Variables used in paper Table 

**sample 
*deriving when/ how often been a survey respondent
*inmcs1 inmcs2 inmcs3 inmcs4 – variables represent if family in corresponding data collection sweeps.

compute lonresp = -1.
if (inmcs1 = 1) lonresp = 1.
if (inmcs2 = 1) lonresp = 2.
if (inmcs3 = 1) lonresp = 3.
if (inmcs4 = 1) lonresp = 4.
if (inmcs1 = 1 and inmcs2 = 1) lonresp = 5.
if (inmcs1 = 1 and inmcs3 = 1) lonresp = 6.
if (inmcs1 = 1 and inmcs4 = 1) lonresp = 7.
if (inmcs2 = 1 and inmcs3 = 1) lonresp = 8.
if (inmcs2 = 1 and inmcs4 = 1) lonresp = 9.
if (inmcs1 = 1 and inmcs2 = 1 and inmcs3 = 1) lonresp = 10.
if (inmcs1 = 1 and inmcs3 = 1 and inmcs4 = 1) lonresp = 11.
if (inmcs2 = 1 and inmcs3 = 1 and inmcs4 = 1) lonresp = 12.
if (inmcs1 = 1 and inmcs2 = 1 and inmcs3 = 1 and inmcs4 = 1) lonresp = 13.
missing values lonresp (-1).
value labels lonresp 1'mcs1' 2'mcs2' 3'mcs3' 4'mcs4' 5'mcs1 mcs2' 6'mcs1 mcs3' 7'mcs1 mcs4' 8'mcs2 mcs3' 9'mcs2 mcs4' 10'mcs1 mcs2 mcs3' 11'mcs1 mcs3 mcs4' 12' mcs2 mcs3 mcs4' 13'mcs1-4'. 
freq lonresp.

*grouping/simplifying above variable
recode lonresp (1,2,3,4 = 1) (5,6,7,8,9 = 2) (10,11,12 = 3) (13 = 4) into lonrespg.
fre lonrespg.

*family returned a tooth.
*number of teeth .
fre tteeth.
recode tteeth (0=0) (sysmis = 0) (1 thru 10 = 1) into teeth.
variable labels teeth 'tooth returned'.
value labels teeth 0'no' 1'yes'.
freq teeth.

*summary of number of teeth returned
compute teethg = -1.
if (tteeth = 0) teethg = 0.
if (tteeth = 1) teethg = 1.
if (tteeth > 1) teethg = 2.
missing values teethg ( ).
value labels teethg 0'no tooth' 1'1 tooth' 2'2+ teeth'.
freq teethg. 

*whether family ‘issued’ at sweep 4 and whether participated and whether returned tooth
compute mcsip = -1.
if (dissued = 1) mcsip = 1.
if (dissued = 1 and teeth = 1) mcsip = 2.
if (dissued = 1 and inmcs4 =1) mcsip = 3.
if (dissued = 1 and inmcs4 = 1 and teeth = 1) mcsip = 4.
value labels mcsip 1'issued mcs4' 2'issued, sent tooth' 3'issued & in mcs4' 4'issued & in mcs4 & returned tooth'.
missing values mcsip (-1).
freq mcsip. 

*these are the valid sample 
recode mcsip (3=0) (4=1) into MCS4tooth.
value labels mcs4tooth 0'no' 1'yes'.
freq mcs4tooth. 

*type of tooth being retuned.
temporary.
select if (teethg = 1).
freq  DK LCAN LCINC LFMOLAR LLINC LSMOLAR UCAN UCINC UFMOLAR ULINC USMOLAR UNKNOWN.

*high – medium – low ‘lead’ areas
missing values lead (-999).
FREQUENCIES VARIABLES=LEAD CORINE
  /NTILES=4
  /NTILES=5
  /STATISTICS=STDDEV MINIMUM MAXIMUM MEAN MEDIAN
  /ORDER=ANALYSIS.
recode lead (19 thru 37.9500000000001 = 1) (37.950000000002 thru 48 = 2) (48.00000000001 thru highest = 3) into leadg.
var labels leadg ' amount of lead'.
value labels leadg 1'<1sd mean' 2'mean +/-1sd' 3'>1sd mean'.
freq leadg.

