I want to perform a multiverse analysis using the multiverse package.
I have many different filters and data transformations to take into account the multiverse analysis.
One of them is the different way to aggregate the data, that can be:
- mean
- median
- no aggregation
My first attempt is the following:
<code>inside(M, {
df <- sim_data %>%
## DATA TRANSFORMATION
mutate( y= branch( data_transform,
"none" ~ y,
"log" ~ log(y+ 0.1),
"sqrt" ~ sqrt(y),
"inverse" ~ 1/(y+ 0.1)
)) %>%
## DATA AGGREGATION
mutate( y= branch( data_aggregation,
"none" ~ y,
"mean" ~ group_by(Condition, ID) %>% summarise(mean = mean(y)),
"median" ~ group_by(Condition, ID) %>% summarise(median = median(y))
))
} )
</code>
<code>inside(M, {
df <- sim_data %>%
## DATA TRANSFORMATION
mutate( y= branch( data_transform,
"none" ~ y,
"log" ~ log(y+ 0.1),
"sqrt" ~ sqrt(y),
"inverse" ~ 1/(y+ 0.1)
)) %>%
## DATA AGGREGATION
mutate( y= branch( data_aggregation,
"none" ~ y,
"mean" ~ group_by(Condition, ID) %>% summarise(mean = mean(y)),
"median" ~ group_by(Condition, ID) %>% summarise(median = median(y))
))
} )
</code>
inside(M, {
df <- sim_data %>%
## DATA TRANSFORMATION
mutate( y= branch( data_transform,
"none" ~ y,
"log" ~ log(y+ 0.1),
"sqrt" ~ sqrt(y),
"inverse" ~ 1/(y+ 0.1)
)) %>%
## DATA AGGREGATION
mutate( y= branch( data_aggregation,
"none" ~ y,
"mean" ~ group_by(Condition, ID) %>% summarise(mean = mean(y)),
"median" ~ group_by(Condition, ID) %>% summarise(median = median(y))
))
} )
Is this code going to transform the data and THEN aggregate the data?
1