Below is my code, which does produce a TIFF image at the dpi I need. The journal specifies that figures are in a .tiff format at 800 dpi. However, lines for the axes around the boxplot and around the blocking variable are very, very faint when I print the graph and on my computer screen when I view the image in Microsoft Office Document Imaging. The lines for the whiskers are also very faint even if I increase the lfactor=8. This problem occurs when I insert the picture into a Word document. When I look at the graph output in SAS, all the lines are thick and crisp. Also, the size of the TIFF produced is small compared to other TIFF files with the sampe dpi.
I've experienced similar problems, and I think there's probably a bug.
I'd recommend calling Tech Support, and have them open an official tech-support track, and maybe a "customer reported problem" (crp) in the defects system (or maybe make defect S0556835 into a crp if they think it sounds like the same problem).
I used the waxis=2 option and that has thickened the lines of the axes around the boxplot so that is better. However, the lines for medians within the boxes and the lines for the whiskers remain very thin in TIFF file. Looking at the graph within SAS the lines for the boxes, medians, and whiskers are all thicker and look good.
I will send this problem to Tech Support.
Yes, some individual pieces of the graph have ways to hard-code a thickness ... but then if you ever want to create output with a different device (eg, dev=png), then you have to un-hardcode all the things you hard-coded to thicker values. Plus, as you've found, some pieces of the graph don't have ways to hardcode a thickness.
Ideally, the lfactor should give you a good/easy way to control the thickness of all lines in a graph ... unfortunately it seems to still have a few issues to work out :\ It will be helpful to get some customer feedback (through tech support) to help give this problem a high(er) priority.