I have done some correlations of 2MASS matches and non-matches with the SDSS early data release. I have focused on the SDSS in the region 352.<RA<55 and -1.0<decl<1.0. I created a subset of the v3 release dataset via:

create table sloan_subset_55 as select cntr,scan_key,ra,decl,j_m,h_m,k_m,rd_flg,cc_flg,ext_key,ph_qual from v3_pscat where (ra between 0. and 55. or ra between 352. and 360.) and decl between -1.0 and 1.0

and then created a table of source in 2MASS which did not have an SDSS counterpart:

create table sloan_nomatch_pscat_55 as SELECT * FROM sloan_subset_55 as a WHERE NOT EXISTS (select b.* FROM sloan_primary AS b WHERE distance(b.ra,b.decl,a.ra,a.decl) < 1.5 AND b.decl between a.decl - 0.0014 and a.decl + 0.0014 AND b.ra   between a.ra - 0.002 and a.ra +.002

	2MASS sources	Non-matching sources
All	208766	4371
Contains "A"	140804	1158
"A",rd="222", cc="000"	132719	586
"AAA", cc="000"	86403	349

The first thing I looked at was the spatial distribution of various selections from the database.

All sources

rd_flg='222' sources

rd_flg!='222' sources

Why does the hemispheric overlap show up in the non-222 sources. This would suggest many are artifacts (and thus there were two chances to make them at the dec boundaries).

The complete list of non-matching sources is here.

A color-color plot of the non-matching AAA sources show that they are mostly stars and suggests they were missed by SDSS. Although there are a handfull of objects with extremely red J-H colors that do not look real.