Any comments on part of our thinking in a survey data analysis stage would
be most welcome.
Our survey data zero order correlations show positive relationships between
our respondent's:
1. 'level of education' and 'trust in other people' (more educated
associated with more trust)
2. 'trust in other people' and 'community organisation membership' (more
trust associated with more membership)
But also show:
3. NO correlation between 'level of education' and 'community organisation
membership.'
All 3 variables are dichotomous categorical (education = 1. Below degree 2.
degree and higher - Trust = 1. No can't be too careful 2. Yes I trust
others - membership = 1. Not a member 2= Yes a member). We have used
polychoric correlations for these tests.
Our question is:
There has been prior debate about whether trust in others leads trusters to
join community organisations or whether joining organisations promotes
higher levels of trust in the joiners (or both). Are we correct in deciding
on the basis of our 3 findings listed above that:
*** joining 'causes' trust and not that trust 'causes' joining?
This is on the basis that:
we assume that education is the IV and trust the DV (Education --> Trust)
and our data show that education is not associated with organisation membership
Therefore if trust led to joining organisations then trust would be an
intervening variable between education and joining (EDUC-->TRUST-->JOINING)
and we would therefore expect to see a bivariate correlation between EDUC
and JOINING.
However as we do not find this correlation we must conclude that both EDUC
and JOINING 'cause' TRUST and so resolve the direction of the arrow between
TRUST and JOINING by this use of the EDUC variable.
I hope this makes sense to you and apologies for the length of the post - we
simply wish to check our thought processes on this one!
Thank you
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|