Oriana version history
Ver. 4.02 - Released 5 April 2013
- A change made in version 4 resulted in circular-linear plots of axial data only plotting one set of arrows/points/etc. rather than the two opposing bidirectional sets, as happened in version 3. Fixed.
- With Analyses|Second Order Stats, if the column containing the means was declared as Axial this was ignored; the stats were calculated as if they were angles. Fixed.
- With axial data the Hotellings test for weighted means (in the Basic Stats output) was being calculated as if they were angles. Fixed.
- In some situations when calculating the confidence intervals of a second order stats grand mean, the opposing angle (e.g. the one 180° around the circle) was reported rather than the correct one. Fixed.
- Fixed display glitch where, when weighted basic statistics are printed out, some results do not line up with their corresponding label in the first column.
- Fixed error that would occur when calculating weighted stats or doing circular-linear plots when all the weights/linear values are 0.
- Trying to calculate multisample tests when only one variable was selected resulted in an "Access violation" message. Fixed so that it won't allow multisample analysis with just one variable.
- Fixed problem calculating Hotelling probability when two samples are identical.
- Fixed the Data|Fill option so that if the minimum or maximum values for random numbers are entered as decimal numbers (e.g. 10.0) the resulting random numbers will also be decimal.
- Fixed crash that could occur when trying to do an analysis after converting a variable to a very large number of subgroups (i.e. 1000s).
- The Monochrome Graph option was being ignored in the Raw Data Plots. Fixed.
- Modified drawing of rose diagram segments so that outer edges are rounded.
- Added a “Change Tab Text” option to the Results window right-click menu.
- Fixed error that occurs when exporting results to Excel when more than one tab has the same name.
- Removed a message that appeared on graphs when the weighted mean option was selected, but the variable being plotted wasn't actually weighted.
- Fixed some problems with attaching data files to error reports.
Ver. 4.01 - Released 4 January 2012
- Fixed program crash at startup when Windows time zone information was not found or was corrupted.
- Fixed error that could occur when importing time data that contained negative values.
Ver. 4.00 - Released 10 October 2011
- New data type: lunar. Data are entered as dates but are treated as circular data based on the lunar month. This is useful when studying phenomena that might be related to lunar phase (e.g. turtles laying eggs during the full moon).
- New method of producing histograms and basic stats. Data can now be entered directly into the data editor as a frequency table, with one variable being the circular data and an associated Linear variable giving the frequency (number of observations for that circular measurement).
- New statistics:
- Hotelling's paired multisample test
- Moore's paired multisample test (nonparametric)
- Hotelling's test for weighted data
- Moore's modified Rayleigh test for weighted data
- Second order statistics:
- Grand mean angle, length and confidence limits
- Hotelling's one sample second order test
- Moore's modified Rayleigh test for second order analyses
- Hotelling's two sample second order test
- Mardia's two sample second order test (nonparametric)
- Hotelling's paired sample second order test
- Moore's paired sample second order test (nonparametric)
- Grand mean angle, length and confidence limits
- Hotelling's paired multisample test
- Added option for circular plots to plot only the data between 0-180°, but spread around the full circumference of the circle. This type of graph is used in fields such as ophthalmology (doubled-angle plots).
- Added an option on circular-linear plots to plot the weighted mean as a centroid, along with an ellipse depicting the standard deviation of the data along the two major axes of the plot. This also is used in ophthalmological studies.
- Added option to save means and mean vector lengths to a file for further second order analysis
- A new data transformation facility was added. Data entered or imported in radians can be converted to degrees for analysis. Also, linear data can be standardized and log or square root transformed.
- Added function to merge several Oriana data files into a single file
- Added option to have the degree symbol printed after the angle labels on graphs
- Added an new option for number of minor divisions on circular graphs (every 15°)
- Added plotting of the grand mean vector when subgroup means are included in histograms.
- Fixed bug that could cause inaccurate stats results (mean, R, etc.) for Day of Year and Week of Year variables. Any users using this type of data should check their results by recalculating them with this new version.
- When filling circular percentage variables with random data, the results were previously whole numbers (e.g. 1.0, 2.0, 3.0, etc.). Now changed so that random decimal portions also included (e.g. 1.23, 2.59, etc.).
- Converting circular percentage variables to linear variables would result in the angular equivalent of the circular percentage. This is now changed so the result is the actual percentage.
- Fixed bug that caused error when converting Week of Year to Day of Year when the week was week 1.
- Fixed bug that caused error when undoing certain data matrix manipulations, like sorting, while filtering is active.
- Two-variable histograms based on special circular types like DOW, MOY and Compass were being labeled with angles rather than the descriptive labels (such as “Monday”, “March” or “NE”). Fixed.
- Fixed Data|Fill Data so that the Start Row will always be the currently selected row, rather than the previously used value.
- Fixed reading of data files so that if an unknown variable type is encountered that variable will be read as a label. Previously it would cause an error and the file would not be read.
- Using the Change Layout command to convert data with a subgrouping variable to several columns of data was rounding the values to the nearest whole number. Fixed.
