yield.active)
## 
## Mixed-Effects Model (k = 513; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -526.1748  1052.3496  1078.3496  1133.1654  1079.0970  
## 
## tau^2 (estimated amount of residual heterogeneity):     0 (SE = 0.0342)
## tau (square root of estimated tau^2 value):             0
## I^2 (residual heterogeneity / unaccounted variability): 0.00%
## H^2 (unaccounted variability / sampling variability):   1.00
## 
## Test for Residual Heterogeneity: 
## QE(df = 501) = 419.8387, p-val = 0.9965
## 
## Test of Moderators (coefficient(s) 1,2,3,4,5,6,7,8,9,10,11,12): 
## QM(df = 12) = 467.2884, p-val < .0001
## 
## Model Results:
## 
##                      estimate      se     zval    pval    ci.lb   ci.ub
## activeIngCleanazo      0.2080  0.1412   1.4732  0.1407  -0.0687  0.4847
## activeIngCleancypr     0.3131  0.2966   1.0559  0.2910  -0.2681  0.8944
## activeIngCleandual     1.1559  0.1049  11.0143  <.0001   0.9502  1.3616
## activeIngCleanflus     0.7149  0.2806   2.5478  0.0108   0.1650  1.2649
## activeIngCleanflut     0.7107  0.0985   7.2124  <.0001   0.5176  0.9038
## activeIngCleanmixed    0.6782  0.0583  11.6359  <.0001   0.5639  0.7924
## activeIngCleanmyc      0.7688  0.2268   3.3902  0.0007   0.3243  1.2133
## activeIngCleanoth      0.5584  0.0990   5.6415  <.0001   0.3644  0.7524
## activeIngCleanpyra     0.6662  0.1104   6.0324  <.0001   0.4497  0.8826
## activeIngCleantebu     0.7252  0.1098   6.6039  <.0001   0.5100  0.9405
## activeIngCleantetra    0.5792  0.1244   4.6544  <.0001   0.3353  0.8231
## activeIngCleanthio     0.4634  0.2377   1.9494  0.0513  -0.0025  0.9292
##                         
## activeIngCleanazo       
## activeIngCleancypr      
## activeIngCleandual   ***
## activeIngCleanflus     *
## activeIngCleanflut   ***
## activeIngCleanmixed  ***
## activeIngCleanmyc    ***
## activeIngCleanoth    ***
## activeIngCleanpyra   ***
## activeIngCleantebu   ***
## activeIngCleantetra  ***
## activeIngCleanthio     .
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Effects of Class on SMD

yield.class <- rma.uni(yi = yi, vi = vi, data=yield.data.SMD,
                     method = "REML", mods = ~classClean-1)
summary(yield.class)
## 
## Mixed-Effects Model (k = 513; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -541.6601  1083.3202  1095.3202  1120.7030  1095.4878  
## 
## tau^2 (estimated amount of residual heterogeneity):     0.0000 (SE = 0.0339)
## tau (square root of estimated tau^2 value):             0.0029
## I^2 (residual heterogeneity / unaccounted variability): 0.00%
## H^2 (unaccounted variability / sampling variability):   1.00
## 
## Test for Residual Heterogeneity: 
## QE(df = 508) = 442.1905, p-val = 0.9838
## 
## Test of Moderators (coefficient(s) 1,2,3,4,5): 
## QM(df = 5) = 444.9295, p-val < .0001
## 
## Model Results:
## 
##                        estimate      se     zval    pval    ci.lb   ci.ub
## classCleanmma            0.7923  0.0511  15.5101  <.0001   0.6922  0.8925
## classCleanother          0.6390  0.1092   5.8520  <.0001   0.4250  0.8530
## classCleanstrobilurin    0.4391  0.0831   5.2835  <.0001   0.2762  0.6020
## classCleanthiophanate    0.4310  0.2514   1.7144  0.0864  -0.0617  0.9236
## classCleantriazole       0.6775  0.0574  11.8011  <.0001   0.5650  0.7900
##                           
## classCleanmma          ***
## classCleanother        ***
## classCleanstrobilurin  ***
## classCleanthiophanate    .
## classCleantriazole     ***
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Effects of Growth stage on SMD

