Prism 8.2.0 Release Notes
Prism 8.2.0 was released on July 21, 2019.
We get it. You have work to do, and reading through a long list of release notes isn't the most exciting thing to do. But you should know that there are a LOT of new and exciting features. Check out some of the highlights below, or see the entire list of improvements, changes, and fixes now available in Prism version 8.2.0!
New Scatter Plot Options
Being able to visualize your data is important! Being able to control HOW to visualize your data may be even more important! When we released Prism 8, we changed the way scatterplots were visualized. We focused on providing a sense of the overall distribution of the data. While many people were happy with this change, others were concerned that with some data sets, many data points began to overlap in the scatterplot. To address this, we made some additional changes in an attempt to prioritize minimizing overlap of data points. However, after these changes, we learned a very important lesson: you can't please everyone! Some people preferred the visual style of scatterplots prior to Prism 8. Others liked scatterplots that provided a good estimate for the overall distribution of the data. Still others liked our newest algorithm focused on minimizing data overlap. We heard you, and now the choice is yours.
We've introduced the ability to choose between three different methods for plotting your data points when creating a scatter plot. These new options are:
- Standard: With this method, the width of the distribution of data points is proportional to the number of individual points at that Y value. This will create a visualization with your data points similar to the appearance of a violin plot. Sort of like a histogram. This is the best choice if you're wanting to visualize the distribution of your data
- Classic: This method prioritizes minimizing the amount of overlap of individual data points over representing the shape of the distribution. In regions of less data, it will try to spread these points out so you can see individual points better. Additionally, this method may create "smiles" in the scatterplot (similar to the appearance of scatterplots in prior versions of Prism)
- Expanded: This method also prioritizes minimizing the amount of overlap of individual points over representing the shape of the distribution. Data points in regions with less data will still be spread out, but without "smiles"
Which visualization you choose all depends on what you're looking for. Want to visualize the data distribution? Go with Standard. Want to avoid overlap of data points in regions of sparse data? Try Classic or Expanded. In fact, try all of them! You can also set a default choice between these three styles in the New Graph tab of Prism's preferences.
Results Tabs (That Look Like Tabs!)
Starting in Prism 8.0, analyses with more than one page of results organized those extra results in tabs found at the top of the results sheet. We've given those tabs a makeover in Prism 8.2, making them easier to find and use. Use the dropdown on the right-hand side of these tabs to choose which tabs to show or hide.
Improved Navigation (Data and Results... Together again)
You asked. We listened. By default, Prism keeps Data sheets and Results sheets in separate folders within the Navigator at the left of the Prism window. But when working with a large numbers of different Data sheets, each with one (or more) analyses and Results sheets, you may want to quickly see which analyses are related to which tables. We've introduced the ability to turn on the "Data with Results" view that keeps each of the Results sheets 'nested' with its associated Data sheet. Access this feature either by right-clicking within the Navigator or on the View tab of Prism preferences.
Our first Prism Labs feature: Unstable parameters rather than ambiguous nonlinear fits
When performing nonlinear regression, Prism will identify some fits as "ambiguous". Prism will make this identification when the Dependency of any parameter is greater than 0.9999. For previous versions of Prism starting with Prism 6, this was the only option.
Our smart statistics programmers found a better approach that will sometimes give useful results when the old approach would have reported: "ambiguous". Instead of labeling entire fits as "ambiguous", you can now choose to have Prism identify individual parameters as "unstable". You can choose between the two approaches on the confidence tab of the nonlinear regression dialog. The new approach is based on determining the "condition" (or ill-condition) of the Hessian matrix that is a key part of the nonlinear regression algorithm. Parameters that are considered to be unstable are so labeled. Prism doesn't report a best-fit value or confidence interval for these parameters because the data simply doesn't define the values for these parameters. However, in some cases this new algorithm will be able to report confidence intervals for stable parameters where the older approach would not.
This is also our first feature release as part of our new Prism Labs. Features coming out of our Prism Labs have been extensively tested, but can still be considered "experimental". Try out our latest and most innovative features as part of Prism Labs, and let us know what you think! We welcome your feedback, and are always working to make these features even better!
