Constructive criticism has led Chuck to place as pending his last post.
The issue is currently under investigation.

Advertisements
This entry was posted in Uncategorized. Bookmark the permalink.

9 Responses to

  1. Chuck says:

    Ok, JL and n/a

    here were the mean AFQT percentiles and color scores by family. each SIDCODE represents a family. the individuals were filtered for being the biological offspring of both household parents and being black. a large percent (> 50%) of the black kids had one or more non bio parents so the # was small. of the families for which there was a sib difference in both color and IQ (presumably because the kids were not twins), in 26 there was a positive association between darkness and IQ and in 28 there was a negative one. the overall correlation between darkness and AFQT between families in this sample was was – 0.42. So I might have been slightly off. I’ll have to look into this more. (My PSPP copy keeps crashing so I’m having trouble doing more than superficial looks.)

    http://lesacreduprintemps19.wordpress.com/?attachment_id=630

  2. n/a says:

    Chuck,

    “in 26 there was a positive association between darkness and IQ and in 28 there was a negative one.”

    Thanks. This strikes me as consistent with no or extremely minimal correlation for between sibling comparisons overall.

    • chuck says:

      N/a & JL

      Can either of you think of a better, more reliable way of decomposing the association between and within families? In terms of the race IQ debate this seems to be an important issue — one deserving of more than my sloppy analysis. Depending on the results, either a colorist hypothesis or a genetic one is imperiled.

      • JL says:

        There are formulas for within- and between-families correlations in The g Factor, p. 141. In this case:

        The between-families correlation is the correlation between the sums of the siblings’ skin color scores and the sums of the siblings’ IQ scores.

        The within-families correlation is the correlation between the signed difference between the siblings’ skin color scores and the signed difference between the siblings’ IQ scores.

        Only one sibling pair per family. No MZ twins.

        • chuck says:

          “The within-families correlation is the correlation between the signed difference between the siblings’ skin color scores and the signed difference between the siblings’ IQ scores.”

          JL, I thought as much. This is (more or less) how I calculated the between family correlation as reported above. Though, I didn’t exclude twins.

          “Only one sibling pair per family. No MZ twins”

          If there are three siblings do I pick one pair at random?
          …..

          Ok, so applying this method to the sample in the attached excel file above, excluding apparent twins (kids that had the same scores), and comparing the last to the first sib in all instances — how do you randomly pick pairs in excel? — the between correlation was -0.38 and the within correlation was 0.02.

          Now, how would we do this in SPSS? If you can help me figure this out, I’ll repeat the above using more inclusive groups: compare the relation in families with full sibs to that in families with halfsibs/ look at outcomes/ look at Hispanics and so on. If this holds this could utterly smash colorism — and my latest attempt at race realism debunking.

          • JL says:

            Yeah, I would pick the sibling pairs randomly, but I’m not familiar with how these things are coded in the NLSY data, so I can’t really begin to think how to do it in SPSS.

            You could pinpoint MZ pairs by looking at concordance of skin color and other biometric traits, but that’s probably too much work (that would enable behavior genetic analyses though).

          • Anonymous says:

            You can precisely identify MZ twins and relationships — it’s just not easy because you have to match ID numbers. To get an idea of what I’m taking about, here’s a sample variable from the family roster section:

            “S13539.00 [HHI_RELY.01] Survey Year: 2002
            PRIMARY VARIABLE

            HHI, HH MEMBER 01 RELATIONSHIP TO R (NUMERIC) (ROS ITEM)

            COMMENT: People living in the household, RELY
            HH member’s relationship to R (numeric)

