acitelli <- read.csv("acitelli.csv")We’ll use the package dplyr. The dplyr package contains the functions for all of the data cleaning verbs: filter(), mutate(), rename(), arrange(), select(), summarize(), and group_by(). You can find a cheat sheet for dplyr here.
#install.packages("dplyr")
library(dplyr)filter()First, let’s filter cases. We can make a dataset of men only. Notice that we used a double equal sign, ==, instead of single, =. When you want to ask if something is equal to some value or another variable, that is, you want to use equal in a logical statement, you need the double equal. You can also use > < >= <= &, which means AND, and finally, | which means OR.
menOnly <- filter(acitelli, gender == 1)We could also use the pipe, %>%.
menOnly <- acitelli %>%
filter(gender ==1)We can save this new data set in our files as a csv. This code will save in the same directory where your .Rmd file is saved. You could give a more specific file path.
write.csv(menOnly, "men.csv")How about only the men who are above the median for Yearsmar. First, find the median years married.
#use a function in the mosaic packageThen, filter for men above that cut off point.
mature_hus <- menOnly %>%
filter(Yearsmar > -1.089)Instead of first finding the median with favstats(), we could ask for the median inside of filter() with the Base R function, median(). Base R has all of the descriptive stats functions you’d expect, mean(), sd(), cor(), but be careful because if you have missing data you’ll have to add rm.na = TRUE as an argument to the function. The syntax also differs from mosaic.
mature_hus <- menOnly %>%
filter(Yearsmar > median(Yearsmar))mutate()Let’s add a new categorical variable that marks the median split on Yearsmar. After you create it, take a look at it.
menOnly <- menOnly %>%
mutate(mature_hus = Yearsmar > median(Yearsmar))How would you get the frequencies on this variable?
#frequenciesNow for a sanity check, how would you get the descriptive stats split by mature and non-mature husbands?
#descriptives splitrename()We copy a variable and give it a new name with a function you already know, mutate().
menOnly <- menOnly %>%
mutate(old_hus = mature_hus)We can rename a variable without creating a new one with rename(new_name = old_name). This is handy if you forget to name variables in Qualtrics!
menOnly <- menOnly %>%
rename(wise_hus = old_hus)We can rename a bunch at the same time. This is handy if you forget to name variables in Qualtrics!
menOnly <- menOnly %>%
rename(self_positivity = self_pos,
other_positivity = other_pos)case_when()Let’s say we want to take gender, which is currently effects coded (men = 1 and women = -1) and make it a dummy variable. We can use the ifelse() function in combination with mutate() to achieve this.
acitelli <- acitelli %>%
mutate(man = ifelse(gender == 1, 1, 0))But what if we wanted to slice up years married to create a string variable that indicated newlyweds, early marriage, and mature marriages? We could nest ifelse() statements, but a better idea is to use the case_when() function.
acitelli <- acitelli %>%
mutate(married = ifelse(Yearsmar < -7, "newlywed",
ifelse(Yearsmar > -7 & Yearsmar < 0, "early marriage", "mature marriage")))These nested ifelse() functions can get out of control. So case_when() to the rescue.
acitelli <- acitelli %>%
mutate(married = case_when(Yearsmar < -7 ~ "newlywed",
Yearsmar > -7 & Yearsmar < 0 ~ "early marriage",
Yearsmar >= 0 ~ "mature marriage"))arrange()First, you should know that you can sort in the viewer by clicking the (faint) arrows just to the right of each variable name. Give it a try. It’s often handy to have a sort command in your code, and/or you might want to sort by more than one variable.
head(acitelli)## cuplid Yearsmar gender self_pos other_pos satisfaction tension simhob
