deletions | additions       diff --git a/Writing Software.tex b/Writing Software.tex  index 1bbf8b0..16bb1ec 100644  --- a/Writing Software.tex  +++ b/Writing Software.tex 
...
In Figure \ref{fig:write1} we explore the breakdown of the answers as a function of career stage. The answers vary slightly between the three groups. One striking result is that the answers of the "early" and "late" career stages closely resemble each other. The reason for such a trend may be that students follow their advisors' recommendations on software practices. In both of these groups, $\sim30\%(\pm3\%$) predominantly write their own software, while $\sim11\%(\pm2\%$) mostly use software written by others. In contrast, a larger portion of postdocs write their own software: $39\%\pm3\%$.   We further consider the breakdown of answers as a function of research area in Figure \ref{fig:write2}. The groups are not fully independent because participants were allowed to choose more than one research area. In all groups the largest portion of astronomers, $\ge50\%$, use both software written by others and write their own. Researchers working in theory and instrumentation are more likely to primarily depend on their own software (42$\pm4%$ and 38$\pm5\%$, respectively) than planetary and observational astronomers (32$\pm7\%$ and 29$\pm2\%$, respectively). The latter two groups are also more likely to use software written by others, 17$\pm5\%$ and 12$\pm1\%$ for planetary and observational researchers, respectively, versus 6$\pm1\%$ and 5$\pm2\%$ for theory and instrumentation.Finally, we break down the answers by country of the researchers' home institution as shown in Figure \ref{fig:write3}. These and the following plots by country are more difficult to interpret because they only contain information about the researchers' current institutions rather than their institutional history. All countries show similar trends with $\ge50\%$ of astronomers choosing the "Both" option. At the extremes, the survey respondents from Germany are most likely to write their own software, $38\pm5\%$, and the respondents from the UK are least likely to use software written by others, $6\pm3\%$.