SDA 4.1 Documentation for UPDATES

NAME

DESCRIPTION

• First there is a summary of changes by SDA version (with links to the next section).
• That summary is followed by a description of the changes by category.

• Version 4.0
  • New User Interface
  • New SDA Manager
  • Modify Underlying Architecture
  • Medians in the MEANS Program
  • Probabilities for LOGIT Program

• Version 4.0.1
  • Modify Interface for Help
  • Accessibility Option
  • Bookmark Analysis Output

• Version 4.0.2
  • Drag and Drop
  • View Variable with Right Click
  • View Has Bars for Frequencies

• Version 4.0.3
  • Percentiles in the MEANS Program
  • Minimum and Maximum in the MEANS Program
  • Codebook Has Bars for Frequencies

• Version 4.1
  • New Input Formats for Creating SDA Datasets
  • CASEID Variable No Longer Required

• Version 4.1.1
  • Data File Can Contain Non-English Text
  • Current Variable List Can Be Downloaded for Editing

• Version 4.1.2
  • TABLES and MEANS Can Output Results in CSV Format

• Version 4.1.3
  • Output of Analysis Programs Made HTML5 compliant
  • XCODEBK Output Was Improved and Made HTML5 compliant

• Version 4.1.4
  • RECODE Map Syntax More Flexible
  • Improvements to the Underlying Architecture

• Version 4.1.5
  • New and Modified Options for SDALOG Program
  • Improvements to SDAMANAGER Usage Report Generation

CHANGES TO THE INTERFACE (Version 4.0 and later)

• In Version 4.0:
  
  • The user interface was substantially modified in Version 4.0, providing a more modern look with rich, dynamic components.

  • This also allows for the simplification of the interface for beginning users by hiding more advanced options until they're needed.

• In Version 4.0.1:
  
  • A question mark now indicates that online help is available by clicking on the question mark. In previous versions of SDA, the online help was accessed by clicking on the name of the option -- a mechanism that was not clear to all users.

• In Version 4.0.2:
  
  • Drag and drop from variable tree to input text boxes
    You can drag and drop a variable from the variable tree to an input field on a form. (Note that although the variables in the tree include both the variable name and the label, only the variable name will be "dropped" into the input field.)
    In addition to "drag and drop," however, the procedure of clicking on a variable to transfer it to the "Selected" box, and then clicking a button to transfer the variable to the appropriate input field continues to function as in previous versions.

  • Right-click on variable tree to view variable description
    You can right-click on a variable in the tree to bring up a context menu where you can click on View variable to view a description of the variable's contents.
    This is a shortcut for the procedure of first selecting a variable to put it in the "Selected" box and then clicking on the "View" button.

ACCESSIBLITY VERSION

• In Version 4.0.1:
  The Version 4.0 interface was modified to include an accessibility interface that was compliant with Section 508. There is a checkbox at the top of the option screen to enter that mode.

RE-RUNNING (BOOKMARKABLE) ANALYSIS OUTPUT

• In Version 4.0.1:
  When the result of one of the analysis programs (TABLES, MEANS, etc.) is sent to your browser to be displayed, the URL in the address window at the top of your browser now contains all of the information necessary to re-run that specific analysis. There are various ways to make use of this new feature:

  • You can bookmark the URL in your browser, in order to run the same analysis later. This is a good way to document compactly the various steps of your analysis.

  • You can also e-mail the URL to someone, and the recipient can simply click on that URL in the e-mail to run the analysis. This method is easier than saving the output in a file and then sending someone the file that you have saved.

  • You can also copy and store one or several URLs in a file, perhaps with some commentary. If you put the URLs in a Word file, for example, you can define the URLs as hyperlinks. Then you can control-click a URL within the file to re-run the analysis. This is another way to describe and document the analysis you are carrying out. You can also e-mail such a file to someone else, who can then read your commentary and re-run the analyses.

  Note that the results for a restricted dataset will still be restricted to those users who have been granted access to that dataset. The analysis will not be re-run on a restricted dataset unless the user is logged in and authenticated.

NEW SDA MANAGER WEB APPLICATION (Version 4.0 and later)

• A new SDA Manager web application is part of the SDA distribution package for Version 4.0 and later. The SDA Manager consolidates all of the management functions for an SDA archive into one web interface. And all of the configuration information for an SDA archive is consolidated into one location -- a MySQL database. Previously, the management functions were handled by separate applications, and the configuration information was dispersed in various files across the system.

• The SDA Manager webapp also supports the configuration of personal user workspaces. These user workspaces enable analysts to create and store recoded and computed variables in their own private storage areas.

• The SDA Manager also allows an archive manager to easily configure dataset-level access control for specifying which users can access which datasets.

DOWNLOAD AND EDIT VARIABLE LIST (in Version 4.1.1)

• The current variable list can now be downloaded in the SDA Manager. That list can then be edited (further) and re-uploaded for the selected study. After re- creating the online codebook, the revised variable list will be in effect for the variable tree used by the analysis programs and for the headings in the online codebook.