yield.growth.stage <- rma.uni(yi = yi, vi = vi, 
                            data=yield.data.SMD[yield.data.SMD$growthStateClean!="unknown",],
                            method = "REML", mods = ~growthStateClean-1)
summary(yield.growth.stage)
## 
## Mixed-Effects Model (k = 490; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -511.1338  1022.2675  1038.2675  1071.7077  1038.5713  
## 
## tau^2 (estimated amount of residual heterogeneity):     0 (SE = 0.0347)
## tau (square root of estimated tau^2 value):             0
## I^2 (residual heterogeneity / unaccounted variability): 0.00%
## H^2 (unaccounted variability / sampling variability):   1.00
## 
## Test for Residual Heterogeneity: 
## QE(df = 483) = 414.1355, p-val = 0.9895
## 
## Test of Moderators (coefficient(s) 1,2,3,4,5,6,7): 
## QM(df = 7) = 443.9602, p-val < .0001
## 
## Model Results:
## 
##                     estimate      se     zval    pval   ci.lb   ci.ub     
## growthStateClean1     0.4066  0.1863   2.1828  0.0291  0.0415  0.7717    *
## growthStateClean1+    0.7522  0.0992   7.5814  <.0001  0.5577  0.9466  ***
## growthStateClean2     0.5839  0.1554   3.7587  0.0002  0.2794  0.8884  ***
## growthStateClean2+    0.7995  0.0643  12.4296  <.0001  0.6734  0.9256  ***
## growthStateClean3     0.6083  0.0540  11.2615  <.0001  0.5024  0.7142  ***
## growthStateClean4     1.6593  0.2212   7.5005  <.0001  1.2257  2.0929  ***
## growthStateClean5     0.5119  0.0934   5.4787  <.0001  0.3288  0.6950  ***
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Effects of application number on SMD

table(yield.data.SMD$applicationsNumb)
## 
##   1   2   3 
## 254 238  21
yield.applications <- rma.uni(yi = yi, vi = vi, 
                            data=yield.data.SMD,
                            method = "REML", mods = ~applicationsNumb)
summary(yield.applications)
## 
## Mixed-Effects Model (k = 513; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -546.4379  1092.8757  1098.8757  1111.5848  1098.9230  
## 
## tau^2 (estimated amount of residual heterogeneity):     0.0000 (SE = 0.0338)
## tau (square root of estimated tau^2 value):             0.0025
## I^2 (residual heterogeneity / unaccounted variability): 0.00%
## H^2 (unaccounted variability / sampling variability):   1.00
## R^2 (amount of heterogeneity accounted for):            0.00%
## 
## Test for Residual Heterogeneity: 
## QE(df = 511) = 448.0084, p-val = 0.9792
## 
## Test of Moderators (coefficient(s) 2): 
## QM(df = 1) = 8.5465, p-val = 0.0035
## 
## Model Results:
## 
##                   estimate      se    zval    pval   ci.lb   ci.ub     
## intrcpt             0.4211  0.0944  4.4582  <.0001  0.2360  0.6062  ***
## applicationsNumb    0.1667  0.0570  2.9234  0.0035  0.0550  0.2785   **
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
# Categorical
yield.data.SMD$applicationsCat <- as.character(yield.data.SMD$applicationsNumb)
yield.applications.cat <- rma.uni(yi = yi, vi = vi, 
                                data=yield.data.SMD,
                                method = "REML", mods = ~applicationsCat-1)
summary(yield.applications.cat)
## 
## Mixed-Effects Model (k = 513; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -544.2503  1088.5007  1096.5007  1113.4383  1096.5799  
## 
## tau^2 (estimated amount of residual heterogeneity):     0.0000 (SE = 0.0338)
## tau (square root of estimated tau^2 value):             0.0027
## I^2 (residual heterogeneity / unaccounted variability): 0.00%
## H^2 (unaccounted variability / sampling variability):   1.00
## 
## Test for Residual Heterogeneity: 
## QE(df = 510) = 444.8633, p-val = 0.9827
## 
## Test of Moderators (coefficient(s) 1,2,3): 
## QM(df = 3) = 442.2577, p-val < .0001
## 
## Model Results:
## 
##                   estimate      se     zval    pval   ci.lb   ci.ub     
## applicationsCat1    0.5676  0.0469  12.1043  <.0001  0.4757  0.6595  ***
## applicationsCat2    0.7965  0.0478  16.6646  <.0001  0.7029  0.8902  ***
## applicationsCat3    0.6833  0.1609   4.2466  <.0001  0.3679  0.9987  ***
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Meta-analysis of seedwt

