Download Data

To download the collected data as a data set, select Collected Data → Download Data in the menu. SoSci Survey provides you with the raw data from the survey: You get a table with one row per filled-out questionnaire and one column per variable.

You can download the data at any time (also while the data collection is still running, and also several times) as a backup or for first analyses. The data stored on the server is neither deleted nor changed when downloading. With the function Collected Data → View Data Set, you can also take a look at the data directly online.

If you encounter problems when downloading/opening the data, please follow the instructions in Problem Solutions for Data Retrieval.

Important: It is likely that at some point you will need the original data – as you downloaded it from the survey server (e.g. for documentation or because an important variable was overwritten in an inattentive moment). Therefore, promptly save the original data on a separate medium and keep it in a safe place. Then create a working copy of the original file and only work with this copy. Never make changes to the original file.

Note: Every person who calls the questionnaire is assigned an interview number (CASE). If a visitor decides to leave the questionnaire right away, an “empty interview” is created, which SoSci Survey deletes by default (you can change this under Project Settings → Privacy). The numbering of the cases in the data set is therefore usually not consecutive, but has gaps. Calls of the questionnaire during development also receive a number. The number of the first “real” interview is therefore usually not 1.

The focus of SoSci Survey is the collection of data; for the statistical analysis, there is suitable specialized software. On the Pro server s2survey.net, a descriptive data analysis is available. When using the standard server and for more complex analyses (e.g. inferential statistics and multivariate methods), please use a suitable program.

• Spreadsheet software (LibreOffice Calc, Microsoft Excel, etc.)
  Such programs allow simple analyses, such as means and standard deviations. With restrictions, even correlations can be calculated. For further analyses (e.g. significance tests), however, these programs are hardly suitable.
• Statistics software (R, Jamovi, SPSS, Stata, etc.)
  These programs are specialized in the statistical analysis of data. Unfortunately, getting started with the open source software GNU R is somewhat demanding. SPSS is not exactly cheap, but available at many universities. For R, SPSS and Stata, SoSci Survey prepares the data sets with variable and value labels. For other software, the universal CSV format is a good choice.
• SQL databases (e.g. MySQL)
  If the data is needed in a database, SoSci Survey also provides the SQL code for the import. As far as possible, the labels of the variables and values are imported as well. Please note that databases only store the data – additional software is required for the statistical analysis.

First, select the file format. To do so, click the corresponding tab.

For Excel (file format XLSX), you can download the data set and a codebook that lists all variables and their values. For the data set, you can set the following options:

• Output of responses
  Shall the answers to closed questions (e.g. selection questions) appear in the table as numeric codes (1, 2, 3) or as text (tea, coffee, cocoa)?
• Missing values
  Shall responses that are not an answer (nothing selected or option “don't know”) appear as codes in the table (e.g. -9 if no option was selected), or shall the corresponding cells of the table remain empty? Means are easier to calculate in Excel if missing values are removed from the data set.
• Include variable labels
  If you activate this option, SoSci Survey inserts a second row with the labels of the variables (e.g. “favorite drink”) below the first row with the technical variable names (e.g. “AB01”). The actual data then only starts from the third row.

Note: Sometimes, Excel shows a message when opening the data set that the file is damaged. Select the option that Excel shall open the file anyway. This warning appears in some Excel versions, but not in others, and especially when the option “Include variable labels” is activated. The data is stored correctly regardless of the warning.

CSV stands for “Character Separated Values”. The values are simply written one after the other into a text file and separated by a predefined character (e.g. comma, semicolon or tab). A CSV file can also be opened in a normal text editor (Notepad++, SciTE or Windows Notepad) to check the content. However, you cannot process the data further there.

The CSV format is a standard format that almost any spreadsheet or statistics program can open. The disadvantage of this format is that it does not allow labels for variables or values (response codes) – therefore, you should select the respective format for the analysis in SPSS or GNU R.

You can define which delimiters and which enclosure are used in the CSV file. However, you will rarely need to change this setting.

The option Shorten long texts is a relic from the time when SPSS 11.5 crashed when importing long texts. Normally, you should not need this option.

• Data import in OpenOffice
• Data Import in Microsoft Excel

The commercial statistics package is fairly widespread. Therefore, SoSci Survey offers the possibility to prepare data for SPSS directly with labels. You can choose between 2 variants.

