Galaxy Zoo Starburst Talk

Characterizing 'the 778': how do the QS objects differ from the matched/paired QC ones?

  • JeanTate by JeanTate

    Following the recommendation in Laura's post (OP in the Dealing with Sample Selection Issues thread), I've begun to characterize these 778 pairs. At the highest level - first question in the classification decision tree - here is the distribution:

       Smoo  FoD  SoA
    QS  527  250   1
    QC  509  267   2
    

    "Smoo": "Smooth"; "FoD": "Features or Disk"; "SoA": "Star or Artifact". Yes, these look similar, and a chi-square(d) test confirms that they are (ignoring the SoA): 0.905, 1 dof, probability 0.341

    Within FoD and within Smoo it's not so ("rfod" = !Eos, i.e. "No" to the question "Could this be a disk viewed edge-on?"; "rsmoo" = "Completely round" AND "In between"; "cig" = "Cigar-shaped"):

        Eos rfod
    QS  107  143
    QC   74  193
    

    Chi-square(d) is 12.9, 1 dof, probability 0.000

        cig rsmoo
    QS  126  401
    QC   68  441
    

    Chi-square(d) is 18.9, 1 dof, probability 0.000

    Relative to the control sample, the QS objects contain more edge-on spirals (per zooites' classifications) and more cigar-shaped ones (which are - very likely - predominantly a mix of unrecognized Eos and highly-inclined spirals/disk galaxies, possibly with small or no bulges).

    What about the distribution of 'BPT types', the relative proportions that are - per their position in a generalized BPT diagram - 'AGN', 'Composite', 'Star-forming', 'Weak AGN', 'Weak SF', 'Unclassifiable'? Rather different than you may have thought (next post)!

    Posted

  • JeanTate by JeanTate in response to JeanTate's comment.

    Comparing BPT type = 'AGN/Composite/SF/weakAGN/weakSF' ("BPT") with 'unclassifiable ("!BPT"), there's no statistically significant difference between QS and the control sample:

       BPT !BPT
    QS 738   40
    QC 732   46
    

    Chi-square(d) = 0.443, 1 dof, p=0.506

    Similarly, the distribution of 'AGN' vs 'Composite' is not statistically significant:

       AGN Composite
    QS 124 144
    QC  55  89
    

    Chi-square(d) = 2.49, 1 dof, p=0.115

    But that's it ... in every other (sensible) comparison I've looked at the difference is statistically significant.

    For example, here's "AGN-like" ('AGN' AND 'Composite' AND 'weak AGN') cf "SF-like" ('SF' AND 'weak SF'):

       AGN-like SF-like
    QS  303      435
    QC  210      522
    

    Chi-square(d) = 10.8, 1 dof, p=0.001

    The right-hand column ("SF-like") hides an astonishing difference, between "SF" and "weak SF":

       SF  weakSF
    QS 432    3
    QC 357  165
    

    Chi-square(d) = 157 ( 😮) , 1 dof, p=0.000

    Some of the finer differences may be new, but the main one is not. In several other posts/threads, the statistically significant "pure BPT" ('pBPT', i.e. 'AGN' AND 'Composite' AND 'SF') comparative prevalence of QS objects (cf "others", i.e. with S/N < 3 for at least one of the four emission lines) has already been noted, albeit not specifically for 'the 778'. Here are the numbers:

       pBPT others
    QS 700    78
    QC 501   277
    

    Chi-square(d) = 145, 1 dof, p=0.000

    Posted