Recoding multiple columns using "long" data

[mfherman]

@tiernan

For some reason the group_by %>% unique_id %>% spread pattern is really unintuitive to me, so every time I reach this point in a project I get stuck.

For sure -- it seems like if you can gather() from wide to long you should be able to spread() back to your original wide data without adding anything. There's a good post here that helps to clarify why spread() needs unique ids to work:

How to spread columns with duplicate identifiers?

Also some more discussion here about the same issue:

Spread- why errors

Many times I get the following error message when I use the spread function: "Error: Duplicate identifiers for rows" and inevitably have to use dcast. Why does dcast work and spread kick back errors when I use the same data set?