• Complete import from file
  This is the convenient way. SoSci Survey creates an SPS file (SPSS syntax) that contains all variables, data and labels. However, this variant may cause problems if you use SPSS 16 (or older) and the participants have answered open questions with long texts.
• Normal import from CSV file
  In this variant, you have to download two files: the data set as a CSV file and the import syntax as SPSS syntax (SPS).

Important: For the CSV import, always download the data set and the import syntax together. If the two files are downloaded at different times, they may not match (e.g. because new variables were added or because some variables were not yet included the first time due to a lack of data).

• Import of Data in SPSS (comfortable) – default
• Import Data into SPSS (via CSV) – if the default import does not work due to overlong texts
• Data import in SPSS (manually from CSV) – not recommended

For Stata, download the data set and the import script. Then you can use the import script to import the CSV file including variable labels.

You may use the open source R Project for Statistical Computing for free, but that does not mean that everyone can use it. This software often supports the most advanced statistical methods, but using it requires a certain learning effort.

Note: R Base only supports one kind of missing values (NA). Depending on the setting for Residual responses, the distinction between an unanswered selection question and an explicit “don't know” is lost.

To import the data into GNU R, you have to download two files: a script for the import and the data set itself as a CSV file. Please note that the CSV file is formatted slightly differently when GNU R is selected as the file type than when CSV was selected.

Important: Always download the data set and the script together. If the two files are downloaded at different times, they may not match (e.g. because new variables were added or because some variables were not yet included the first time due to a lack of data).

• Import Data to R

Alternatively, you can Activate API Import. You then get one line of R script that you simply run in R or RStudio to import the data.

If you want to import the collected data into an SQL database, SoSci Survey provides the corresponding SQL commands. The labels for variables and values are transferred – as far as possible in SQL.

Further down, you will find more tabs. Here you can define which records and which variables are downloaded.

If you only want to retrieve part of the data, please select Collected Data → Filter Criteria for Valid Cases. There you can filter the cases by different criteria.

Note: As soon as the questionnaire is in the field, interviews from the questionnaire development are no longer downloaded by default (you can change this under Filter Criteria for Valid Cases). An interview is recognized as “questionnaire development” if it was called directly from the project administration () or if it was conducted in the pretest.

• Variables selection
  • Any data stored in the data set, including those from deleted questions/items
    This downloads all variables, i.e. also variables that were never asked, and variables whose questions or items were deleted afterwards. Activate this option if you have unintentionally deleted an item after the data collection. The variables of deleted questions and items use the name “DEL” with a consecutive number.
  • All variables that have been defined in the list of questions
    This includes all variables for which an entry exists in the Variables Overview.
  • All variables with data in any data set (incl. questionnaire development/testing)
    Variables that have never been asked in the questionnaire are hidden here.
  • Variables that contain any data within in the downloaded data
    This hides variables for which there are no answers in the downloaded data set. This would, for example, also exclude a text input field in which no answer was typed in any of the selected interviews.
  • Variables that show (at least two) different values within the downloaded data
    This hides variables that have the same code across all cases, for example -9 for “not answered” throughout.
• Sort variables
  • Default (like in the list of questions)
    The variables are output in the order in which they appear in the Variables Overview.
  • Internal (order of first use)
    The variables are sorted in the order in which they were first asked in the questionnaire. If the data set was deleted before the start of the actual data collection and if the questionnaire does not use filter questions, this order corresponds to the order in the questionnaire.
• Time spent per page and other time variables
  SoSci Survey stores how long a participant needed to fill out the individual pages. These variables TIME001 etc. are downloaded by default (see Completion Times).
• Download data quality parameters
  The data set is extended by an index for the completion speed (TIME_RSI) and by information on the proportion of missing answers (see Quality Indicators).
• Download mailing address entries (only in privacy mode “individual-related”)
  If you use the serial mail function with personal address entries, you can have the contact data of the address entries belonging to a case shown in the data set.

In multilingual survey projects, you can select which language is used for the labels of the variables or values. Please note that the descriptions of questions can only be specified in one language, while, e.g., the labels of answer options vary with the language.

Once you have selected a file format, you will find one or more buttons for downloading the data at the bottom (depending on format and settings). The file is delivered like an ordinary download – you should first save the file to your hard disk and not open it directly.

Important: After the click, SoSci Survey first prepares the file internally. Please wait until the download starts (for very large data sets, this may take a few minutes) and do not click the button again, otherwise it will take even longer.

Opening/importing the downloaded data is explained in the chapters linked above.