fix missing item

@AnafNeves it should work now - let me know if there are further issues

analysis/study1.qmd

@@ -14,7 +14,6 @@ format:
 execute: 
   cache: true
 ---
-
 ```{r}
 #| message: false
 #| warning: false
@@ -28,7 +27,6 @@ library(tidySEM)
 library(EGAnet)
 ```
 
-
 ## Data Preparation
 
 ::: {.panel-tabset}
@@ -47,7 +45,6 @@ df1b <- haven::read_sav("../data/Murphy2020/Study 6 IAS.sav") |>
   mutate(Gender = as.character(ifelse(Gender == 1, "Male", ifelse(Gender == 2, "Female", "Other"))))
 ```
 
-
 ### Sample 2
 
 Gaggero (2021) (https://osf.io/5x9sg)
@@ -56,6 +53,7 @@ Gaggero (2021) (https://osf.io/5x9sg)
 load("../data/Gaggero2021/data.rda")
 df2 <- data |> 
   mutate(Gender = as.character(gender)) |> 
+  mutate(across(starts_with("IAS "), as.numeric)) |>
   rename(
     Age=age, Heart = `IAS 1`, Hungry = `IAS 2`, Breathing = `IAS 3`, Thirsty = `IAS 4`,
     Urinate = `IAS 5`, Defecate = `IAS 6`, Taste = `IAS 7`, Vomit = `IAS 8`, Sneeze = `IAS 9`,
@@ -228,25 +226,6 @@ df9 <- df9 |>
   )
 ```
 
-<!-- ##### Sample 10 -->
-
-<!-- Bowling (2024) - https://osf.io/tx8h9/ -->
-
-<!-- ```{r} -->
-<!-- df10 <- haven::read_sav("../data/Bowling2024/TouchAttitudesData.sav") |> -->
-<!--   mutate_all(as.numeric) |> -->
-<!--   mutate(Gender = as.character(ifelse(Gender_coded == 1, "Male", ifelse(Gender_coded == 2, "Female", "Other")))) |> -->
-<!--   rename( -->
-<!--     Heart = IAS1, Hungry = IAS2, Breathing = IAS3, Thirsty = IAS4, -->
-<!--     Urinate = IAS5, Defecate = IAS6, Taste = IAS7, Vomit = IAS8, Sneeze = IAS9, -->
-<!--     Cough = IAS10, Temp = IAS11, Sex_arousal = IAS12, Wind = IAS13, Burp = IAS14, -->
-<!--     Muscles = IAS15, Bruise = IAS16, Pain = IAS17, Blood_Sugar = IAS18, -->
-<!--     Affective_touch = IAS19, Tickle = IAS20, Itch = IAS21 -->
-<!--   )  -->
-
-<!-- ``` -->
-
-
 ### Sample 10
 
 Makowski
@@ -283,7 +262,6 @@ data <- list(df1a=df1a, df1b=df1b, df2=df2, df3=df3, df4=df4, df5=df5, df6=df6,
 save(data, file = "../data/data.RData")
 ```
 
-
 :::
 
 
@@ -310,7 +288,6 @@ save(data, file = "../data/data.RData")
 
 Total N = `r nrow(df1a) + nrow(df1b) + nrow(df2) + nrow(df3) + nrow(df4) + nrow(df5) + nrow(df6) + nrow(df7a) + nrow(df7b) + nrow(df7c) + nrow(df8a) + nrow(df8b) + nrow(df9) + nrow(df10a) + nrow(df10b)`.
 
-
 ## Descriptive
 
 ### Distribution
@@ -399,7 +376,6 @@ data_all <- rbind(
 ) 
 ```
 
-
 ### Correlations
 
 An overall positive intercorrelation pattern, with no clear structure emerging.
@@ -528,8 +504,6 @@ make_correlation(data10b)
 
 
 :::
-
-
 ## Graph Analysis
 
 Exploratory Graph Analysis (EGA) is a recently developed framework for psychometric assessment, that can be used to estimate the number of dimensions in questionnaire data using network estimation methods and community detection algorithms, and assess the stability of dimensions and items using bootstrapping.
@@ -662,8 +636,6 @@ uva
 uva$keep_remove
 ```
 
-
-
 :::
 
 ### Networks
@@ -877,7 +849,6 @@ egas10b$p
 
 :::
 
-
 ### Structure Stability
 
 ::: {.panel-tabset}
@@ -1026,7 +997,6 @@ patchwork::wrap_plots(lapply(egas10a$egas, plot), nrow = 3)
 patchwork::wrap_plots(lapply(egas10b$egas, plot), nrow = 3)
 ```
 
-
 :::
 
 ### Scores
@@ -1058,7 +1028,6 @@ ega_scores_10a <- get_scores(data10a, ega)
 ega_scores_10b <- get_scores(data10b, ega)
 ```
 
-
 ## Factor Analysis
 
 ### How Many Factors
@@ -1215,7 +1184,6 @@ knitr::kable(as.data.frame(n)[n$n_Factors %in% c(1, 3, 4, 5), ], row.names = FAL
 :::
 
 
-
 ### EFA
 
 ::: {.panel-tabset}
@@ -1474,18 +1442,27 @@ is_converged <- function(m) {
 }
 
 compare_mods <- function(data) {
-  m1 <- lavaan::cfa(m1, data = data)
-  m4 <- lavaan::cfa(m4, data = data)
-  m5 <- lavaan::cfa(m5, data = data)
-  m6 <- lavaan::cfa(m6, data = data)
-  m7 <- lavaan::cfa(m7, data = data)
   
-  anova(m1, m4, m5, m6, m7) |>
+  fit_model <- function(f, data) {
+    if(!"Cough" %in% names(data)) {
+      f <- str_remove(f, fixed("Cough +"))
+    }
+    lavaan::cfa(f, data = data)
+  }
+  
+  
+  cfa1 <- fit_model(m1, data = data)
+  cfa4 <- fit_model(m4, data = data)
+  cfa5 <- fit_model(m5, data = data)
+  cfa6 <- fit_model(m6, data = data)
+  cfa7 <- fit_model(m7, data = data)
+  
+  anova(cfa1, cfa4, cfa5, cfa6, cfa7) |> 
     parameters::parameters() |>
     mutate(AIC = format(round(AIC, 0), nsmall = 0),
            BIC = format(round(BIC, 0), nsmall = 0),
            Chi2 = format(round(Chi2, 2), nsmall = 2),
-           Converged = sapply(list(m1, m4, m5, m6, m7), is_converged))
+           Converged = sapply(list(cfa1, cfa4, cfa5, cfa6, cfa7), is_converged))
 }
 
 display(compare_mods(data_all))
@@ -1641,6 +1618,7 @@ display(compare_mods(data10b))
 
 :::
 
+
 Conclusion: The 7th factor solution was preferred in most datasets, including by indices that penalize increased number of parameters (such as BIC).