8.4 Example Data

Chapter 7

In Chapter 7 we use data from Dunn, Aknin, and Norton (2007), who examined the relationship between spending habits and happiness using OLS regression. To capture spending habits self-reported monthly spending was categorized as being either personal or prosocial, and then summed to create a category-specific total. Measures of happiness were obtained using a 5-item ordinal scale.

The dependent variable in this analysis (GeneralHappiness) was self-reported general happiness. Participants rated their general happiness by answering the question “Do you feel happy, in general?” by selecting from five possible response options (no, rarely, sometimes, most of the time, and yes), which were then scored from 1-5. Here, higher numbers were indicative of greater happiness. For the purpose of our analysis we will dichotomize GeneralHappiness depending on whether a person was not happy vs happy.

The sample was selected to be a nationally representative sample of 632 Americans (287 males and 345 females). Participants responded as part of a larger, online survey, in return for points that could be redeemed for prizes (Dunn, 2008). No further details on sampling were available.

8.4.1 Variables

• Happy: Dichotomous variable indicating whether the subject responded “yes” when asked if they felt happy, in general.
• PersonalSpending: Self-reported dollars spent per month on (a) bills and expenses, and (b) gifts for themselves.
• ProsocialSpending: Self-reported dollars spent per month on (a) gifts for others, and (b) donations to charity.
• PersonalIncome: Participants selected their personal income category from 6 options: less than $20,000, $20,000-$35,000, $35001-$50,000, $50,001-$65000, $65,001-$80,000, $80,001+.

library("ggplot2")
dunn2008 <- read.csv("data/DUNN2008.csv")
dunn2008$Happy <- ifelse(dunn2008$GeneralHappiness == "yes", 1, 0)
dunn2008$PersonalSpending  <- dunn2008$PersonalSpending/100
dunn2008$ProsocialSpending <- dunn2008$ProsocialSpending/100
dunn2008$PersonalSpending_star <- as.numeric(scale(dunn2008$PersonalSpending, scale = FALSE))
dunn2008$ProsocialSpending_star <- as.numeric(scale(dunn2008$ProsocialSpending, scale = FALSE))
dunn2008$Income <- dplyr::recode(dunn2008$PersonalIncome, 
                            "20001-35000" = "20-35K", 
                            "35001-50000" = "35-50K", 
                            "50001-65000" = "50-65K", 
                            "65001-80000" = "65-80K", 
                            "80001andup" = "> 80K", 
                            "less20000" = "< 20K")
inc_lev_order <- c("< 20K","20-35K", "35-50K","50-65K","65-80K","> 80K")
dunn2008$Income <- factor(dunn2008$Income, levels=inc_lev_order)