SQLITE3 - Why is counting 'null' records over a 100x faster than counting 'not null' records in an indexed column?

I have a table with 300,000 rows, with an indexed column being either a large text block or NULL. Currently around 250,000 rows are NULL, with the other 50,000 rows filled with text.

c.execute('SELECT count(indexed_column) FROM table WHERE indexed_column IS NULL)

This does a row count in under one second.

c.execute('SELECT count(indexed_column) FROM table WHERE indexed_column IS NOT NULL)

c.execute('SELECT count(indexed_column) FROM table)

Both of these statements take around 180 seconds each to do their counts.

Why is there such a huge speed discrepancy for identifying NULL values vs NOT NULL? Is there a faster way to count NOT NULL values?