Analyses.qmd

---
title: "Pig Personality"
author: "Noah Stetson"
date: "`r Sys.Date()`"
format: html
editor: visual
editor_options: 
  chunk_output_type: inline
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE) #to be honest I have no idea what this does
```

# Personality In Pigs: Data Analysis

#### This is the data analysis for my undergraduate thesis on personality in pigs 🐷

### Goals:

1.  Load library and datasets

    -   Load datasets from GitHub so people can look at them without having to download a billion CSV files

2.  Clean data so it can be analyzed in a variety of ways

3.  Look at survey data, how it varies, how pigs are scoring, etc.

    -   Internal reliability

        -   Inter-item correlation: Do certain test items correlate with each other? Do the test items correlate as expected?

        -   Inter-total correlation: Do certain personality scores correlate with each other?

        -   Inter-rater correlation: See how the 2 raters (D and T) correlate with how they scores the pigs

4.  Look at behavioral data, how it varies, what pigs are doing, etc.

5.  Compare survey data to behavioral data. Do certain behaviors correlate with certain survey scores?

6.  Compare pig demographics to their behaviors and personality survey ratings/scores. Do certain demographics (like sex and background) correlate with any behaviors or personality survey results?

------------------------------------------------------------------------

## Loading library

#### Make sure to download these if you don't already have them

```{r}
library(tidyverse)
library(dplyr)
library(ggplot2)
library(psych)
library(readr)
library(corrplot)
library(irr)
library(lares)
```

## Loading datasets

#### Pulled from GitHub so **you don't need to change these or download the CSV files**!!

```{r}
obs <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/behavior_data/Obs_Clean.csv", show_col_types = FALSE)  # behavior observations

behavior_counts <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/behavior_data/behavior_counts.csv", show_col_types = FALSE) # behavior observations but tidyer

behavior_cat_counts <-  read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/behavior_data/behavior_cat_counts.csv", show_col_types = FALSE) # behavior observations but tidyer

svy <-read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/survey_data/Survey_Nov7_Clean.csv", show_col_types = FALSE)  # survey data

svy_by_rater <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/survey_data/svy_with_averages.csv", show_col_types = FALSE) # survey data (but with scores split between raters)

D_svy <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/survey_data/svy_D.csv", show_col_types = FALSE) # survey data from rater D

T_svy <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/survey_data/svy_T.csv", show_col_types = FALSE) # survey data from rater T

Avg_svy <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/survey_data/svy_Avg.csv", show_col_types = FALSE) # survey data from averages of raters

meta <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/PigSample.csv", show_col_types = FALSE) # meta data (pig demographics)
```

## Cleaning Data

Put meta data into df based on pig id number

```{r}
svy_m <- inner_join(meta, svy, by = "PigID")
obs_m <- inner_join(meta, obs, by = "PigName")
svy_avg_m <- inner_join(meta, svy_by_rater, by = "PigID")
```

### Survey data

Test items only version

```{r}
svy_items <- svy_avg_m %>% select(1:14,'Time.Known.D','Time.Known.T','Time.Known.Avg','D1.D':'D35.D','D1.T':'D35.T','D1.Avg':'D35.Avg')
```

Facet scores only version

```{r}
svy_f <- svy_avg_m %>% select(1:14,'Time.Known.D','Time.Known.T','Time.Known.Avg','F1.D':'F4.D','F1.T':'F4.T','F1.Avg':'F4.Avg')
```

Sub-facet scores only version

```{r}
svy_sf <- svy_avg_m %>% select(1:14,'Time.Known.D','Time.Known.T','Time.Known.Avg','SC1.D':'SC12.D','SC1.T':'SC12.T','SC1.Avg':'SC12.Avg') # All of them
```

### Make tidy version of survey results

Items

```{r}
items_tidy <-svy_items %>% 
  pivot_longer(cols = 'Time.Known.D':'Time.Known.Avg', names_to = "Time.Known", values_to = 'Time.Known.Value')
items_tidy <-items_tidy %>% 
  pivot_longer(cols = 'D1.D':'D35.D', names_to = "item.D", values_to = 'item.D.value')
items_tidy <-items_tidy %>% 
  pivot_longer(cols = 'D1.T':'D35.T', names_to = "item.T", values_to = 'item.T.value')
items_tidy <-items_tidy %>% 
  pivot_longer(cols = 'D1.Avg':'D35.Avg', names_to = "item.Avg", values_to = 'item.Avg.value')
```

Facet scores

