deletions | additions       diff --git a/survey results.tex b/survey results.tex  index 450f28a..bd32c3a 100644  --- a/survey results.tex  +++ b/survey results.tex 
...
\end{enumerate}   Answers to the first question are summarized in a graphical format in the bar plot of Figure \ref{fig:fig3}. The age distribution shows that roughly half of the respondents graduated before year 2005 (81 respondents) while the remainder graduated in 2006 and after or are still in graduate school (94 respondents: early stage researchers). As for the second question, 122 respondents identified themselves as observers, 43 as theorists, and 42 as numericists. (Note that some counts do not add up as expected because respondents could select more than one category.) Some respondents also specified a principal field of work and responses were roughly balanced between Radio, IR, Optical, and X-ray Astronomy with 10 to 20 in each group. As for the third question, 145 respondents indicated that they use data from large archives: 80 of them use it often (once a month or more), while 65 use it only sometimes (less than once a month). The remainder of the respondent group (30 respondents) never use data from large archives. Overall, the data collected in the first part of the survey shows that the survey population is rather evenly-distributed in terms of principal field of work, representing evenly various subdomains of astronomical research. Also, the vast majority of survey respondents have worked with large data and data archives. The population is skewed towards younger astronomers with about half of respondents being scientists in their early career stage (Ph.D. completed in the last 6 years or yet to be completed).  In the second part of the survey, researchers were asked to respond to two questions, one about data use practices and the other about data sharing practices (Full questions listed in \ref{sec:surveyquestions}). Results from these survey questions are summarized in tabular format in Figures \ref{fig:fig4} and \ref{fig:fig5}. Note that because respondents could choose more than one sharing method, the percentages in the last column of these figures are of \textit{repsonses}, \textit{responses},  not of \textit{respondents}. Figure \ref{fig:fig4} shows how respondents have used in their research data they have learned about reading a journal article. The most common method for researchers to obtain data from journal articles is the rather antiquated technique of manually copying and pasting (or transcribing) it from a table into another. More automated methods (such as downloading the data from an archive where it is made available or from the journal site as an e-table of ASCII data) are the second most popular techniques. Other techniques used by over half of the respondents include contacting the paper author asking directly for the data and manually extrapolating the data from a plot or graph. Results from Figure \ref{fig:fig5} show respondents' data sharing practices. The vast majority of respondents indicated that they have emailed the data to a colleague upon request. Over half of the respondents has used a ftp site or a personal website to store the data and make it publicly available. One third of the respondents have used a project-based website. The number of respondents who have used the option of an institutional archive is much lower (only about 1 in 10 respondents). Although a much rarer response, there are however some respondents who have opted in some cases not to share their data. The most common reasons for not sharing data are that: it is too much effort; the dataset is too large in size; or due to the perception that no one will want it.