survey_analysis_cog_appraisal_exploratory.Rmd

---
title: "R Notebook"
output: html_notebook
---

First let's load the survey data.

```{r}
library(dplyr)
library(ggplot2)
library(rstatix)
load("~/Dropbox (University of Oregon)/UO-SAN Lab/Berkman Lab/Devaluation/analysis_files/data/scored_data_coded.RData")
```

## Compare Session 2 and Session 1


```{r}

#grab the cognitive appraisal group
# stop("need to run this manually, and look up the specific Randlabel that applies for your session :-)")
# cog_appraisal_group <- ppt_list_clean[ppt_list_clean$arm_session_randlabel=="Kirk",]
# 
# 

```

```{r}

#iterate through each of our scales, and test their correlation with FCI and FFQ

promote_minus_prevent <- scored %>% 
  #filter for FFQ
  filter(scale_name %in% c("FFQ","FCI")) %>%
  #filter for the two cols we're interested in
  filter(scored_scale %in% c("cancer_promoting","cancer_preventing")) %>%
  #throw it out wide
  select(subid,scored_scale,score,survey_name,scale_name) %>% tidyr::pivot_wider(names_from = scored_scale,values_from = score) %>%
  #create a composite measure
  mutate("cancer_promoting_minus_preventing"=cancer_promoting-cancer_preventing) %>%
  #wide across the surveys so that we can compare the surveys
  select(scale_name,survey_name,subid,cancer_promoting_minus_preventing) %>% 
  tidyr::spread(survey_name,cancer_promoting_minus_preventing)

pmp_fci <- promote_minus_prevent %>% filter(scale_name=="FCI") %>% select(-scale_name) %>% 
  mutate(fci_s2_minus_s1 = `DEV Session 2 Surveys` - `DEV Session 1 Surveys`)
pmp_ffq <- promote_minus_prevent %>% filter(scale_name=="FFQ") %>% select(-scale_name) %>% 
  mutate(ffq_s2_minus_s1 = `DEV Session 2 Surveys` - `DEV Session 1 Surveys`)

outcome_difference_scales <- merge(pmp_fci,pmp_ffq, by="subid") %>% select(subid,fci_s2_minus_s1,ffq_s2_minus_s1)
```

## Understand the other scales

```{r}
#iterate through each scale; let's pick the ones that we think might be worth checking...
#should capture three scales: BFI, NCS, IMI, BIS-11, and PLAN
#if we've missed any of those we should feel free to add it in later.
scales_to_check <- c(
  "abstract_thinking",
  "avoid_depth",
  "cognitive",
  "cognitive_strategies",
  "deliberating_issues",
  "effort_importance",
  "extraversion","agreeableness","conscientiousness","neuroticism","openness",
  "get_job_done",
  "intellectual_task",
  "interest_enjoyment",
  "like_responsibility",
  "mental_forecasting",
  "new_solutions_to_problems",
  "non_planning",
  "perceived_choice","perceived_competence",
  "prefer_complex","prefer_little_thought",
  "relief_not_satisfaction","satisfaction_in_deliberating",
  "small_daily_projects","solve_puzzles",
  "tasks_little_thought","temporal_orientation","think_minimally","thinking_not_exciting","thinking_not_fun",
  "thought_appealing",
  "total_impulsiveness",
  "value_usefulness")

scored %>% filter(scored_scale %in% scales_to_check) %>% .$scale_name %>% table
#scales in the first survey are BFI, NCS, and PLAN
session_0_surveys <- scored %>% filter(survey_name=="DEV Session 0 Surveys")%>% filter(scale_name %in% c("BFI", "NCS","PLAN")) %>%
  filter(n_items>1) %>% #just get the constructs, not the individual questions
  select(scale_name,scored_scale,score,subid)

session_1_surveys <- scored %>% 
  filter(survey_name=="DEV Session 1 Surveys")%>% 
  filter(scale_name %in% c("BIS-11")) %>%
  filter(n_items>1) %>% #just get the constructs, not the individual questions
  select(scale_name,scored_scale,score,subid)

session_2_surveys <- scored %>% 
  filter(survey_name=="DEV Session 2 Surveys")%>% 
  filter(scale_name %in% c("IMI")) %>%
  filter(n_items>1) %>% #just get the constructs, not the individual questions
  select(scale_name,scored_scale,score,subid)

#now combine those

selected_trait_survey_results_long <- 
  #combine
  rbind(session_0_surveys,session_1_surveys,session_2_surveys) %>%
  #create a label
  mutate(scale_subscale_name = paste0(scale_name,"_",scored_scale))

#find out how many scales exactly we're dealing with
scale_count <- length(unique(selected_trait_survey_results_long$scale_subscale_name))

selected_trait_survey_results <- 
  selected_trait_survey_results_long %>%
  #select just the cols we want
  select(score,subid,scale_subscale_name) %>% 
  #throw it out wide
  tidyr::pivot_wider(names_from = scale_subscale_name,values_from = score)




