adding a new column to pandas slow without reset_index

Question

I have data1 which contains about 20,000 rows:

data2=data1[data1.column=="ABC"] #about 100 rows

data1["temp"]=1  #takes no time even though it has 20k rows

data2["temp"]=1 #takes a second!

If I call the following, it executes quickly.

data2 = data2.reset_index(drop=True);
data2["temp"]=1

What am I missing here? I clearly don't understand what index thing is for in pandas and how to use it.