Table 1.

Effects of treatment on performance

Dependent variable:
ParameterOptimum found logistic (1)Best solution OLS (2)No. of unique solutions Poisson (3)Mean solution OLS (4)
CT0.900*0.285***0.240***0.098
(0.418)(0.053)(0.059)(0.107)
NT0.4440.0740.103*0.351***
(0.426)(0.051)(0.040)(0.104)
CT with storage0.5840.238***0.532***0.400***
(0.350)(0.047)(0.061)(0.121)
IT with storage0.5410.133***0.344***0.271*
(0.389)(0.048)(0.054)(0.116)
NT with storage0.672*0.125**0.121*0.114
(0.328)(0.043)(0.050)(0.113)
Problem 22.200***0.479***0.0290.571***
(0.306)(0.032)(0.034)(0.063)
Problem 31.225***0.130**0.097**0.038
(0.279)(0.046)(0.034)(0.078))
Problem 40.2880.345***0.0190.064
(0.326)(0.042)(0.033)(0.064)
Problem 50.5310.270***0.0060.285***
(0.330)(0.037)(0.034)(.062)
Log(prob. order)0.0930.0420.237***0.459***
(0.174)(0.024)(0.018)(0.038)
Best pretest0.1020.068***0.025**0.046*
(0.063)(0.008)(0.008)(0.020)
Own pretest0.078***
(0.016)
Round0.191***
0.009
Round20.005***
0.000
Constant0.8651.000***3.719***4.371***
(0.514)(0.074)(0.065)0.184
Observations51451451426,214
  • Columns 1–3: unit of observation is a whole trial; column 4: unit of observation is a single solution. For columns 2 and 4, the dependent variable is solution distance [measured as log(1+ difference from optimal distance)], so lower numbers correspond to better performance.*P < 0.05; **P < 0.01; ***P < 0.001