Widening Black-White IQ gap with Parental Education in the Add Health data

Some large datasets available, at this address, allows us to investigate some characteristics of the Add Health respondents. What I am interested here is the magnitude of the black-white (verbal) IQ difference at different levels of parental education. I have downloaded the WAVE I and WAVE III Public Use datasets for SPSS. In the WAVE I, the respondents’ mean age is 16, in WAVE III, 22 yrs-old.

But before running the ‘comparison of means’ program, there is some recoding to do.

Name : PA12
Label : A12 LEVEL OF EDUCATION-PQ
Value :
1 = 8th grade or less
2 = >8th grade/didn’t graduate high school
3 = Business/trade/voc. school instead high school
4 = High school graduate
5 = Completed a GED
6 = Business/trade/voc. school after high school
7 = College/didn’t graduate
8 = Graduated from college/university
9 = Prof training beyond 4-year college/univ
10 = Never went to school

From this, I have recoded the PA12 (parental) variable into a 3-categories variable. In your SPSS (or at least PSPP) program, go to “Transform”, “Recode into Different Variables…”. Then, choose your PA12 variable. Give a name and a label. Check the box “Old and New Values…”.

Here’s how it’s done. Click on “Range”, 1 through 3. Click on “Value” in the right panel labelled “New Value” and enter 1. Hit the box “Add”. Click on “Range”, 4 through 6. Click on “Value” in the right panel labelled “New Value” and enter 2. Hit the box “Add”. Click on “Range”, 7 through 9. Click on “Value” in the right panel labelled “New Value” and enter 3. Hit the box “Add”. Click on “Continue”, and then “OK”. Now you have your 3-categories variables. Your window should look like this.

SPSS Recode into Different Variable Old and New Values

Then, select “Analysis”, “Comparison of Means”, “Means”.

List of dependent variables :
ADD HEALTH PICTURE VOCABULARY TEST SCORE [AH_PVT]
CROSS-SECT/LONG PVT STANDRDIZED SCORE-W1 [PVTSTD1]
CROSS-SECTIONAL STANDARDIZED SCORE-W3 [PVTSTD3C]

List of independent variables :
Layer 1 : BWrace (see below syntax)
Layer 2 : Parental_Education [ParEduc3] (see below syntax)

Here is the table I have generated. The N row displays large sample size because weights generally inflate the original Ns. And so, we need to re-run the comparison of means program without weights to get the original Ns, and apply it in Excel, and then compute the gaps.

Add Health - Black-White verbal IQ gap by parents' education

And below is the computation of the pooled d gap by parental education level. It can be seen that the BW d gap is widening (but not linearly) with parental education.

Excel - Black-White verbal IQ gap by parents' education in Add Health

The environmental hypothesis could not predict this pattern because (1) health is better at higher levels of SES and (2) culture which also varies within groups is cognitively stimulating at higher levels of SES. The result from the Add Health data is somewhat consistent with other data showing a modest to large increase in the black-white IQ difference when the SES increases (Herrnstein & Murray, 1994, pp. 287-288; Jensen, 1998, p. 358; Murray, 1999, Figure 3; Gottfredson, 2003, Table 12; Chuck, Jan. 19, 2012). The GSS (Hu, Jan. 1, 2013) and the NLSY97 (Hu, May 11, 2013) also confirmed the pattern of an increasing IQ gap with SES level.

SPSS syntax for the above analysis :

IF H1GI6B=1 BWrace=0.
IF H1GI6A=1 BWrace=1.

RECODE PA12 (1 thru 3=1) (4 thru 6=2) (7 thru 9=3) (ELSE=SYSMIS) INTO ParEduc3.
VARIABLE LABELS ParEduc3 ‘Parental_Education’.
EXECUTE.

WEIGHT BY GSWGT1.

MEANS TABLES=AH_PVT PVTSTD1 PVTSTD3C BY BWrace BY ParEduc3
/CELLS MEAN COUNT STDDEV.

WEIGHT OFF.

MEANS TABLES=AH_PVT PVTSTD1 PVTSTD3C BY BWrace BY ParEduc3
/CELLS MEAN COUNT STDDEV.

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