This paper presents results from work concerning development of the installation technique of lime-cement columns for ground improvement by deep mixing, as a part of the work of the Swedish Deep Stabilization Research Centre. A second field test was performed in order to test the findings from a previous work published by Larsson et al. The same methodology was adopted as in the previous work. However, the experimental programme was somewhat modified. Statistical analysis of variances, ANOVA, was used to investigate the influence of a number of factors in the mixing process: retrieval rate; number of mixing blades; rotation speed; air pressure in the binder tank; and diameter of the binder outlet hole. The analysis was performed with respect to the stabilisation effect and the coefficient of variation evaluated from hand-operated penetrometer tests on excavated column sections in open test pits. The retrieval rate and the number of mixing blades were found to have a significant effect. The effect of rotation speed, the binder tank air pressure, and the diameter of the outlet hole were insignificant. The two field studies illustrate the importance of performing a sufficient number of tests when studying influencing factors, because the strength is highly stochastic with large variations.