```


```{r}
results_traits_and_outcomes <- merge(outcome_difference_scales,selected_trait_survey_results,by="subid")

library(GGally)
ggpairs(results_traits_and_outcomes %>% select(-subid),columns=c(1,2))
```




```{r}
library(corrplot)

data_to_corr <- results_traits_and_outcomes %>% select(-subid)

corrplot.mixed(cor(data_to_corr,use = "pairwise.complete.obs"), order="hclust", tl.col="black")



```





This should show us any correlations that we should be paying attention to.




```{r}
cor_results <- cor(data_to_corr,use = "pairwise.complete.obs")[,c("fci_s2_minus_s1","ffq_s2_minus_s1")]

dframe <- NULL
for (x in c("fci_s2_minus_s1","ffq_s2_minus_s1")){
  for (y in colnames(data_to_corr)[3:ncol(data_to_corr)]){
    #print(paste0("corr between",x,y))
    cor_test_result <- cor.test(data_to_corr[,x],data_to_corr[,y])
    dframe_row <- data.frame("x"=x,"y"=y, "r"=cor_test_result$estimate, "pval"=cor_test_result$p.value)
    if(is.null(dframe)){
      dframe <- dframe_row
    }else{
      dframe <- rbind(dframe,dframe_row)
    }
    
  }
}

```



```{r}
min(p.adjust(dframe$pval,method="BH"))
```


```{r}
results_traits_and_outcomes2 <- merge(outcome_difference_scales %>% mutate(s2_minus_s1 = fci_s2_minus_s1 + ffq_s2_minus_s1) %>% select("subid","s2_minus_s1"),selected_trait_survey_results,by="subid")
data_to_corr <- results_traits_and_outcomes2 %>% select(-subid)


dframe <- NULL
for (x in c("s2_minus_s1")){
  for (y in colnames(data_to_corr)[3:ncol(data_to_corr)]){
    #print(paste0("corr between",x,y))
    cor_test_result <- cor.test(data_to_corr[,x],data_to_corr[,y])
    dframe_row <- data.frame("x"=x,"y"=y, "r"=cor_test_result$estimate, "pval"=cor_test_result$p.value)
    if(is.null(dframe)){
      dframe <- dframe_row
    }else{
      dframe <- rbind(dframe,dframe_row)
    }
    
  }
}

```

```{r}
min(p.adjust(dframe$pval,method="BH"))
```

No, there is no scale that predicts s2 minus s1 performance....


## Separate out conditions

OK, but what about within each condition?

```{r}
results_traits_and_outcomes3 <- merge(
  ppt_list_clean %>% select(subid,arm_session_randlabel),
  results_traits_and_outcomes2)


groups <- unique(results_traits_and_outcomes3$arm_session_randlabel)
groups_no_control <- setdiff(groups,control_group)
dframe <- NULL
for (g in groups_no_control){
  group_results <- results_traits_and_outcomes3 %>% filter(arm_session_randlabel==g) %>% select(-arm_session_randlabel)
  data_to_corr <- group_results %>% select(-subid)
  
  for (x in c("s2_minus_s1")){
    for (y in colnames(data_to_corr)[3:ncol(data_to_corr)]){
      #print(paste0("corr between",x,y))
      cor_test_result <- cor.test(data_to_corr[[x]],data_to_corr[[y]])
      dframe_row <- data.frame("group"=g, "x"=x,"y"=y, "r"=cor_test_result$estimate, "pval"=cor_test_result$p.value)
      if(is.null(dframe)){
        dframe <- dframe_row
      }else{
        dframe <- rbind(dframe,dframe_row)
      }
      
    }
}
}


```
```{r}
dframe$pval_bhadjust <- p.adjust(dframe$pval,method="BH")
```