[EpiData-list] problems with calculations during double entry

[epidata-list at lists.umanitoba.ca]

Hi,

I have a problem with results of calculation during double entry. During
double entry, all calculations give same results as first entry except only
for few subjects. I checked my formulae with other software and they look
OK. This bug is not systematic that’s why I have difficulties finding
errors. I don’t understand why this bug appears only for few subjects. They
have not the same characteristics.

Who can tell me what king of error it could be, because I don’t know anymore
what to looking for?

Below, calculations :

pspef
  AFTER ENTRY
    IF (sex = 1) and (height < 162.6) THEN
      pspdpef=(height * 0.073) - 5.98
    ENDIF
    IF (sex = 1) and (height > 162.5) THEN
      pspdpef=(height * 0.125) - 13.14
    ENDIF
    IF (sex = 0) and (height < 152.6) THEN
      pspdpef=(height * 0.079) - 6.79
    ENDIF
    IF (sex = 0) and (height > 152.5) THEN
      pspdpef=(height * 0.064) - 3.94
    ENDIF
    IF (sex = 1) THEN
      pssdcpef=(pspef - pspdpef) / (pspdpef * 0.18)
    ENDIF
    IF (sex = 0) THEN
      pssdcpef=(pspef - pspdpef) / (pspdpef * 0.2)
    ENDIF
    psppef=(pspef/pspdpef)*100
  END
END

pspdpef
  NOENTER
END

pssdcpef
  NOENTER
END

psppef
  NOENTER
END

Example : sex = 1 , height = 132.8, pspef = 4.90

First entry (good results)
pspdpef = 3.71
pssdcpef = 1.78
psppef = 132.1

Double entry
pspdpef = 3.71
pssdcpef = -1.15
psppef = 79.2

Thank you in advance for your help.

Marie-Pierre

Strippoli Marie-Pierre
Institut of social and preventive medicine
Department of Epidemiology and Biostatistic
Paediatric Epidemiology
Finkenhubelweg 11
3012 Bern
Tel: +41 (0)31 631 35 07
Fax: +41 (0)31 631 35 20
http://www.ispm.unibe.ch