Overall mean difference

overall.seedwt <- rma.uni(yi = yi, vi = vi, data = seedwt.data.MD, method = "REML")
summary(overall.seedwt)
## 
## Random-Effects Model (k = 207; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -275.9736   551.9472   555.9472   562.6030   556.0063  
## 
## tau^2 (estimated amount of total heterogeneity): 0.0028 (SE = 0.0014)
## tau (square root of estimated tau^2 value):      0.0529
## I^2 (total heterogeneity / total variability):   14.28%
## H^2 (total variability / sampling variability):  1.17
## 
## Test for Heterogeneity: 
## Q(df = 206) = 221.5038, p-val = 0.2183
## 
## Model Results:
## 
## estimate       se     zval     pval    ci.lb    ci.ub          
##   0.1302   0.0119  10.9485   <.0001   0.1069   0.1535      *** 
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Effects of Active Ingredient on SMD

seedwt.active <- rma.uni(yi = yi, vi = vi, data=seedwt.data.SMD,
                       method = "REML", mods = ~activeIngClean-1)
summary(seedwt.active)
## 
## Mixed-Effects Model (k = 207; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -178.4013   356.8026   372.8026   399.1891   373.5565  
## 
## tau^2 (estimated amount of residual heterogeneity):     0 (SE = 0.0534)
## tau (square root of estimated tau^2 value):             0
## I^2 (residual heterogeneity / unaccounted variability): 0.00%
## H^2 (unaccounted variability / sampling variability):   1.00
## 
## Test for Residual Heterogeneity: 
## QE(df = 200) = 112.2592, p-val = 1.0000
## 
## Test of Moderators (coefficient(s) 1,2,3,4,5,6,7): 
## QM(df = 7) = 125.1656, p-val < .0001
## 
## Model Results:
## 
##                      estimate      se    zval    pval    ci.lb   ci.ub
## activeIngCleanazo      0.1393  0.2361  0.5899  0.5553  -0.3235  0.6020
## activeIngCleanflut     0.7203  0.1159  6.2129  <.0001   0.4931  0.9475
## activeIngCleanmixed    0.5333  0.0853  6.2502  <.0001   0.3661  0.7005
## activeIngCleanoth      0.4773  0.1372  3.4792  0.0005   0.2084  0.7461
## activeIngCleanpyra     0.5750  0.1668  3.4481  0.0006   0.2482  0.9018
## activeIngCleantebu     0.7533  0.1704  4.4197  <.0001   0.4192  1.0873
## activeIngCleantetra    0.3335  0.1751  1.9042  0.0569  -0.0098  0.6767
##                         
## activeIngCleanazo       
## activeIngCleanflut   ***
## activeIngCleanmixed  ***
## activeIngCleanoth    ***
## activeIngCleanpyra   ***
## activeIngCleantebu   ***
## activeIngCleantetra    .
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Effects of Class on SMD

seedwt.class <- rma.uni(yi = yi, vi = vi, data=seedwt.data.SMD,
                      method = "REML", mods = ~classClean-1)
summary(seedwt.class)
## 
## Mixed-Effects Model (k = 207; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -181.2380   362.4760   372.4760   389.0421   372.7806  
## 
## tau^2 (estimated amount of residual heterogeneity):     0 (SE = 0.0530)
## tau (square root of estimated tau^2 value):             0
## I^2 (residual heterogeneity / unaccounted variability): 0.00%
## H^2 (unaccounted variability / sampling variability):   1.00
## 
## Test for Residual Heterogeneity: 
## QE(df = 203) = 114.2499, p-val = 1.0000
## 
## Test of Moderators (coefficient(s) 1,2,3,4): 
## QM(df = 4) = 123.1749, p-val < .0001
## 
## Model Results:
## 
##                        estimate      se    zval    pval    ci.lb   ci.ub
## classCleanmma            0.5990  0.0832  7.1965  <.0001   0.4358  0.7621
## classCleanother          0.2338  0.1761  1.3275  0.1844  -0.1114  0.5789
## classCleanstrobilurin    0.3470  0.1460  2.3757  0.0175   0.0607  0.6332
## classCleantriazole       0.6278  0.0785  7.9987  <.0001   0.4740  0.7816
##                           
## classCleanmma          ***
## classCleanother           
## classCleanstrobilurin    *
## classCleantriazole     ***
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Effects of Growth stage on SMD