```{r}
facets_tidy <-svy_f %>% 
  pivot_longer(cols = 'Time.Known.D':'Time.Known.Avg', names_to = "Time.Known", values_to = 'Time.Known.Value')
facets_tidy <-facets_tidy %>% 
  pivot_longer(cols = 'F1.D':'F4.D', names_to = "facet.D", values_to = 'facet.D.value')
facets_tidy <-facets_tidy %>% 
  pivot_longer(cols = 'F1.T':'F4.T', names_to = "facet.T", values_to = 'facet.T.value')
facets_tidy <-facets_tidy %>% 
  pivot_longer(cols = 'F1.Avg':'F4.Avg', names_to = "facet.Avg", values_to = 'facet.Avg.value')
```

Sub-facet scores

```{r}
subfacets_tidy <-svy_sf %>% 
  pivot_longer(cols = 'Time.Known.D':'Time.Known.Avg', names_to = "Time.Known", values_to = 'Time.Known.Value')
subfacets_tidy <-subfacets_tidy %>% 
  pivot_longer(cols = 'SC1.D':'SC12.D', names_to = "subfacet.D", values_to = 'subfacet.D.value')
subfacets_tidy <-subfacets_tidy %>% 
  pivot_longer(cols = 'SC1.T':'SC12.T', names_to = "subfacet.T", values_to = 'subfacet.T.value')
subfacets_tidy <-subfacets_tidy %>% 
  pivot_longer(cols = 'SC1.Avg':'SC12.Avg', names_to = "subfacet.Avg", values_to = 'subfacet.Avg.value')
```

### Behavior Observations

```{r}
# Make wide version
behavior_counts_wide <-behavior_counts %>% 
  pivot_longer(cols = 2:21 , names_to = "Behavior", values_to = 'Count')


# Also make version with meta data
behavior_m <- inner_join(behavior_counts, meta, by = "PigName")
```

```{r}
# Make wide version
behavior_cat_counts_wide <-behavior_cat_counts %>% 
  pivot_longer(cols = 2:8 , names_to = "Behavior.Cat", values_to = 'Count')


# Also make version with meta data
behavior_cat_m <- inner_join(behavior_cat_counts, meta, by = "PigName")
```

okay done with that phew

------------------------------------------------------------------------

# Preliminary Analyses

## Overview of Personality Survey Scores

Item ratings

```{r}
all_items <- svy_items %>% select(16:120)
describe(all_items)
```

Sub-facet scores

```{r}
all_sf <- svy_sf %>% select(16:51)
describe(all_sf)
```

### As a graph

### Sort by main facet

```{r}
facets_tidy$facet.Avg <- factor(facets_tidy$facet.Avg, levels=rev(unique(facets_tidy$facet.Avg)))


ggplot(facets_tidy, aes(x= facet.Avg, y= facet.Avg.value)) +
  geom_boxplot(color = "darkgoldenrod", fill='lightgoldenrod1') +
  theme_classic() +
  xlab("Personality Facet") +
  ylab("Score") +
  coord_flip() +
  ggtitle("Variation in Personality Scores Across 20 Pigs")
```

### Graph of sub-facets

```{r}
subfacets_tidy$subfacet.Avg <- factor(subfacets_tidy$subfacet.Avg, levels=rev(unique(subfacets_tidy$subfacet.Avg)))


ggplot(subfacets_tidy, aes(x= subfacet.Avg, y= subfacet.Avg.value)) +
  geom_boxplot(color = "darkgoldenrod", fill='lightgoldenrod1') +
  theme_classic() +
  xlab("Personality Sub-Facet") +
  ylab("Score") +
  coord_flip() +
  ggtitle("Variation in Personality Scores Across 20 Pigs")
```

## Internal Reliability

### Inter-item correlation

(correlations between each pig's average raw item scores)

### Overall

UGLY !!!!

but good to look at the patterns or whatever

This shows how test item scores correlate across all the pigs