*TABLE 2 in paper.
*Biological
Variables: dhcsex00  M4ethnic (ddc06e00)  agemum
Recode adgmab00 (13 thru 19 = 1) (20 thru 29 = 2) (30 thru 39 = 3) (40 thru highest = 4) into agemum.


**Behavioural.
Variables: ambfeva0 dmberea0 dmhodi00  m4clsl m4cghg 

*long-standing limiting illness
do if (dmclsia0 >= 1).
compute m4clsl = 0.
if (dmclsia0 = 1)   m4clsl  = 1.
if (dmclsia0 = 1 and dmclsla0 = 1) m4clsl = 2.
end if.

*original general health status variable = dmcghea0 .

recode m4cghe (1=1) (2,3=2) (4,5=3) into m4cghg.
value labels m4cghg 1'ex' 2'vg-g' 3'f-p'.
freq m4cghg.


**environmental.
Variables: PTTY00 ddregn00 ddcnty00 dimd dhadsa00
*where live in UK.
freq  ddregn00.
missing values  ddregn00 (13).

*level of deprivation in area where live.
compute dimd =  sum(dimdscoe,dimdscow,dimdscos,dimdscon).
freq dimd.

**social
Variables: m4homeg1 doedp000 ddtots dlang dmphq dmp05s DCWRK DHTYP 

*parent highest qualification
recode dmdnvq00 dpdnvq00 (96=0) (95=0.5) (else=copy) into dmhq dphq.
compute dmphq = dmhq.
if ((missing(dmhq)) or (dphq > dmhq)) dmphq = dphq.
freq dmphq.
recode dmphq (-1=sysmis).
freq dmphq.

**parental social class (NSSEC).
compute dmp05s = dmd05s00 .
if ((missing(dmd05s00)) or (dpd05s00 > dmd05s00)) dmp05s = dpd05s00.
if (dcwrk >= 4) dmp05s = 0.
freq dmp05s.

**number of children in household.
recode ddtots00 (1 thru 4 = copy) (5 thru 13 = 5) into ddtots.
freq ddtots.

*household type (parents).
recode DHTYP0000 (1=1) (2 thru 12=2) (15 thru 24 = 3) into dhtyp.
freq dhtyp.

*number of people working in household.
recode DCWRK0000 (1=1) (2,3=2) (5=3) (4=4) (6=5) into dcwrk.
freq dcwrk.

*language spoken in home
recode  ddhlan00 (1=1) (2,3,4=2) (5=3) into dlang.

*housing in sweep 1
freq dmroow00.
recode dmroow00 (1,2,3=1) (4,5=2) (6=3) (7 thru 95 = 4) into m4homeg1.
value labels m4homeg1 1'own' 2'rent la/ha' 3'rent - private' 4'other'.
Freq  m4homeg1.

**income poverty
fre doedp000.  

*MCS data needs to be weighted.
*SPSS
*In spss the data needs to be weighted using an appropriate .csaplan file – see appropriate documentation on CLS website on how to derive these.
http://www.cls.ioe.ac.uk/page.aspx?&sitesectionid=880&sitesectiontitle=Survey+Design
*on the right-hand side of this page there is a document on how to weight data in SPSS or STATA.

*SPSS cross-tab code
CSTABULATE
  /PLAN FILE='/Users/kkkkkk/Documents/MCSdata/mcs4csp.csaplan'
  /TABLES VARIABLES= m4clsl m4cghe  M4ethnic PTTY00 ddregn00 ddcnty00 dimd dhadsa00 dmhodi00   dmberea0  m4homeg1 doedp000 ddtots dlang dmphq dmp05s DCWRK DHTYP  dhadsa00  BY mcs4tooth / CELLS colPCT
  /STATISTICS SE COUNT
  /MISSING SCOPE=TABLE CLASSMISSING=EXCLUDE.