Re: [Gretl-users] boxplot accuracy