```{r}
item_matrix_table <- cor(svy_by_rater[,c(3:37)]) #3:37 is the test items
corrplot(item_matrix_table, addCoef.col = 1,
         number.cex = 0.3, tl.cex = 0.5)
```

```{r}
avg_items <- Avg_svy %>% select(3:37)
inter_item <- avg_items %>% correlate() %>% select(-term)
inter <- colMeans(inter_item, na.rm = TRUE)

mean(inter)
# 0.00679244

inter
# this shows the item correlations as a table

```

### By facet

#### Fearfulness

This looks at all the test items that go with the "fearfulness" facet to see how well they correlate with each other

You would think that if all these test items are looking for the same thing, then they would mostly correlate with each other

```{r}
avg_f1 <- Avg_svy %>% select(D1.Avg, D5.Avg, D23.Avg, D3.Avg, D8.Avg, D19.Avg, D10.Avg, D18.Avg, D33.Avg, D13.Avg, D28.Avg, D34.Avg)
inter_f1 <- avg_f1 %>% correlate() %>% select(-term)

inter_f1 <- colMeans(inter_f1, na.rm = TRUE)

mean(inter_f1)
# 0.08263553 (low)

psych::alpha(avg_f1) # alpha value
# 0.47 (unacceptable)
```

#### Aggression Towards Humans

```{r}
avg_f2 <- Avg_svy %>% select(D6.Avg, D15.Avg, D31.Avg, D21.Avg, D25.Avg, D29.Avg)
inter_f2 <- avg_f2 %>% correlate() %>% select(-term)

inter_f2 <- colMeans(inter_f2, na.rm = TRUE)

mean(inter_f2)
# 0.3979195 (low)

psych::alpha(avg_f2)
# alpha: 0.75 (questionable)
```

#### Activity/Excitability

```{r}
avg_f3 <- Avg_svy %>% select(D12.Avg, D26.Avg, D32.Avg, D7.Avg, D14.Avg, D4.Avg, D11.Avg, D20.Avg, D17.Avg, D22.Avg, D30.Avg)
inter_f3 <- avg_f3 %>% correlate() %>% select(-term)

inter_f3 <- colMeans(inter_f3, na.rm = TRUE)

mean(inter_f3)
# 0.06783288 (low)

psych::alpha(avg_f3)
# 0.55 (poor)
```

#### Aggression/Dominance towards Pigs

```{r}
avg_f4 <- Avg_svy %>% select(D2.Avg, D16.Avg, D27.Avg, D9.Avg, D24.Avg, D35.Avg)
inter_f4 <- avg_f4 %>% correlate() %>% select(-term)

inter_f4 <- colMeans(inter_f4, na.rm = TRUE)

mean(inter_f4)
#0.1235088 (low)

psych::alpha(avg_f4)
# 0.56 (poor)
```

Hmm so it looks like the items that are supposed to be for the same construct actually correlate pretty low

### Inter-total correlation

(correlations between calculated personality sub-facet scores)

This could help me see how well different personality traits correlate with each other

For example, maybe fear of humans (SC4) and companionability (SC10) have a negative correlation idk

```{r}
svy_matrix_table <- cor(svy_by_rater[,c(38:49)]) #38:39 is the sub-facets

corrplot(svy_matrix_table, addCoef.col = 1,
         number.cex = 0.6, tl.cex = 0.7)
```

#### Strongish correlations I am noticing:

-   1 and 10 (-.78) (fear of humans negatively correlated with companionability)

-   3 and 12 (-.73) (fear of pigs negatively correlated with dominance over other pigs)

-   4 and 10 (-.75) (fear of handling negatively correlated with companionability)

-   6 and 12 (.79) (situational aggression positively correlated with dominance over other pigs)

#### Same thing but as a table