seedwt.growth.stage <- rma.uni(yi = yi, vi = vi, 
                             data=seedwt.data.SMD[seedwt.data.SMD$growthStateClean!="unknown",],
                             method = "REML", mods = ~growthStateClean-1)
summary(seedwt.growth.stage)
## 
## Mixed-Effects Model (k = 198; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -161.0261   322.0522   336.0522   358.8546   336.6609  
## 
## tau^2 (estimated amount of residual heterogeneity):     0 (SE = 0.0546)
## tau (square root of estimated tau^2 value):             0
## I^2 (residual heterogeneity / unaccounted variability): 0.00%
## H^2 (unaccounted variability / sampling variability):   1.00
## 
## Test for Residual Heterogeneity: 
## QE(df = 192) = 86.8329, p-val = 1.0000
## 
## Test of Moderators (coefficient(s) 1,2,3,4,5,6): 
## QM(df = 6) = 149.4960, p-val < .0001
## 
## Model Results:
## 
##                     estimate      se    zval    pval    ci.lb   ci.ub     
## growthStateClean1+    0.8785  0.1589  5.5275  <.0001   0.5670  1.1900  ***
## growthStateClean2     0.2754  0.1561  1.7640  0.0777  -0.0306  0.5813    .
## growthStateClean2+    0.3292  0.1077  3.0565  0.0022   0.1181  0.5402   **
## growthStateClean3     0.5470  0.0823  6.6499  <.0001   0.3858  0.7082  ***
## growthStateClean4     1.4361  0.1977  7.2647  <.0001   1.0486  1.8235  ***
## growthStateClean5     0.5853  0.1900  3.0811  0.0021   0.2130  0.9576   **
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Effects of application number on SMD

table(seedwt.data.SMD$applicationsNumb)
## 
##   1   2 
## 116  91
seedwt.applications <- rma.uni(yi = yi, vi = vi, 
                             data=seedwt.data.SMD,
                             method = "REML", mods = ~applicationsNumb)
summary(seedwt.applications)
## 
## Mixed-Effects Model (k = 207; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -183.4827   366.9653   372.9653   382.9344   373.0847  
## 
## tau^2 (estimated amount of residual heterogeneity):     0 (SE = 0.0527)
## tau (square root of estimated tau^2 value):             0
## I^2 (residual heterogeneity / unaccounted variability): 0.00%
## H^2 (unaccounted variability / sampling variability):   1.00
## R^2 (amount of heterogeneity accounted for):            NA%
## 
## Test for Residual Heterogeneity: 
## QE(df = 205) = 116.4111, p-val = 1.0000
## 
## Test of Moderators (coefficient(s) 2): 
## QM(df = 1) = 4.3252, p-val = 0.0376
## 
## Model Results:
## 
##                   estimate      se    zval    pval    ci.lb   ci.ub   
## intrcpt             0.2443  0.1555  1.5708  0.1162  -0.0605  0.5491   
## applicationsNumb    0.2140  0.1029  2.0797  0.0376   0.0123  0.4156  *
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
# Categorical
seedwt.data.SMD$applicationsCat <- as.character(seedwt.data.SMD$applicationsNumb)
seedwt.applications.cat <- rma.uni(yi = yi, vi = vi, 
                                 data=seedwt.data.SMD,
                                 method = "REML", mods = ~applicationsCat-1)
summary(seedwt.applications.cat)
## 
## Mixed-Effects Model (k = 207; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -183.4827   366.9653   372.9653   382.9344   373.0847  
## 
## tau^2 (estimated amount of residual heterogeneity):     0 (SE = 0.0527)
## tau (square root of estimated tau^2 value):             0
## I^2 (residual heterogeneity / unaccounted variability): 0.00%
## H^2 (unaccounted variability / sampling variability):   1.00
## 
## Test for Residual Heterogeneity: 
## QE(df = 205) = 116.4111, p-val = 1.0000
## 
## Test of Moderators (coefficient(s) 1,2): 
## QM(df = 2) = 121.0136, p-val < .0001
## 
## Model Results:
## 
##                   estimate      se    zval    pval   ci.lb   ci.ub     
## applicationsCat1    0.4583  0.0673  6.8045  <.0001  0.3263  0.5903  ***
## applicationsCat2    0.6722  0.0778  8.6436  <.0001  0.5198  0.8246  ***
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Meta-analysis of Cercospora severity

