ess <- foreign::read.spss("Baza ESS1.sav")
re-encoding from UTF-8
ess <- as.data.frame(ess)
data <- dplyr::mutate_if(ess[,517:537],is.factor, as.character)
# covert string represenation of variables to numeric values
data[data == "Very much like me"] <- 1
data[data == "Like me"] <- 2
data[data == "Somewhat like me"] <- 3
data[data == "A little like me"] <- 4
data[data == "Not like me"] <- 5
data[data == "Not like me at all"] <- 6
data[data == "Don't know"] <- NA
data[data == "Refusal"] <- NA
data[data == "No answer"] <- NA
#handling mising data
VIM::aggr(data)
# extract basic statistic for dataset
summary_data <- describe(data)
summary_data <- as.data.frame(summary_data)
# add rownames from dataset to names of variables in tibble format
var_names <- rownames(summary_data)
summary_data$var_names <- var_names
# select only relevant data from summary
summary_data <- summary_data %>% dplyr::select(var_names,sd, min,max,skew,kurtosis,se)
# reorganize data to make var_names as first column in dataset
summary_data <- summary_data %>% dplyr::select(var_names, everything())
#round floating numeric data in dataset to 2 digits
dplyr::mutate_if(summary_data, is.numeric, round,digits = 2)
# get output in tabular form
summary_data %>%gt::gt()
| var_names |
sd |
min |
max |
skew |
kurtosis |
se |
| Q1* |
1.3771286 |
1 |
6 |
0.5790435 |
-0.5862819 |
0.03275166 |
| Q2* |
1.2263964 |
1 |
6 |
-0.9549656 |
0.3144621 |
0.02905207 |
| Q3* |
1.0818659 |
1 |
6 |
1.3889382 |
1.9383678 |
0.02565710 |
| Q4* |
1.5148398 |
1 |
6 |
0.2040218 |
-1.1666722 |
0.03596581 |
| Q5* |
1.1547939 |
1 |
6 |
1.2873721 |
1.2723298 |
0.02736355 |
| Q6* |
1.5692259 |
1 |
6 |
0.1345306 |
-1.2065370 |
0.03725707 |
| Q7* |
1.5429304 |
1 |
6 |
0.2224145 |
-1.1582473 |
0.03674685 |
| Q8* |
1.0294342 |
1 |
6 |
1.1761362 |
1.4268837 |
0.02439308 |
| Q9* |
1.2996622 |
1 |
6 |
0.6751729 |
-0.3219481 |
0.03079630 |
| Q10* |
1.3113493 |
1 |
6 |
0.5377279 |
-0.5507595 |
0.03110819 |
| Q11* |
1.1155229 |
1 |
6 |
1.0674035 |
0.7711320 |
0.02641818 |
| Q12* |
0.9646045 |
1 |
6 |
1.2560505 |
1.8694577 |
0.02283128 |
| Q13* |
1.3625622 |
1 |
6 |
0.5218108 |
-0.6268963 |
0.03230490 |
| Q14* |
1.1848085 |
1 |
6 |
1.1725628 |
0.8610250 |
0.02811426 |
| Q15* |
1.4325961 |
1 |
6 |
-0.8191307 |
-0.3494632 |
0.03391770 |
| Q16* |
1.0845117 |
1 |
6 |
1.0367127 |
0.8147598 |
0.02568376 |
| Q17* |
1.3988355 |
1 |
6 |
0.5686463 |
-0.6813000 |
0.03312768 |
| Q18* |
0.8679235 |
1 |
6 |
1.2903332 |
2.1089633 |
0.02053718 |
| Q19* |
0.9558014 |
1 |
6 |
1.2544859 |
1.6364556 |
0.02261025 |
| Q20* |
1.3444597 |
1 |
6 |
1.0235842 |
0.3273875 |
0.03183101 |
| Q21* |
1.5009709 |
1 |
6 |
0.3251656 |
-0.9690770 |
0.03558642 |
#exclude NAs from dataset
ess.data <- na.omit(data)
#convert dataset coulmns that were coverted to character during impot back o numeric form
ess.data <- mutate_if(ess.data,is.character, as.numeric)
ess.model <-'
SE =~ Q2 + Q17 + Q4 + Q13
TD =~ Q3 + Q8 + Q19 + Q12 + Q18
OP =~ Q1 + Q11 + Q6 + Q15 + Q10 + Q21
CON =~ Q5 + Q14 + Q7 + Q16 + Q9 + Q20'
#fit the cfa function to model
ess.fit <- lavaan::cfa(ess.model, data = ess.data)
#get modification indices with higest suggested impact on model (in this case more than 50 lower value of chi-square test)
modification_indices <- modificationindices(ess.fit) %>%
as_data_frame()%>%
arrange(-mi)%>%
filter(mi > 50 & op == "=~")%>%
dplyr::select(lhs, op, rhs, mi,sepc.all)
modification_indices %>%gt::gt()
| lhs |
op |
rhs |
mi |
sepc.all |
| TD |
=~ |
Q11 |
249.13276 |
0.4612053 |
| CON |
=~ |
Q15 |
223.61241 |
-0.4254639 |
| CON |
=~ |
Q2 |
216.31693 |
-0.4722874 |
| TD |
=~ |
Q15 |
215.62325 |
-0.4239874 |
| CON |
=~ |
Q11 |
200.19414 |
0.4071356 |
| TD |
=~ |
Q2 |
192.51620 |
-0.4109670 |
| CON |
=~ |
Q17 |
96.79052 |
0.3075721 |
| OP |
=~ |
Q17 |
90.58969 |
-0.5418538 |
| OP |
=~ |
Q13 |
88.53532 |
0.6390393 |
| SE |
=~ |
Q10 |
56.10605 |
0.4032900 |
#plot path diagram
lavaanPlot::lavaanPlot(mode = ess.fit,coefs = T,covs = T,stand = T,stars = c("latent","covs"))