```{r}
avg_sf <- Avg_svy %>% select(38:49)
inter_total_sf <- avg_sf %>% correlate() %>% select(-term)
inter_total_sf_no_na <- colMeans(inter_total_sf, na.rm = TRUE)
```

## Inter-rater Reliability

Do rater 1 and rater 2 score each pig in a similar way

Make a new df because my brain is so smooth I literally cannot figure out how else to do it

```{r}
svy_wide <-svy_m %>% 
  pivot_longer(cols = 18:52, names_to = "item", values_to = 'item.value')
svy_wide <- svy_wide %>% 
  pivot_longer(cols = 18:29, names_to = "sub.facet", values_to ='subfacet.value')
svy_wide <- svy_wide %>% 
  pivot_longer(cols = 18:21, names_to = "facet", values_to = 'facet.value')
```

### See how rater D and T compare in how they overall scored the pigs

(between different personality sub-facets)

```{r}
svy_wide$sub.facet <- factor(svy_wide$sub.facet, levels=rev(unique(svy_wide$sub.facet)))
#this reorders the tick marks so it's not out of order
# why didn't it do it

ggplot(svy_wide, aes(x= sub.facet, y= subfacet.value, fill = Rater)) +
  geom_boxplot() +
  theme_classic() +
  scale_fill_manual(values = c("powderblue", "goldenrod1")) + #change colors and add outline
  xlab("Personality Sub-Facet") +
  ylab("Score") +
  ggtitle("Variation in Personality Scores Across 20 Pigs")
```

Overall, rater D appears to have more variation in their ratings than rater T

### Cohen's Kappa

I generated all these data frames by hand because I am stupid and dumb

#### How to interpret based on what Cohen said:

0 or less: No agreement

0.01-0.20: None to slight

0.21-0.40: Fair

0.41-0.60: Moderate

0.61-0.80: Substantial

0.81-1.00: Almost perfect

#### SC1 (Fear of Humans)

Kappa = 0.27

```{r}
irr_sc1 <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/irr_dfs/irr_sc1.csv", show_col_types = FALSE)

kappa2(irr_sc1[,c(1,2)], "equal")
```

#### SC2 (Nonsocial Fear)

Kappa = 0.167

```{r}
irr_sc2 <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/irr_dfs/irr_sc2.csv", show_col_types = FALSE)

kappa2(irr_sc2[,c(1,2)], "equal")
```

#### SC3 (Fear of Pigs)

Kappa = 0.0832

```{r}
irr_sc3 <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/irr_dfs/irr_sc3.csv", show_col_types = FALSE)
 
kappa2(irr_sc3[,c(1,2)], "equal")
```

#### SC4 (Fear of Handling)

Kappa = 0.27

```{r}
irr_sc4 <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/irr_dfs/irr_sc4.csv", show_col_types = FALSE)
kappa2(irr_sc4[,c(1,2)], "equal")
```

#### SC5 (General Aggression)

Kappa = 0.202

```{r}
irr_sc5 <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/irr_dfs/irr_sc5.csv", show_col_types = FALSE)

kappa2(irr_sc5[,c(1,2)], "equal")
```

#### SC6 (Situational Aggression)

Kappa = -0.0165

```{r}
irr_sc6 <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/irr_dfs/irr_sc6.csv", show_col_types = FALSE)

kappa2(irr_sc6[,c(1,2)], "equal")
```

#### SC7 (Excitability)

Kappa = -0.132

```{r}
irr_sc7 <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/irr_dfs/irr_sc7.csv", show_col_types = FALSE)

kappa2(irr_sc7[,c(1,2)], "equal")
```

#### SC8 (Playfulness)

Kappa = 0.0446

```{r}
irr_sc8 <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/irr_dfs/irr_sc8.csv", show_col_types = FALSE)

kappa2(irr_sc8[,c(1,2)], "equal")
```

#### SC9 (Active Engagement)

Kappa = 0.199

```{r}
irr_sc9 <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/irr_dfs/irr_sc9.csv", show_col_types = FALSE)

kappa2(irr_sc9[,c(1,2)], "equal")
```

#### SC10 (Companionability)

Kappa = 0.13

```{r}
irr_sc10 <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/irr_dfs/irr_sc10.csv", show_col_types = FALSE)

kappa2(irr_sc10[,c(1,2)], "equal")
```

#### SC11 (Aggression Towards Pigs)

Kappa = 0.101