- Fixed problem with weighted means where missing data in the weighting variable could give extremely large values for the weighted statistics. Now any cases with missing weighting values are assumed to be unweighted.
- Circular-linear graphs of axial data were plotting two sets of points or vectors on either side of the circle, as is done with histograms, but this could be confusing. These graphs will now only plot the original data.
- Width of thicker lines on graphs are now more accurately displayed on printouts.
- Variables that had many hundreds of blank rows at the top of the column were not being included as possible second variables in the dialog boxes of 2-variable histograms and circular-linear plots. Fixed.
- If the Preferences dialog was used to customize the compass abbreviations, but some were left blank or duplicate entries were included, this would cause the program to crash at startup. Fixed.
- Fixed the data file Save As dialog to remember the last used folder.
- Fixed problem importing numeric data from dBase and Paradox database files.
- Minor bug fixes in import procedures.
- Fixed a bug that, in certain circumstances with new installations, could cause Oriana to shut down immediately after starting.
Ver. 3.20 - Released 30 September 2010
- Three new visual themes have been added: Office 2010 Blue, Silver and Black.
- Most simple information, warning and error dialog boxes now use Windows Vista/7-style Task Dialogs.
- Data import and export now supports Excel 2007/2010 file format (.XLSX and .XLSM files). Export of results can now also be done to the newer Excel format.
- New Excel import routines more reliably detect times and dates in various international formats.
- Added a Guess Column Names option to find most likely row in an Excel file that contains column names.
- Selection of worksheet to import from Excel now done with tabs at the bottom of the preview grid rather than a drop-down box as before.
- Dialogs boxes that suggest alternative variable types for imported data (e.g. circular proportional if data are between 0-1) now have a "Don't show again" checkbox.
- Automatic updating of the software at startup will now show the new features of the new version before downloading.
- When converting a variable between linear and date types, if the values appear to be years the data will now be converted to 1 Jan of that year. Previously the value would be interpreted as the number of days since 30 Dec 1899.
- When importing numeric data that could be special circular types such as circular percentage, day of year, month of year, etc., the user is asked if the data are that specialized type. If the answer is "no" Oriana used to import the data as linear. It now imports the data as angles.
- Fixed an error that would occur when label variables created from imported data were converted to numeric types.
- Fixed a problem where the decimal proportion of Circular Proportional and Circular Percentage variables were not being displayed if the Windows system was set up to use a comma for the decimal separator.
- Converting labels to Circular Proportional or Circular Percentage variable types would treat the value in the label as an angle, then convert it to a proportion or percentage (e.g. a label of "0.5" would become a proportion of 0.00139 [0.5/360] rather than 0.5). Fixed.
- Data filtering was not working correctly when filtering was based on a Circular Proportional or Circular Percentage variable type. Fixed.
- If a filtering variable had missing cases and the Greater Than or Not Equal options were in use then the cases with missing data were wrongly being included. They are now excluded.
- When opening an Oriana version 1 file the program would offer to convert it to the version 2/3 format, but it did not always do as told. It is now behaving.
- An error message would occur when the OK button was pressed on the Options|Preferences dialog box when a file with just a single variable was in use. Fixed.
- The Import Preview box was ignoring the user's last used options (like Maximum Preview Rows) unless the Update Preview button was pressed. Fixed.
- The Convert X/Y to Vector dialog box wasn't listing all numeric variables as possible sources of the X/Y data. Fixed.
- Fixed Excel export of results so that, if the Export All box is not ticked the current page will be exported, not the first one.
- Minor changes to various dialog boxes.
Ver. 3.13 - Released 4 February 2010
- Fixed a problem with calculation of weighted means which could cause incorrect results. Anyone using weighted means should install this update and recalculate their statistics.
- Modified plotting of the weighted mean vector so that user can choose between having just the mean plotted, as previously, or so that the length of the arrow reflects the length of the weighted mean vector. The length on 2-variable histograms will be the r value, scaled 0-1 like the unweighted r, and that on the circular-linear graphs will be in the native units of the weighting variable.
- Enabled the plotting of separate weighted means for each subgroup.
- Enabled the plotting of weighted means in regular (single variable) histograms.
- When weighted mean was plotted for bidirectional/axial data only one of the mean arrows was weighted - the opposing one was unweighted. Fixed.
- The statistics for the weighting variable displayed in the Results window were for all cases, even if results were based on subgrouped data. Fixed.
- Fixed an error message that could occur when calculating weighted statistics on data with a single case.
- On linear 2-variable histograms the unweighted mean was being drawn even when weighting selected. Fixed.
- Fixed manual scaling of Y axis of histograms and circular-linear graphs so that when the percentage scale is used the maximum percent can be entered in the dialog, rather than maximum frequency.
- Fixed problem when creating half and double width linear histograms with manual percentage scaling, where scales were wrong.
- Fixed automatic scaling of percentage values on the Y axis to give neater results (e.g. maximum 8% rather than 7.59%).
