No, assuming [-6,13] is a 95% confidence interval on a normal distribution with mean of 3.5 (substitute your own reasonable assumptions about distribution and sample size), it's plausibly something like 3x more likely that the effect is positive.