```{r}
irr_sc11 <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/irr_dfs/irr_sc11.csv", show_col_types = FALSE)

kappa2(irr_sc11[,c(1,2)], "equal")
```

#### SC12 (Dominance Over Other Pigs)

Kappa = 0.269

```{r}
irr_sc12 <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/irr_dfs/irr_sc12.csv", show_col_types = FALSE)

kappa2(irr_sc12[,c(1,2)], "equal")
```

Overall it looks like the raters don't really agree on their assessments, though it does vary between personality sub-facets

### Graph comparing Kappa scores between sub-facets

Line shows where there is "fair" inter-observer reliability

```{r}
irr_results <- read_csv("https://raw.githubusercontent.com/nwstetson/PigPersonality/main/survey_data/irr_dfs/irr_results.csv", show_col_types = FALSE)

ggplot(irr_results, aes(x = Sub.facet, y = Kappa) ) + 
  geom_col(color = "navy", fill='lightsteelblue1') +
  theme_classic() +
  xlab("Personality Sub-Facet") +
  ylab("Kappa") +
  ggtitle("Inter-Rater Reliability Across Personality Scores") +
  geom_hline(yintercept=0.21, linetype="dashed", color = "red")
```

Mean IRR (not high)

```{r}
mean(irr_results$Kappa)
```

So that's not great but I will compare the pigs scores by using their average score between the 2 raters

## Overview of Behaviors

Yes I recorded everything in counts instead of times big mistake I don't want to talk about it 😍

### As a graph

they nudge the ground and sleep

```{r}
ggplot(behavior_counts_wide, aes(x= Behavior, y= Count)) +
  geom_boxplot(color = "navy", fill='lightsteelblue1') +
  theme_classic() +
  xlab("Behavior") +
  ylab("Counts") +
  ggtitle("Differences in Behavior Between Pigs")
```

Categories version

```{r}
ggplot(behavior_cat_counts_wide, aes(x= Behavior.Cat, y= Count)) +
  geom_boxplot(color = "navy", fill='lightsteelblue1') +
  theme_classic() +
  xlab("Behavior") +
  ylab("Counts") +
  ggtitle("Differences in Behavior Between Pigs")
```

------------------------------------------------------------------------

# Primary Analyses

### Do personality scores correlate with observable behaviors?

Make DF with behaviors and scores

```{r}
behavior_and_scores <- inner_join(behavior_m, Avg_svy, by = "PigID")
behavior_cat_and_scores <- inner_join(behavior_cat_m, Avg_svy, by = "PigID")
```

## Top correlations overview

AHH IT'S TERRIFYING

```{r}
svy_behav_matrix_table <- cor(behavior_and_scores[,c(2:21, 37:87)]) #takes out demographics

describe(svy_behav_matrix_table) # overview of correlations
corrplot(svy_behav_matrix_table, addCoef.col = 1,
         number.cex = 0.6, tl.cex = 0.7) # all the correlations

corrplot(svy_behav_matrix_table, addCoef.col = 1,
         number.cex = 0.4, tl.cex = 0.5, order="hclust", type="upper") # top correlations
```

Looking closer at top correlations

```{r}
svy_behav <- behavior_and_scores %>% select(2:21, 37:84)

corr_cross(svy_behav, # dataset
           max_pvalue = 0.05, # show only sig. correlations at selected level
           top = 25)# display top 10 correlations, any couples of variables 
```

### Fearfulness

```{r}
ggplot(behavior_cat_and_scores, aes(x=Anxious, y=F1.Avg)) +
  geom_point() +
  stat_smooth(method = "lm") +
  xlab("Number of Fearful Behaviors") +
  ylab("Fearfulness Score") +
  ggtitle("Fearfulness Score Versus Fearful Behaviors")

# Pearson's correlation
cor.test(behavior_cat_and_scores$Anxious, behavior_cat_and_scores$F1.Avg)
# cor = -0.1384581 (very weak negative correlation)
# p-value = 0.5605
```

### Activity/Excitability

