Data storage type matters

Despite most sources tell that the storage type in stata should not matter, it is worth checking whether this is the case for your dataset. I just came across a situation where two identically constructed datasets (one stored in default type (float) and one stored in double) generated different output. Also before that i encountered a problem with person identifiers in the GSOEP if using the default data storage. If your dataset is not huge (with the GSOEP it still works quite ok) it might be worth to take the safe side and use

set type double

before you assemble your data set. This saves the data in the most precise way stata offers.

Turning Footnotes into Endnotes …

Submitting (LaTeX) written papers has some advantages: one usually does not have to change any formatting manually. LaTeX does that for us if it is instructed to do so. Journals often want the submissions to be formatted in a peculiar way. My challenge was to change footnotes to endnotes. I did remember that it was easy in LaTeX, but not how it was done, so Google had to help me …
Continue reading

Preamble when switching between OS X and Windows

A problem when working on one and the same project on different platforms (here: Windows and Mac/OS X) is that path-names differ. There are two straightforward solutions to this:

1) When defining a number of different path (e.g. one path where data is stored, one where results/output is stored), it is handy to define the paths as globals and to add an “if” condition. The platform can be detected by the local `c(os)': Continue reading