Reduction count as efficiency measure?

[Jan Skibinski]

        Could anyone explain what is a relation between
        number of reductions and execution time of any
        given Haskell function? Is there any linear
        relationship between the two on any given machine
        and compiler/interpreter?

        May I at least assume the implication:
        reduction_count1 > reduction_count2 ==> time1 > time2 ?
        when assessing efficiency of two similar algorithms
        running in Hugs?

        Jan