```{r}
ggplot(behavior_cat_and_scores, aes(x=Anxious, y=F3.Avg)) +
  geom_point() +
  stat_smooth(method = "lm") +
  xlab("Number of Active Behaviors") +
  ylab("Activity/Excitability Score") +
  ggtitle("Activity/Excitability Score Versus Active Behaviors")

# Pearson's correlation
cor.test(behavior_cat_and_scores$Activity, behavior_cat_and_scores$F3.Avg)
# cor = 0.3857312 (moderate positive correlation)
# p-value = 0.09302
```

### Aggression Towards other Pigs

```{r}
ggplot(behavior_cat_and_scores, aes(x=G.agg, y=F2.Avg)) +
  geom_point() +
  stat_smooth(method = "lm") +
  xlab("Number of Aggressive Behaviors") +
  ylab("Aggression/Dominance Towards Pig Score") +
  ggtitle("Aggression/Dominance towards Pigs Score Versus Agressive Behaviors")

# Pearson's correlation
cor.test(behavior_cat_and_scores$G.agg, behavior_cat_and_scores$F2.Avg)
# cor = -0.1326601 (very weak negative correlation)
# p-value = 0.5772
```

```{r}
ggplot(behavior_and_scores, aes(x=Attempt.disp, y=SC11.Avg)) +
  geom_point() +
  stat_smooth(method = "lm") +
  xlab("Times Displaced") +
  ylab("Aggression Score") +
  ggtitle("Displacements vs Aggression Score")

# Pearson's correlation
cor.test(behavior_and_scores$Attempt.disp, behavior_and_scores$SC11.Avg)
# cor = 0.285504 (weak positive correlation)
# p-value = 0.2224

```

```{r}
ggplot(behavior_and_scores, aes(x=Get.disp, y=SC11.Avg)) +
  geom_point() +
  stat_smooth(method = "lm") +
  xlab("Times Displaced") +
  ylab("Aggression Score") +
  ggtitle("Do Pigs that Score Higher on Aggression Get Displaced Less?")

# Pearson's correlation
cor.test(behavior_and_scores$Get.disp, behavior_and_scores$SC11.Avg)
# cor = -0.397797 (moderate negative correlation)
# p-value = 0.08239
```

------------------------------------------------------------------------

## Behaviors (but not in categories this time) vs personality results

#### Get.disp vs SC12 (dominance over pigs)

```{r}
ggplot(behavior_and_scores, aes(x=Get.disp, y=SC12.Avg)) +
  geom_point() +
  stat_smooth(method = "lm") +
  xlab("Times Displaced") +
  ylab("Dominance Score") +
  ggtitle("Do Pigs that Score Higher on Dominance Get Displaced Less?")


# Pearson's correlation
cor.test(behavior_and_scores$Get.disp, behavior_and_scores$SC12.Avg)
# -0.2143021 (weak negative correlation)
```

#### Attempt.disp vs SC12 (dominance over pigs)

```{r}
# Correlation between pig's SC12 vs # of times they attempted to displace another pig
ggplot(behavior_and_scores, aes(x=Attempt.disp, y=SC12)) +
  geom_point() +
  stat_smooth(method = "lm") +
  xlab("Times Attempt Displace") +
  ylab("Dominance Score") +
  ggtitle("Do Pigs that Score Higher on Dominance Displace Pigs More?")

# Pearson's correlation
cor.test(behavior_and_scores$Attempt.disp, behavior_and_scores$SC12)
# 0.1807212 #also weak
```

#### Attempt.disp vs SC8 (playfulness)

```{r}
ggplot(behavior_and_scores, aes(x=Attempt.disp, y=SC8.Avg)) +
  geom_point() +
  stat_smooth(method = "lm") +
  xlab("Times Attempt Displace") +
  ylab("Playfulness Score") +
  ggtitle("Pigs that Displace Pigs More Score Lower in Playfulness")

# Pearson's correlation
cor.test(behavior_and_scores$Attempt.disp, behavior_and_scores$SC8.Avg)
# cor = -0.4235986
```

### Travel vs SC9 (active engagement)

```{r}
ggplot(behavior_and_scores, aes(x=Travel, y=SC9.Avg)) +
  geom_point() +
  stat_smooth(method = "lm") +
  xlab("Times Traveling") +
  ylab("Active Engagement Score") +
  ggtitle("Do Pigs that Score High on Active Engagement Travel More?")

