The Trail Making Test (TMT; R. M. Reitan, 1958, 1992) is extensively used i
n research in neuropsychology and in aging, in part because it has been pos
tulated to reflect executive processes, such as planning and switching. How
ever, neurocognitive and individual-difference-based analyses of this test
are complicated because of different spatial arrangements of targets, the u
se of letters only in Version B, and potential order effects when Version A
is administered prior to Version B. The present article examines a Variant
of a TMT (called the Connections Test) that attempts to remedy these defic
iencies. A structural equation model suggested that there were no direct ef
fects of age on either the nonalternating or alternating Versions of the Co
nnections Test (analogous to TMT Versions A and B, respectively); rather, a
ll age-related effects were mediated through effects on perceptual speed. M
oreover, although the nonalternating and alternating versions were strongly
related to one another, only the latter had significant independent relati
ons with measures of higher order cognition.