This paper summarizes how the partisan influence literature assesses the re
lationship between the left-right party composition of government and polic
y outputs through a meta-analysis of 693 parameter estimates of the party-p
olicy relationship published in 43 empirical studies. Based on a simplified
`combined tests' meta-analytic technique, we show that the average correla
tion between the party composition of government and policy outputs is not
significantly different from zero. A mutivariate logistic regression analys
is examines how support for partisan theory is affected by a subset of medi
ating factors that can be applied to all the estimates under review. The an
alysis demonstrates that there are clearly identifiable conditions under wh
ich the probability of support for partisan theory can be substantially inc
reased. We conclude that further research is needed on institutional and so
cio-economic determinants of public policy.