Overall mean difference

overall.cerco <- rma.uni(yi = yi, vi = vi, data = cerco.data.MD, method = "REML")
summary(overall.cerco)
## 
## Random-Effects Model (k = 40; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -151.1431   302.2862   306.2862   309.6134   306.6196  
## 
## tau^2 (estimated amount of total heterogeneity): 0 (SE = 57.1784)
## tau (square root of estimated tau^2 value):      0
## I^2 (total heterogeneity / total variability):   0.00%
## H^2 (total variability / sampling variability):  1.00
## 
## Test for Heterogeneity: 
## Q(df = 39) = 14.4324, p-val = 0.9999
## 
## Model Results:
## 
## estimate       se     zval     pval    ci.lb    ci.ub          
##  -9.6508   2.5198  -3.8300   0.0001 -14.5894  -4.7121      *** 
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Meta-analysis of Target spot severity

Overall mean difference

overall.target.spot <- rma.uni(yi = yi, vi = vi, data = target.spot.data.MD, method = "REML")
summary(overall.target.spot)
## 
## Random-Effects Model (k = 31; tau^2 estimator: REML)
## 
##    logLik   deviance        AIC        BIC       AICc  
## -135.3497   270.6994   274.6994   277.5018   275.1439  
## 
## tau^2 (estimated amount of total heterogeneity): 285.6250 (SE = 125.3878)
## tau (square root of estimated tau^2 value):      16.9004
## I^2 (total heterogeneity / total variability):   58.82%
## H^2 (total variability / sampling variability):  2.43
## 
## Test for Heterogeneity: 
## Q(df = 30) = 72.8437, p-val < .0001
## 
## Model Results:
## 
## estimate       se     zval     pval    ci.lb    ci.ub          
## -18.9129   3.9579  -4.7785   <.0001 -26.6703 -11.1555      *** 
## 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Save for graphing

save(overall.rust, overall.yield, overall.seedwt, overall.cerco, overall.target.spot,
     rust.active, rust.class, rust.growth.stage, rust.applications, rust.applications.cat,
     yield.active, yield.class, yield.growth.stage, yield.applications, yield.applications.cat,
     seedwt.active, seedwt.class, seedwt.growth.stage, seedwt.applications, seedwt.applications.cat,
     file="data/output_data/analysis_results.R")

Footer

Spun with ezspin(“programs/data_analysis.R”, out_dir=“output”, fig_dir=“figures”, keep_md=FALSE)

Session Info:

sessionInfo()
## R version 3.3.0 (2016-05-03)
## Platform: x86_64-apple-darwin13.4.0 (64-bit)
## Running under: OS X 10.10.5 (Yosemite)
## 
## locale:
## [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
## 
## attached base packages:
## [1] stats     graphics  grDevices utils     datasets  methods   base     
## 
## other attached packages:
## [1] doBy_4.5-15   metafor_1.9-8 Matrix_1.2-6  knitr_1.13    ezknitr_0.4  
## 
## loaded via a namespace (and not attached):
##  [1] lattice_0.20-33   R.methodsS3_1.7.1 MASS_7.3-45      
##  [4] grid_3.3.0        formatR_1.4       magrittr_1.5     
##  [7] evaluate_0.9      stringi_1.1.1     R.oo_1.20.0      
## [10] R.utils_2.3.0     tools_3.3.0       stringr_1.0.0    
## [13] markdown_0.7.7