# Pearson's correlation
cor.test(behavior_and_scores$Travel, behavior_and_scores$SC9)
#0.3479849
```

### Resting in proximity vs fear of humans

```{r}
cor.test(behavior_and_scores$Rest.proxim, behavior_and_scores$SC1.Avg)
#SC1 = -0.2909014
```

### Getting displaced vs companionability

```{r}
# Pearson's correlation
cor.test(behavior_and_scores$SC10.Avg, behavior_and_scores$Get.disp)
# 0.461596

ggplot(behavior_and_scores, aes(x=SC10.Avg, y=Get.disp)) +
  geom_point() +
  stat_smooth(method = "lm") +
  xlab("Times Pig Gets Displaced") +
  ylab("Companionability Score") +
  ggtitle("Pigs that Score Higher in Companionability Get Displaced More")
```

so I don't think behaviors really correlate with the personality scores but maybe it's because I had to change my survey last-minute and my ethogram no longer matched up with it like it was supposed to 😥

------------------------------------------------------------------------

# Additional Analyses

### Are there any sex differences in behavior?

Oh hmm not really

```{r}
corr_var(behavior_m, # dataset
         Sex_num, # name of variable to focus on
         top = 10) # display top 10 correlations
```

### Is there a sex difference in personality scores?

Maybe a little bit for aggression, but weak

```{r}
corr_var(svy_avg_m, # dataset
         Sex_num, # name of variable to focus on
         top = 10) # display top 10 correlations
```

### What about fearfulness?

Cor = -0.2600817

p = 0.2681

t= -1.1428

```{r}
cor.test(svy_avg_m$Sex_num, svy_avg_m$F1.Avg)
```

Sex vs aggression towards humans

Cor = -0.5578186

t = -2.8515

p= 0.0106

```{r}
cor.test(svy_avg_m$Sex_num, svy_avg_m$F2.Avg)

# Significant for F2. Here's what the mean scores are: 
aggregate(svy_avg_m$F2.Avg, list(svy_avg_m$Sex), FUN=mean) 
```

Activity/Excitablity

cor = 0.3940262

t = 1.8189

p = 0.08561

```{r}
cor.test(svy_avg_m$Sex_num, svy_avg_m$F3.Avg)
```

Aggression/Dominance Towards Pigs

t = -0.13546

cor = -0.03191228

p = 0.8938

```{r}
cor.test(svy_avg_m$Sex_num, svy_avg_m$F4.Avg)
```

Dominance?

Cor = 0.3450894

also very small

```{r}
cor.test(svy_avg_m$Sex_num, svy_m$F3)
```

### Overview of sex differences in personality scores

In SC1 it looks like males have greater variation than females

```{r}
ggplot(svy_wide, aes(x= sub.facet, y= subfacet.value, fill = Sex)) +
  geom_boxplot() +
  theme_classic() +
  scale_fill_manual(values = c("powderblue", "goldenrod1")) +
  xlab("Personality Sub-Facet") +
  ylab("Score") +
  ggtitle("Sex Differences in Personality Scores Across 20 Pigs")
```

Same thing but in big main facets instead of sub-facets

I mean it looks like there's kind of a difference but the correlations are small, I guess because the sample size is so small too

```{r}
ggplot(svy_wide, aes(x= facet, y= facet.value, fill = Sex)) +
  geom_boxplot() +
  theme_classic() +
  scale_fill_manual(values = c("powderblue", "goldenrod1")) +
  xlab("Personality Facet") +
  ylab("Score") +
  ggtitle("Sex Differences in Personality Scores Across 20 Pigs")

```

### Sex differences in getting displaced???

