Ford_2018_Replication
Ashish
10/26/2019
Libraries and reading data
rm(list=ls())
library(MASS)
library(tidyverse)## ── Attaching packages ─────────────────────────────── tidyverse 1.2.1 ──## ✔ ggplot2 3.2.1 ✔ purrr 0.3.2
## ✔ tibble 2.1.3 ✔ dplyr 0.8.3
## ✔ tidyr 1.0.0 ✔ stringr 1.4.0
## ✔ readr 1.3.1 ✔ forcats 0.4.0## ── Conflicts ────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
## ✖ dplyr::select() masks MASS::select()library(corrplot)## corrplot 0.84 loadedlibrary(interactions)
library(BayesFactor)## Loading required package: coda## Loading required package: Matrix##
## Attaching package: 'Matrix'## The following objects are masked from 'package:tidyr':
##
## expand, pack, unpack## ************
## Welcome to BayesFactor 0.9.12-4.2. If you have questions, please contact Richard Morey (richarddmorey@gmail.com).
##
## Type BFManual() to open the manual.
## ************df_raw <- read_csv("/Users/ashish/files/research/perab/data/raw/PerabW1_V4_Transformed_IntuitiveNames.csv")## Parsed with column specification:
## cols(
## .default = col_double(),
## recentpol_open = col_character(),
## news_open1 = col_character(),
## news_open2 = col_character(),
## news_open3 = col_character(),
## jail_repclassinfo = col_character(),
## jail_partviolprotest = col_character(),
## jail_corruption = col_character(),
## jail_unclassinfo = col_character(),
## jail_orgviolrally = col_character(),
## jail_engageviolspeech = col_character(),
## jail_advocateviolspeech = col_character(),
## jail_harrassofficial = col_character(),
## punish_comey = col_character(),
## punish_govofficials = col_character(),
## punish_offobstruct = col_character(),
## libcon_self = col_character(),
## libcon_trump = col_character(),
## partisan_fam = col_character(),
## partisan_friends = col_character(),
## race_other = col_character()
## # ... with 174 more columns
## )## See spec(...) for full column specifications.Creating groups of columns
Action columns
These columns have the stem: > “In the past six months, how often have you…?”
The columns are:
- Written a letter, email or made a phone call to a public official
- Signed a petition
- Participated in a political march, rally or protest
- Joined or became more active in a political group
- Encouraged someone else to engage in political activity (contacting officials, signing petitions, joining a group or march, etc.)
direct_action_cols <- c("action_contofficial",
"action_petition",
"action_rally",
"action_joingroup",
"action_encourage")This column has the stem: > “Since the election last November, would you say that your are…?”
The answer values are categorical and correspond to:
- 1 = more active
- 2 = less active
- 3 = same
pol_activity_cf_cols <- c("pol_activity")Reappraisal columns
These columns most directly correspond to items on the ERQ. The question stem is: > “Think about how you dealt with recent political news that was upsetting to you. How often, if at all, did you use the following strategies to manage or control your emotions to feel better? Some of the questions may seem similar but they are different in important ways so please consider each one carefully.”
The columns are:
- To feel more positive emotion (hope, happiness, pride), I changed the way I thought about upsetting political news
- To feel less negative emotion (worry, anger, sadness, disgust), I changed the way I thought about upsetting political news
reappraisal_basic_cols <- c("per_posreapp",
"per_negreapp") These columns I selected based on an exploratory factor analysis where three factors emerged. I interpreted this set of columns to all be attempting to engage in some type of cognitive change. (The other two factors were avoidance and social strategies.) The stem for these columns are the same as the previous section. The indidivual items are as follows:
- To feel more positive emotion (hope, happiness, pride), I changed the way I thought about upsetting political news
- To feel less negative emotion (worry, anger, sadness, disgust), I changed the way I thought about upsetting political news
- To feel better, I focused on the longer term meaning of the situation
- To feel better, I tried to think objectively about situation
- I tried to accept the political situation as it is rather than get emotional about it
- I tried to understand how people with different opinions feel about the situation
- I sought out information to help me place upsetting events in a larger context
reappraisal_like_cols <- c("per_posreapp",
"per_negreapp",
"per_longterm",
"per_objective",
"per_accept",
"per_understand",
"per_seekinfo")Felt emotion
These columns had the stem:
“When you think about recent political news, how much do you feel the following emotion?”
