Data problem (drop certain data)? to perform signrank test.

[lianlian]

I want to conduct a signrank test for the following data:
premax1 postmax1
1 1
1 1
1 1
1 1
1 1
1 1
.3333333 1
.3333333 1
.3333333 1
.3333333 1
.3333333 1
.3333333 1
.5 .6666667
.5 .6666667
.5 .6666667
.5 .6666667
.5 .6666667
.5 .6666667
.8333333 .6666667
.8333333 .6666667
.8333333 .6666667
.8333333 .6666667
.8333333 .6666667
.8333333 .6666667

....
and so on.
How could I delete the repeated variables in the data?
Make the data becomes like this:?
premax1 postmax1
1 1
.3333333 1
.5 .6666667
.8333333 .6666667


Thank you very much

[Maarten Buis]

bys premax1 postmax1 : keep if _n == 1

[lianlian]

thank you very much. I have other variables in the data. Can I generate a new variable out of it but not delete all the other observations?

[Maarten Buis]

bys premax1 postmax1 : gen byte mark = (_n == 1)

[Nick Cox]

Maarten's code assumes that a given pair of values can occur just once. But from your example blocks of identical values occur as blocks of 6. That's a different problem for which

if mod(_n, 6) == 1

is one solution. Actually any value from 0 to 5 would work.

[lianlian]

Thank you! it is not the results I want but I will work on the bysort command.

[daniela Albers]

I do not know if this might help you, but you can write: help duplicates_dlg and it gives you several options on what to do with duplicates.
Hope it helps

[Nick Cox]

The duplicates command is one I am fond of, but I don't think it helps here. What remains unclear is why observations occur in blocks of 6; why the poster only wants to work with one in each block; and whether the two values in a block could occur in another block.