## 1 3 8.202667 -1 4.8 4.6 4.000000 1.5 0
## 2 3 8.202667 1 3.8 4.0 3.666667 2.5 1
## 3 10 10.452667 -1 4.6 3.8 3.166667 4.0 0
## 4 10 10.452667 1 4.2 4.0 3.666667 2.0 0
## 5 11 -8.297333 -1 5.0 4.4 3.833333 2.5 0
## 6 11 -8.297333 1 4.2 4.8 3.833333 2.5 0
## man married
## 1 0 mature marriage
## 2 1 mature marriage
## 3 0 mature marriage
## 4 1 mature marriage
## 5 0 newlywed
## 6 1 newlywedSay we want to take a peak at the women with the bottom 6 self_pos scores.
acitelli %>%
arrange(gender, self_pos) %>%
head()## cuplid Yearsmar gender self_pos other_pos satisfaction tension simhob
## 1 160 8.7026667 -1 3.2 3.8 3.333333 4.0 0
## 2 52 13.1193333 -1 3.4 3.8 3.833333 2.0 1
## 3 441 0.1193333 -1 3.4 4.4 4.000000 3.0 0
## 4 70 11.3693333 -1 3.6 4.4 3.833333 1.5 0
## 5 116 4.7860000 -1 3.6 4.2 2.333333 4.0 0
## 6 178 -7.0473333 -1 3.6 3.6 2.666667 3.0 0
## man married
## 1 0 mature marriage
## 2 0 mature marriage
## 3 0 mature marriage
## 4 0 mature marriage
## 5 0 mature marriage
## 6 0 newlywedWe could also save the arranged dataset.
acitelli <- acitelli %>%
arrange(gender, self_pos)
head(acitelli)## cuplid Yearsmar gender self_pos other_pos satisfaction tension simhob
## 1 160 8.7026667 -1 3.2 3.8 3.333333 4.0 0
## 2 52 13.1193333 -1 3.4 3.8 3.833333 2.0 1
## 3 441 0.1193333 -1 3.4 4.4 4.000000 3.0 0
## 4 70 11.3693333 -1 3.6 4.4 3.833333 1.5 0
## 5 116 4.7860000 -1 3.6 4.2 2.333333 4.0 0
## 6 178 -7.0473333 -1 3.6 3.6 2.666667 3.0 0
## man married
## 1 0 mature marriage
## 2 0 mature marriage
## 3 0 mature marriage
## 4 0 mature marriage
## 5 0 mature marriage
## 6 0 newlywedWhat about the top 6? We can use the desc() function inside of arrange().
acitelli %>%
arrange(gender, desc(self_pos)) %>%
head()## cuplid Yearsmar gender self_pos other_pos satisfaction tension simhob
## 1 11 -8.297333 -1 5 4.4 3.833333 2.5 0
## 2 98 -9.214000 -1 5 4.2 4.000000 2.0 1
## 3 114 12.619333 -1 5 3.4 3.666667 2.5 0
## 4 127 3.619333 -1 5 4.6 3.833333 2.0 0
## 5 135 7.786000 -1 5 5.0 4.000000 1.5 0
## 6 177 11.619333 -1 5 5.0 4.000000 1.0 1
## man married
## 1 0 newlywed
## 2 0 newlywed
## 3 0 mature marriage
## 4 0 mature marriage
## 5 0 mature marriage
## 6 0 mature marriageselect()Save a smaller subset of variables.
small <- acitelli %>%
select(cuplid, gender, satisfaction, self_pos)We can also save everything but some variable(s).
no_tension <- acitelli %>%
select(-tension)summarize()acitelli %>%
summarize(mean = mean(satisfaction),
sd = sd(satisfaction),
min = min(satisfaction))## mean sd min
## 1 3.60473 0.4964205 1.166667group_by()We can split the file and view results grouped by some variable.
acitelli %>%
group_by(gender) %>%
summarize(mean = mean(satisfaction),
sd = sd(satisfaction),
min = min(satisfaction))## # A tibble: 2 x 4
## gender mean sd min
## <int> <dbl> <dbl> <dbl>
## 1 -1 3.591216 0.5300260 1.500000
## 2 1 3.618243 0.4617875 1.166667You can use group_by() to create aggregated variables, this is handy if you have nested data. We actually do have married couples here, so let’s create a dyad mean tension variable.
acitelli <- acitelli %>%
group_by(cuplid) %>%
mutate(tension_mean = mean(tension)) %>%
ungroup()
#this last command is not entirely nessesary, but good practiceWe now seen our first pipelines, using group_by(). Now we can make a pipeline of many of the commands I did above. The last thing I do is drop useless gender variable, because the resulting dataset if all men.
mature_hus2 <- acitelli %>%
filter(gender == 1) %>%
mutate(wise_hus = Yearsmar > median(Yearsmar)) %>%
rename(self_positivity = self_pos,
other_positivity = other_pos,
personID = cuplid) %>%
arrange(wise_hus) %>%
select(-gender)Save a dataset of women who are perceiving above the mean tension, and drop the simhob variable.
#above the meanWhat are the couple ID’s of the couples with the lowest 3 average satisfaction scores?
#3 lowest