pos_emo_cols <- c("emo_excited", "emo_proud")
neg_emo_cols <- c("emo_angry", "emo_disgusted", "emo_anxious", "emo_shock", "emo_sad")All ER strategies
These are all the emotion regulation strategies queried.
er_strategy_cols <- c("per_avoid", "per_distract", "per_expressed",
"per_posreapp", "per_negreapp", "per_longterm",
"per_objective", "per_keeptoself", "per_notexpress",
"per_accept", "per_understand", "per_socialsupport", "per_seekinfo")Attention check
This was an attention check that 30% of the sample failed. The results look nearly the same when you omit the failures.
attention_check <- c("NCC3_6_5", "NCC3_7_5")Creating aggregate variables
Here I am making columns that aggregate the
df <- df_raw %>%
mutate(
direct_action = rowMeans(df_raw[,direct_action_cols], na.rm = T),
neg_emo_mean = rowMeans(df_raw[,neg_emo_cols], na.rm = T),
reappraisal_basic = rowMeans(df_raw[,reappraisal_basic_cols], na.rm = T),
reappraisal_like = rowMeans(df_raw[,reappraisal_like_cols], na.rm = T),
pol_activity_cf = fct_recode(as.factor(pol_activity),
"more"= "1",
"less" = "2",
"same" = "3"),
pol_activity_cf_more = (pol_activity_cf == "more"),
pol_activity_cf_less = (pol_activity_cf == "less"),
pol_extremity = abs(as.numeric(libcon_self) - 4)
)# %>% filter(NCC3_6_5 == 2 | NCC3_7_5 == 2)## Warning: NAs introduced by coercionSubsetting self-identified liberals and conservatives
df_lib <- df %>%
filter(libcon_self %in% 1:3)
df_conservative <- df %>%
filter(libcon_self %in% 5:7)Visualizing action
How much action did liberals take?
Individual types of action
df_lib %>%
pivot_longer(cols = direct_action_cols, names_to = "action_type", values_to = "action_amount") %>%
ggplot(aes(x = action_amount, fill = action_type)) +
geom_histogram(position = position_dodge(), binwidth = .2) +
scale_x_continuous(labels = c("Never", "Once or twice", "Several times", "Many times"))
Aggregated action variable
df_lib %>%
ggplot(aes(x = direct_action)) +
geom_histogram(binwidth = .2)
Ford Study 1a
Participants
140 students who voted for Clinton.
Procedure
- Filled out various questionnaires about personality, self-regulation, and well-being.
- Participants reported their voting behavior, and experiences learning about the election outcome.
- Participants then reported the extent to which they had been using election-focused reappraisal since the election.
- Filler items asked participants about their attitudes, goals, and social interactions before participants completed the measure of political action
- All data were collected between November 12, 2016 and December 1, 2016
Political action
- The items were (a) “participated in any protest, movement, or rally,” (b) “donated money,” and (c) “volunteered time.”
- Donated money and volunteered time were binned into 1-11 scale points
- Summed to create composite index (M = 1.15, SD = 2.28)
Election focused reappraisal
- Modified ERQ questions referring to election related emotion. (1-7 scale)
Results
- Individuals who used reappraisal more (vs. less) to manage their emotional responses to the election outcome were less likely to have engaged in political action since hearing about the election outcome, b = -.27, 95% CI[-.43, -.11], p = .001.
- All effects remain after controlling for age, gender, ethnicity, and SES.