            0 0 Identity
            144 1 Wife
            309 2 Husband
            4751 3 Mother
            358 4 Father
            32 5 Adoptive mother
            1 6 Adoptive father
            3 7 Step-mother
            18 8 Step-father
            8 9 Foster mother
            0 10 Foster father
            17 11 Mother-in-law
            4 12 Father-in-law
            131 13 Sister (FULL)
            98 14 Brother (FULL)
            28 15 Sister (HALF – Same mother)
            1 16 Sister (HALF – Same father)
            0 17 Sister (HALF – don’t know)
            18 18 Brother (HALF – Same mother)
            0 19 Brother (HALF – Same father)
            0 20 Brother (HALF – don’t know)
            3 21 Sister (STEP)
            1 22 Brother (STEP)
            1 23 Sister (ADOPTIVE)
            0 24 Brother (ADOPTIVE)
            0 25 Sister (FOSTER)
            0 26 Brother (FOSTER)
            0 27 Brother-in-law
            1 28 Sister-in-law
            126 29 Maternal Grandmother
            12 30 Paternal Grandmother
            1 31 Social Grandmother
            0 32 Grandmother (don’t know or refused)
            10 33 Maternal Grandfather”

            Now, you can use these types of variables to create kinship links … but it would take quite a bit of computing. It’s nothing that I could do alone. Here was from the NLSY handbook:

            “The round 1 household roster further established the relationship of each person in the household to the youth and to each other. Follow-up questions verified the exact relationship; see Figure 1 for definitions of relationships. For example, if the household informant identified an occupant only as a “mother,” an additional question asked the household informant if this person was a biological, adoptive, step-, or foster mother. The survey collected the same type of information for a person only identified as a “father.” If occupants were listed as half siblings, the interviewer questioned the household informant on whether they shared a biological mother or a biological father. Another set of questions determined if full siblings whose reported birth dates differed by a month or less were multiple births; if they were the same gender, the household informant was asked if they were identical or fraternal twins. In addition, a person listed only as a “grandmother” was further identified as a maternal, paternal, or social grandmother. The survey solicited similar information for a household occupant listed only as a grandfather, a great-grandmother, or a great-grandfathe”
            http://www.nlsinfo.org/nlsy97/97guide/family.htm

          • Chuck says:

            Anyways, for all practical purposes we are left with the following variables — which I have been using — to identify relatives:

            (1)R12053.00 [CV_YTH_REL_HH_CURRENT] Survey Year: 1997 PRIMARY VARIABLE

            RS RELATIONSHIP TO HOUSEHOLD PARENT FIGURE

            Relationship of the parent figure(s)/guardian(s) in household to nthe youth as of the survey date. Note: This variable was renamed in round 2; the round 1 name was
            CV_HH_REL_BIRTH.

            4395 1 Both biological parents
            991 2 Two parents, biological mother
            214 3 Two parents, biological father
            2531 4 Biological mother only
            295 5 Biological father only
            103 6 Adoptive parent(s)
            44 7 Foster parent(s)
            197 8 No parents, grandparents
            114 9 No parents, other relatives
            69 10 Anything else
            ——-
            8953

            Refusal(-1) 0
            Don’t Know(-2) 0
            Invalid Skip(-3) 31

            And

            (2)

            R11930.00 [SIDCODE] Survey Year:1997 PRIMARY VARIABLE HOUSEHOLD IDENTIFICATION CODE

            COMMENT: HOUSEHOLD IDENTIFICATION CODE

            0 0
            1149 1 TO 999
            1194 1000 TO 1999
            1208 2000 TO 2999
            1204 3000 TO 3999
            1213 4000 TO 4999
            1205 5000 TO 5999
            1226 6000 TO 6999
            585 7000 TO 7999
            0 8000 TO 8999
            0 9000 TO 9999
            ——-
            8984
            …….

            (Apparently, 73% of Backs in this sample did not live with both biological parents. Of those that did, not all had color ratings.)

            If you have a better Idea of how to look at this, let me know.

            When I break results down by family this is what it looks like:
            http://occidentalascent.wordpress.com/?attachment_id=6463

  3. JL says:

    I did not realize that they used a color chart for skin color ratings in the NLSY. The reliability issue I mentioned earlier is moot then, unlike apparently in the Add Health.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s