NEW INPUT FORMATS FOR CREATING SDA DATASETS

• In Version 4.1:
  • The SDA Manager now accepts new data formats, in addition to a text data file and a DDL metadata file, as input for creating SDA datasets. The following formats are now accepted:
    • An SPSS system file (for example, x.sav)
    • A Stata system file (x.dta)
    • A comma-separated values file (x.csv) (with variable names in the first row)
    • A tab-separated values file (x.tsv) (with variable names in the first row)

  • A variable named CASEID is no longer required for creating an SDA dataset. The new data formats now accepted (SPSS, Stata, etc.) do not always have a CASEID variable, so this requirement was dropped for all SDA datasets. However, if you define a CASEID variable, it is still used for some checking, as when new variables are added to a pre-existing SDA dataset.

• In Version 4.1.1:
  • Non-English text can now be included in character data variables. The text should be encoded in UTF-8. This works best when data input comes from CSV or TSV files.

IMPROVEMENTS TO USAGE REPORT GENERATION

• In Version 4.1.5:
  The "Generate Usage Report" tab added an option to use the various -c options (for displaying client addresses). A date picker widget was added to choose date ranges. An option was added to filter client addresses. Wildcards can now be used with client addresses and dataset names.

CHANGES TO ANALYSIS PROGRAMS (after Version 3.5)

• ALL PROGRAMS (in Version 4.1.3)

  Output made HTML5 compliant
  The HTML output of the various analysis programs was modified to be compliant with the HTML5 standard. However, this does not affect the appearance of the output in any significant way.

• MEANS (in Version 4.0)

  Calculate a median for each cell
  An option was added to calculate and display the median of the dependent variable, in addition to the mean. If there is a very large number of cases and table cells, there may not be enough memory to calculate the exact median for all the cells of the table. In that case an estimate of the median, called the remedian, will be computed for some of the cells. An asterisk next to the median statistic indicates that it was estimated by the remedian.

• MEANS (in Version 4.0.3)

  Calculate a percentile for each cell
  An option was added to calculate and display a specified percentile (from the 1st to the 99th percentile) of the dependent variable in each cell. Similar to the median, an estimate for each percentile may be computed and indicated by an asterisk, if the there is a very large number of cases and table cells.

  Display the minimum and/or maximum value for each cell
  The minimum and/or maximum value of the dependent variable for the cases in each cell of the table can be displayed.

• TABLES and MEANS (in Version 4.1.2)

  Output results as a CSV file
  SDA includes a new option in TABLES and MEANS to output statistical results in CSV format for importing into other applications (such as Excel) for creating custom charts. CSV files can also be useful for preparing tables for inclusion in manuscripts.

• LOGIT (in Version 4.0)

  Contribution to probability of '1' rather than '0'
  An option was added to convert the logit or probit coefficients to the scale of a probability, to show how much each independent variable contributes to an increase in the probability that the dependent variable is predicted to be '1' rather than '0'.

• VIEW/DESCRIBE (in Version 4.0.2)

  Bars to show relative frequencies
  When a frequency distribution is displayed (for variables with 256 categories or fewer), each category is now displayed with a horizontal bar to show the relative size of the percent of cases falling into each category.

• RECODE (in version 4.1.4)

  Recode map made more flexible
  Recode rules for character missing data and system missing can be combined with other recode rules on the same line.

CHANGES TO THE XCODEBK PROGRAM

• In Version 4.0.3:
  Bars to show relative frequencies
  When a frequency distribution is displayed (for variables with 40 categories or fewer -- or a number that can be specified with the XCODEBK program), each category is now displayed with a horizontal bar to show the relative size of the percent of cases falling into each category.

• In Version 4.1.3:
  Codebook output improved
  
  • The style of the output was improved and made compliant with HTML5.
  • The "frames = no" option was made the default output mode and made easier to navigate.
  • Accessibility improvements were made for those using screen readers

CHANGES TO THE SDALOG PROGRAM

• In Version 4.1.5:
  New and modified options
  SDALOG added a -r [daterange] option so output can be limited to a range of dates (or a single date). Also added a -a [client_address] option that limits output to specified hostnames or numeric IPs. Address strings can contain wildcards (*). The -s [study_name] option was modified so the name can contain wildcards (*). These options make the previous -f and -F options obsolete. The -f and -F options will no longer be documented but will not yet be removed.

CHANGES TO THE UNDERLYING ARCHITECTURE

Upgrading from SDA Version 3.5
What does this mean for organizations that already have SDA archives? What changes in the new SDA version -- and what doesn't? Many of the pieces of the current SDA architecture remain the same as before. In particular, the SDA assets that organizations have developed -- such as SDA datsets, codebooks, etc. -- are fully compatible with SDA 4.x.

The things that DO NOT CHANGE in the new architecture include:

• The SDA dataset format
• The SDA codebook format
• The SDA metadata syntax (DDL)
• All the non-CGI SDA programs (analysis programs, codebook creation program, etc.)

The things that DO CHANGE are:

• The CGI programs (hsda, hsda2, etc.) are completely replaced by a servlet-based web user interface.
• The configuration information that was previously stored in HARC files is now stored in a MySQL database. (However, a simple procedure, HARCimport, is provided in the SDA Manager to import the information from existing HARC files into the database.)

Improvements in Version 4.1.4

• The C++ analysis programs are 64-bit executables in SDA 4.1.4. (In previous versions of SDA they were 32-bit.)
• The configuration for automatically sending emails about automated nightly maintenance tasks is now done interactively via the SDAMANAGER. (Previously this configuration had to be done manually by editing an XML file on the server.)
• MySQL 8 is now supported (along with the continuing support of MySQL 5.7)
• The password encoding algorithm used by SDAMANAGER has been upgraded to the more secure bcrypt algorithm.