Replication of Study 1a
Are the two general reappraisal items predictive of the mean of the action columns?
df_lib %>%
ggplot(aes(x = reappraisal_basic, y = direct_action)) +
geom_jitter(alpha = .2) +
geom_smooth(method="loess") +
geom_smooth(method="lm", linetype="dotted")
df_lib %>%
lm(direct_action ~ reappraisal_basic, .) %>%
summary##
## Call:
## lm(formula = direct_action ~ reappraisal_basic, data = .)
##
## Residuals:
## Min 1Q Median 3Q Max
## -0.8495 -0.6340 -0.2223 0.5660 2.1816
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.810621 0.060473 29.941 <2e-16 ***
## reappraisal_basic 0.007782 0.024669 0.315 0.752
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.7615 on 1162 degrees of freedom
## Multiple R-squared: 8.563e-05, Adjusted R-squared: -0.0007749
## F-statistic: 0.0995 on 1 and 1162 DF, p-value: 0.7525with(df_lib,
correlationBF(direct_action, reappraisal_basic))## Bayes factor analysis
## --------------
## [1] Alt., r=0.333 : 0.07223411 ±0%
##
## Against denominator:
## Null, rho = 0
## ---
## Bayes factor type: BFcorrelation, Jeffreys-beta*Does a linear combination of reappraisal-like columns that emerged in the factor analysis predictive of mean of action columns?
df_lib %>%
ggplot(aes(x = reappraisal_like, y = direct_action)) +
geom_jitter(width = .05, alpha = .2) +
geom_smooth(method="loess") +
geom_smooth(method = "lm", linetype = "dotted")
df_lib %>%
lm(direct_action ~ reappraisal_like, .) %>%
summary##
## Call:
## lm(formula = direct_action ~ reappraisal_like, data = .)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1.0638 -0.6537 -0.1891 0.5342 2.2433
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.51430 0.09200 16.459 < 2e-16 ***
## reappraisal_like 0.11314 0.03217 3.517 0.000453 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.7575 on 1162 degrees of freedom
## Multiple R-squared: 0.01053, Adjusted R-squared: 0.009683
## F-statistic: 12.37 on 1 and 1162 DF, p-value: 0.0004527with(df_lib,
cor.test(direct_action, reappraisal_like))##
## Pearson's product-moment correlation
##
## data: direct_action and reappraisal_like
## t = 3.5174, df = 1162, p-value = 0.0004527
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
## 0.04544896 0.15915904
## sample estimates:
## cor
## 0.1026393with(df_lib,
correlationBF(direct_action, reappraisal_like))## Bayes factor analysis
## --------------
## [1] Alt., r=0.333 : 31.32176 ±0%
##
## Against denominator:
## Null, rho = 0
## ---
## Bayes factor type: BFcorrelation, Jeffreys-beta*How do the individual reappraisal items predict action? Note the caveat that we’re assuming the frequency of strategy usage is a continuous linear measure (it’s not).
df_lib %>%
lm(as.formula(paste("direct_action ~ ", paste(reappraisal_like_cols, collapse = "+"))), .) %>%
summary##
## Call:
## lm(formula = as.formula(paste("direct_action ~ ", paste(reappraisal_like_cols,
## collapse = "+"))), data = .)
##
## Residuals:
## Min 1Q Median 3Q Max
## -1.5331 -0.5433 -0.1434 0.4346 2.3583
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.4355771 0.0881546 16.285 < 2e-16 ***
## per_posreapp -0.0007294 0.0266495 -0.027 0.9782
## per_negreapp -0.0065992 0.0258421 -0.255 0.7985
## per_longterm 0.0318169 0.0234097 1.359 0.1744
## per_objective 0.0020786 0.0241064 0.086 0.9313
## per_accept -0.1553631 0.0205102 -7.575 7.33e-14 ***
## per_understand 0.0473567 0.0236916 1.999 0.0459 *
## per_seekinfo 0.1911902 0.0210027 9.103 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 0.7047 on 1156 degrees of freedom
## Multiple R-squared: 0.148, Adjusted R-squared: 0.1428
## F-statistic: 28.68 on 7 and 1156 DF, p-value: < 2.2e-16