Question

> imqd = read.csv("csv/quest/IMQ.csv")
> demod = read.csv("csv/DEMO.csv")
> mcqd = read.csv("csv/quest/MCQ.csv")
> 
> length(demod)
[1] 145
> length(demod[[1]])
[1] 9965
> length(mcqd)
[1] 168
> length(mcqd[[1]])
[1] 9493
> length(imqd)
[1] 5
> length(imqd[[1]])
[1] 9965
> 
> mydata = merge(imqd, demod)
> length(mydata)
[1] 148
> length(mydata[[1]])
[1] 9965

So far, so good. But, if I try to merge mcqd with anything, I lose most of my rows, even though the data looks good to me.

> intersect(intersect(names(imqd), names(mcqd)), names(demod))
[1] "X"    "seqn"

> finaldata = merge(mydata, mcqd)
> length(finaldata)
[1] 314
> length(finaldata[[1]])
[1] 18

Why are there only 18 rows now?

If you want to play along at home, you can get the csv files here.

Answer 1

merge is trying to return only those rows which match on each of the common columns.
Looking at MCQ.csv, we see that the 20th row starts off:

"19",20,2,NA,

merge will not use this row. Both of the common columns, the first and the second columns, do not match for every file. The pk of the data to merge on is clearly seqn. So, we can simply use the by argument to merge:

> finaldata = merge(mydata, mcqd, by="seqn")