Packages
library(prcr)
library(tidyverse)
library(car)
library(rstatix)
library(effsize)
library(effectsize)
library(openxlsx)
Data
dados <- read.xlsx('C:/Users/user/Desktop/Vida acadêmica/Submissões/Netwine/Nova pasta/dadosEnraizamento.xlsx')
By Size
# Por Porte
dados %>% group_by(Porte) %>% summarise(mean = mean(Fator_inovação))
## # A tibble: 3 x 2
## Porte mean
## <chr> <dbl>
## 1 Porte_grande 0.332
## 2 Porte_médio -0.0529
## 3 Porte_pequena -0.190
dados %>% group_by(Porte) %>% summarise(mean = mean(autoridade1))
## # A tibble: 3 x 2
## Porte mean
## <chr> <dbl>
## 1 Porte_grande 0.472
## 2 Porte_médio 0.261
## 3 Porte_pequena 0.243
dados %>% group_by(Porte) %>% summarise(mean = mean(hub1))
## # A tibble: 3 x 2
## Porte mean
## <chr> <dbl>
## 1 Porte_grande 0.349
## 2 Porte_médio 0.281
## 3 Porte_pequena 0.350
Table Size X Position
table(dados$Porte, dados$blok1)
##
## Centro Periferia
## Porte_grande 8 4
## Porte_médio 10 8
## Porte_pequena 8 8
By Position
dados %>% group_by(blok1) %>% summarise(mean = mean(Fator_inovação))
## # A tibble: 2 x 2
## blok1 mean
## <chr> <dbl>
## 1 Centro 0.140
## 2 Periferia -0.183
t.test(Fator_inovação ~ blok1, data = dados) # p-value = 0.2144
##
## Welch Two Sample t-test
##
## data: Fator_inovação by blok1
## t = 1.2602, df = 43.133, p-value = 0.2144
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -0.1938356 0.8397988
## sample estimates:
## mean in group Centro mean in group Periferia
## 0.1404268 -0.1825548
dados %>% group_by(blok1) %>% summarise(mean = mean(autoridade1))
## # A tibble: 2 x 2
## blok1 mean
## <chr> <dbl>
## 1 Centro 0.436
## 2 Periferia 0.145
t.test(autoridade1 ~ blok1, data = dados) # 3.064e-06
##
## Welch Two Sample t-test
##
## data: autoridade1 by blok1
## t = 5.1321, df = 38.983, p-value = 8.257e-06
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## 0.1766425 0.4064613
## sample estimates:
## mean in group Centro mean in group Periferia
## 0.4364584 0.1449065
dados %>% group_by(blok1) %>% summarise(mean = mean(hub1))
## # A tibble: 2 x 2
## blok1 mean
## <chr> <dbl>
## 1 Centro 0.479
## 2 Periferia 0.120
t.test(hub1 ~ blok1, data = dados) # 9.068e-07
##
## Welch Two Sample t-test
##
## data: hub1 by blok1
## t = 6.2384, df = 36.179, p-value = 3.274e-07
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## 0.2419636 0.4750104
## sample estimates:
## mean in group Centro mean in group Periferia
## 0.4789219 0.1204349
Embeddeness X Innovation
dadosCluster <- dados[,c(59:68)]
dados$Embedded_Sum <- rowSums(dadosCluster)
dados$Embedded_Sum
## [1] 3 2 5 6 4 7 4 5 4 3 4 6 5 2 2 4 2 4 10 2 2 4 4 5 1
## [26] 8 0 0 3 4 3 1 4 5 1 4 0 4 3 9 2 9 3 2 3 2
# Quanto maior o score, menos enraízado nas redes locais
2 Steps Cluster
m3 <- create_profiles_cluster(dadosCluster, C13_Fornecedores,
C13_Clientes, C13_Concorrentes, C13_Consultores,
C13_Lab_Inst_Pesq, C13_Universidades,
C13_Grupo_Empr_Sind, C13_Imprensa_Esp,
C13_Reg_NorTec, C13_Leg_NormCodTrab,
n_profiles = 2)
plot_profiles(m3, to_center = TRUE)
m3$cluster
## [1] 1 2 2 1 1 2 2 2 2 1 1 2 2 1 1 1 1 2 2 1 1 2 2 1 1 2 1 1 2 2 1 1 2 2 1 2 1 1
## [39] 1 2 1 2 1 1 1 1
dados$TSCA <- m3$cluster
Summary
dados %>% group_by(TSCA) %>% summarise(mean = mean(Embedded_Sum))
## # A tibble: 2 x 2
## TSCA mean
## <int> <dbl>
## 1 1 2.46
## 2 2 5.3
# Grupo 1 = Mais Enraizado
# Grupo 2 = Menos enraízado
dados$TSCA <- factor(dados$TSCA,
labels = c("More Embedded", "Less Embedded"))
Lets see if the group Less Embedded have a superior Innovation Activity
dados %>% group_by(TSCA) %>% summarise(mean = mean(Fator_inovação)) # Means for groups
## # A tibble: 2 x 2
## TSCA mean
## <fct> <dbl>
## 1 More Embedded -0.229
## 2 Less Embedded 0.297
dados %>% group_by(TSCA) %>% shapiro_test(Fator_inovação) # Normality test for both groups
## # A tibble: 2 x 4
## TSCA variable statistic p
## <fct> <chr> <dbl> <dbl>
## 1 More Embedded Fator_inovação 0.959 0.371
## 2 Less Embedded Fator_inovação 0.948 0.341
leveneTest(Fator_inovação ~ TSCA, dados, center = mean) # Levene Test for homogeneity
## Levene's Test for Homogeneity of Variance (center = mean)
## Df F value Pr(>F)
## group 1 0.6078 0.4398
## 44
t.test(Fator_inovação ~ TSCA, data = dados, var.equal = T) # p-value = 0.04344, yep
##
## Two Sample t-test
##
## data: Fator_inovação by TSCA
## t = -2.0794, df = 44, p-value = 0.04344
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -1.03604919 -0.01620644
## sample estimates:
## mean in group More Embedded mean in group Less Embedded
## -0.2287512 0.2973766
cohens_d(Fator_inovação ~ TSCA, data = dados) # -0.62
## Cohen's d | 95% CI
## --------------------------
## -0.62 | [-1.21, -0.02]
##
## - Estimated using pooled SD.
d_to_common_language(-0.62)
## $`Cohen's U3`
## [1] 0.2676289
##
## $Overlap
## [1] 0.756561
##
## $`Probability of superiority`
## [1] 0.3305459
interpret_d(-0.62, rules = "cohen1988") # Medium effect, sounds good
## [1] "medium"
## (Rules: cohen1988)
interpret_d(-0.62, rules = "sawilowsky2009")
## [1] "medium"
## (Rules: sawilowsky2009)
Table
table(dados$Porte, dados$TSCA)
##
## More Embedded Less Embedded
## Porte_grande 5 7
## Porte_médio 10 8
## Porte_pequena 11 5
table(dados$blok1, dados$TSCA)
##
## More Embedded Less Embedded
## Centro 13 13
## Periferia 13 7
# Conclusion:
# It's seems that wineries that share more links outside the cluster indeed have a better innovation activity as Morrison 2009 argued