# How to process Binomial-Binomial data on SPSS for hypothesis testing?

There are 2 groups of 100 pregnant women: the first are malnourished and the second are well-nourished. A COHORT research is being done on them to determine if malnutrition is a risk factor for low birth weight.

Out of 100 malnourished women there are 80 LBW infants, in contrast with only 10 LBW within the 100 well-nourished group.

I used the alpha = 0.05 and make this research ONE-TAILED. I assume that they have normal distribution curve.

H0: there are no significant relationship between ... H1: there is a signif ...

P1 = 0.1 P2 = 0.8

P' = (80 + 10)/(100+100) P' = 0.45 , Q' = 0.55

SE = SQRT[ (0.45*0.55) (n1 + n2) / (n1*n2) ] SE = SQRT(99/20k) SE = -0.07036

Z = (P1-P2)/SE Z = -0.7/0.07036 Z = -9.9488

Calculated probability (by chance) from the Z table is 0.

Failed to reject H1: THERE IS A STRONG CORRELATION BETWEEN MALNOURISHED PREGNANCY AND LBW.

====================

I have inputted the data into the SPSS using the values for each variable; NUTSTATS: 1 = normal, 2 = malnourished LBW: 1 = no, 2 = yes

How do I produce the same results? What test do I need?

I'm not sure about whether this should be continued with a Chi-squared test, because I'm new to this and this is a nominal-nominal data type.

Thank you for any responses.

You can do a $\chi^2$ test, as it is equivalent to the two proportion test in a 2 by 2 contingency table. Here is a quick example in SPSS.

DATA LIST FREE / Mal LBW Weight.
BEGIN DATA
1 0 10
1 1 90
2 0 80
2 1 20
END DATA.
FORMATS Mal LBW (F1.0).
VALUE LABELS Mal 1 'Well-nourished' 2 'Malnourished'
/LBW 0 'Low Birth Weight' 1 'Normal Weight'.
WEIGHT BY Weight.
DATASET NAME Nourishment.

*Chi-square test.
CROSSTABS Mal BY LBW /STATISTICS=CHISQ.