```{r}
ggplot(behavior_m, aes(x=Get.disp, fill = Sex)) + 
   theme_classic() +
  xlab("Times Getting Displaced") +
  ylab("umm math i think") +
  ggtitle("Do Male Pigs Get Displaced Less Than Female Pigs?") +
  geom_density(alpha = 0.4) +
  theme_minimal() +
  scale_fill_manual(values = c("goldenrod1", "darkorchid4"))


ggplot(behavior_m, aes(x=Get.disp, fill = Sex)) + 
   theme_classic() +
  xlab("Times Getting Displaced") +
  ylab("umm math i think") +
  ggtitle("Do Male Pigs Get Displaced Less Than Female Pigs?") +
  geom_hist() +
  theme_minimal() +
  scale_fill_manual(values = c("goldenrod1", "darkorchid4"))
```

### Background differences maybe?

#### Behaviors

umm please just look at behaviors R stop doing that

anyway it looks like it doesn't really correlate with anything

```{r}
corr_var(behavior_m, # dataset
         Background_num, # name of variable to focus on
         top = 25) # display top 10 correlations
```

### Background

```{r}
cor.test(svy_avg_m$Background_num, svy_avg_m$F1.Avg)
cor.test(svy_avg_m$Background_num, svy_avg_m$F2.Avg)
cor.test(svy_avg_m$Background_num, svy_avg_m$F3.Avg)
cor.test(svy_avg_m$Background_num, svy_avg_m$F4.Avg)

# Significant for F2. Here's what the mean scores are: 
aggregate(svy_avg_m$F2.Avg, list(svy_avg_m$Background), FUN=mean) 
```

### Group

```{r}
cor.test(svy_avg_m$Group_num, svy_avg_m$F1.Avg)
cor.test(svy_avg_m$Group_num, svy_avg_m$F2.Avg)
cor.test(svy_avg_m$Group_num, svy_avg_m$F3.Avg)
cor.test(svy_avg_m$Group_num, svy_avg_m$F4.Avg)

```

### Time Informant has Known Pig

```{r}
cor.test(svy_avg_m$Time.Known.Avg, svy_avg_m$F1.Avg)
cor.test(svy_avg_m$Time.Known.Avg, svy_avg_m$F2.Avg)
cor.test(svy_avg_m$Time.Known.Avg, svy_avg_m$F3.Avg)
cor.test(svy_avg_m$Time.Known.Avg, svy_avg_m$F4.Avg)
```

```{r}
ggplot(svy_avg_m, aes(y= F4.Avg, x= Time.Known.Avg)) +
  geom_point() +
  theme_classic() +
  stat_smooth(method = "lm") +
  scale_fill_manual(values = c("powderblue", "goldenrod1")) +
  xlab("Months Informant Has Known Pig For") +
  ylab("Aggression/Dominance Towards Pigs Score") +
  ggtitle("Months Known vs Aggression/Dominance Towards Pigs Score")
```

#### Survey results vs background

lol again it's just showing it correlates more with other demographics than survey items/scores

nothing over .5 so it doesn't look like it matters

```{r}
corr_var(svy_avg_m, # dataset
         Background_num, # name of variable to focus on
         top = 10) # display top 10 correlations
```

#### Background differences in getting displaced

```{r}
ggplot(behavior_m, aes(x=Get.disp, fill=Background)) + #fix name in DF later
  theme_classic() +
  xlab("Get Displaced") +
  ylab("Proportion") +
  ggtitle("Do Pigs From Abuse/Neglect Backgrounds Get Displaced More?") +
  geom_density(alpha = 0.4) +
  theme_minimal() +
  scale_fill_manual(name= "Background", labels=c("Abuse/Neglect", "Other"), values = c("brown2", "darkblue"))
```

#### As a graph

also an issue with this is idk if the owner surrendered pigs experienced abuse/neglect, probably a lot of them were neglected so that might be why there isn't much of a pattern

```{r}
ggplot(svy_avg_wide, aes(x= sub.facet, y= subfacet.value, fill = Background)) +
  geom_boxplot() +
  theme_classic() +
  scale_fill_manual(values = c("seagreen1", "lightblue1")) +
  xlab("Personality Sub-Facet") +
  ylab("Score") +
  ggtitle("Variation in Personality Scores Across 20 Pigs")
```