- If the number of bars in a histogram was set to more than 90, then the graph edited, the number of bars was reset to 90. Fixed.
- For rose diagrams with the frequency shown as area, the printed frequency table was showing the square roots of the frequencies. Fixed.
- Removed some options from the Y-Axis section of the raw data plot dialog that were not used in this type of graph.
- Removed the Print Frequency Table option from circular-linear graph dialog, as it wasn’t used.
- Fixed problem that could cause Access Violation error in data with vector pairs assigned when rows of the data matrix are deleted or inserted.
- Fixed problem causing error message when R calculated to be zero.
- Changed message on raw data plots when each symbol represents more than one observation to be clearer.
- The coloring of the confidence interval arc for inaccurate standard errors on circular diagrams with subgrouped means always reflected the inaccurate status of the entire data set, rather than for each subgroup of data. Fixed.
- Improvements to importing data to better guess the correct data type.
Ver. 3.12 - Released 15 December 2009
- Circular-linear scatterplots were always being drawn clockwise with 0º at top, even if different orientations and rotations were chosen. Fixed.
- The "Rotate Symbol Around Origin" option was being ignored in Circular-linear scatterplots. Fixed.
- Fixed error message that could occur when updating through the Check Updates menu item when run as a limited user.
Ver. 3.11 - Released 30 September 2009
- Opening some dialog boxes could result in an Access Violation error and not be displayed. Fixed.
- Main window title bar still had "3.0" as the version. Fixed.
Ver. 3.10 - Released 21 September 2009
- Added two new variables types, Circular Proportional and Circular Percentage. These are circular variables, but rather than setting and reporting the angle as values between 0°-360°, the proportion (0-1) or percentage (0-100) of the circumference of the circle are used. For example, an angle of 90° would be a proportion of 0.25 or a percentage of 25. Data are entered or imported in this format and all graphs and statistical results are labeled with these values.
- Added a “No Grid” option to the Minor Divisions option on the Angular Axis options on histogram dialogs to suppress all angular gridlines, even the ones labeled 0, 90, 180 and 270.
- Allowed the Number of Gridlines option to be set to 0 for the Circular grid on circular graphs, and for scatterplot axes.
- Basic statistics now reports standard deviation, standard error & confidence interval for time variables are reported as a time as well as an angle. The new variable types are also reported as both angles and proportion/percent.
- Changed the “Class width” option on histogram dialogs, which set the width of bars on histograms, to “Number of bars”. This lets users specify the exact number of bars. The equivalent class width is also displayed in the native format (either degrees, time, proportion or percentage).
- Added an option to circular graphs to draw a circle showing the critical value of the Rayleigh test. The length of the r vector can then be compared to this circle to assess its significance.
- Fixed the Edit Frequency dialog box to not allow invalid angles to be entered.
- Fixed an Access Violation error that could occur when attempting to delete a graph page by right clicking on the tabs at the left.
- The Options|Format dialog box wasn't always reflecting the current status of the Show Seconds option. Fixed.
- Printing of frequency tables wasn't taking the Show Seconds option into account. Fixed.
- Fixed a problem with checking for valid values when doing search and replace on angles.
- Fixed error that could occur when subgrouping variable is changed after a filtering condition has been set.
- Fixed problems in setting the number of gridlines in QQ and von Mises distribution plots.
- If manual scaling was being used in scatterplots and some data values were outside the set range these points would be plotted anyway, but outside of the graph (e.g. below the axis or off to the right beyond the axis). Now fixed so values outside the range are ignored and not plotted.
- Labeling of scatterplot axes for compass variables was wrong - the second North tick at the end of the axis was missing. Fixed.
- Printed frequency tables for histograms weren’t showing the ranges correctly when the “Center bars on zero option” was selected. Fixed.
- Fixed error that could occur when manual scaling scatterplots with very large linear data values.
- Fixed range check error that could occur when a large number of vector pairs were declared.
- Fixed the Define Vector Pair dialog to allow the same linear variable to be used as a vector length with more than one circular variable.
- Settings of the statistics lines colors, thicknesses and styles were not being saved in the desktop file and weren't being undone properly. Fixed.
Ver. 3.02 - Released 23 June 2009
- Circular histograms drawn in a counter-clockwise direction had the bars misplaced (they were drawn one class-width too far in the counter-clockwise direction). Fixed.
Ver. 3.01 - Released 27 April 2009
- Fixed a problem where some basic statistics and pairwise comparisons were not being displayed for samples with extremely large concentrations (> 10,000).
- Fixed a problem in drawing raw data graphs that could very rarely result in a “Division by zero” error.
- Minor changes to the Check for Updates facility to avoid problems if the user tries to save the downloaded update to a folder that is not writeable.
- Editing graphs loaded from old desktop files created with Oriana version 2 could cause a “Range check” error. Fixed.
- The Outline Symbol option on raw and scatterplots was sometimes not outlining the symbols. Fixed.
- The von Mises distribution plots were not drawing any grid lines when the Number of Gridlines option was set to Auto. Fixed.
Ver. 3.00 - Released 5 February 2009