Complete Release Notes for Prism 8.2.0
- New: Choose from three different visualization styles when creating scatter plots. Set a preferred style within Prism preferences to be the default for future graphs
- New: A new algorithm available during nonlinear regression provides for improved detection of "unstable" parameters
- New: Absolute IC50 equation added to Nonlinear Regression Equation library. The absolute IC50 is the concentration that gives a response of 50%, which is distinct from the relative IC50, which is the concentration that gives a response halfway between the top and bottom plateaus
- New: All dose-response equations now include versions for X as concentration and X as Log(concentration)
- New: Added the Natural Log (Ln) axis scale option (Prism already allowed log10 and log2 axes)
- New: Introduced the ability to add multiple annotations for a single bar
- Appearance of results tabs has been improved to increase visibility and usability
- The "Data with Results" view can now be activated from the Navigator to display all linked Data and Results sheets together instead of within separate folders. Setting this view as default can be done using Prism's Preferences
- Tweaked grid line colors and weights to make the look consistent in Windows and Mac and to make the grid lines in results sheet a bit more prominent
- Introduced "Data summary" section to results of ordinary Two-way ANOVA displaying number of columns (Column Factor), rows (Row Factor) and values
- Split the "Dose-response Special" folder of equations into two folders, one for when X is concentration and one for when X is Log(concentration)
- (Windows) Enabled exporting of semi-transparent objects on a graph to PDF
- (Windows) Made it possible to re-size the "Software Update" dialog in order to make it easier to read the release notes
Performance (Speed) Improvements
- Included analytical derivatives for the "Asymmetric Sigmoidal, 5PL, X is concentration" and "Sigmoidal, 4PL, X is concentration" equations. This will make fitting these models faster, as Prism won't have to compute numerical derivatives. Improved performance of Two-way Repeated Measures ANOVA
- Improved Navigator performance for large projects
- (Mac) Improved performance on copying and pasting data sets with huge numbers of rows
Fixed Analysis Bugs That Resulted in Incorrect Results
- Corrected results displayed for RM Two-way ANOVA if there was no variation between subjects
- Fixed formatting results of Deming Linear regression when there were too few points or all the data had the same X value
Other Analysis Bugs Fixed
- Prism no longer freezes when attempting to perform RM Two-way ANOVA on huge data sets
- Fixed an issue causing Prism to crash when performing Ordinary Two-way ANOVA for data with huge range
- Fixed an issue causing Prism to crash when performing Repeated Measures Three-way ANOVA if the source data table contained a huge amount of data
- Added an error message when Prism fails to generate a spline curve: "For some data sets, Prism failed to generate a curve because there are not enough data."
- Fixed an issue in which it was impossible to perform a "ROC Curve" analysis with two selected data sets if the data table contained more than two data sets in total
- Changed the "Mean Diff" section title to "Predicted (LS) mean diff" in "Multiple Comparisons' results tab of "Two-way ANOVA" analysis if missing values are present in the source data table
- Fixed an issue in which the "Are lines different?" results sheet would not show "You have not entered enough data to compare regression lines."
- Fixed an issue in which the "Two-way ANOVA" results sheet was empty and no floating note appeared when: the source data was formatted as Mean,SD/SEM,%CV,N and contained identical values
- Fixed an issue in which the results were empty with no floating note when creating a LOWESS curve from a data set with fewer than 12 data points
- Fixed an issue in which "Hougaard's Measure of Skewness" was not reported in the 'Nonlinear Regression' results sheet when using the equation "Asymmetric Sigmoidal, 5PL, X is log(concentration)"
- Added an empty "Effect Size" section with floating note in "Contingency" results sheet if one row/column contained only zeroes and some effect sizes were selected to be reported
- Fixed Prism so that it shows "The Woolf logit interval was calculated instead" floating note in "Contingency" results sheet if you requested the Baptista-Poke method but the total sample size is larger than 100,000
- Fixed the issue in which "Likelihood Ratio" result unexpectedly changed from 0.00 to expected value after re-opening the "Parameters: Chi-Squared (and Fisher's exact) test" dialog several times
- Fixed the issue in which "P value and statistical significance" and "Data analyzed" sections were absent in "Contingency" results sheet if the source data table contained missing, negative, or non-integer values
- Improved output for post-tests after ANOVA when some of the individual t-tests are invalid by adding a floating note with brief explanation
- Fixed the issue with incomplete output for post-tests after two-way ANOVA in 'Summary' and 'Adjusted p-values'
- Added the missing "%" symbol in the "Narrative results" tab of Two-way ANOVA
- Fixed a typo in the Nonlinear Regression dialog: "The X values should be plotted on an axis with a logarithmic scale, ..."
- Fixed an issue when Ordinary Two-way ANOVA results sheet appeared instead of RM Two-way ANOVA results sheet after changing format of Y values from replicate values to Mean, SD/SEM/%CV, N
- Added a floating note to the results sheet of the Correlation matrix analysis when only one row contains data for both variables or one of the variables has the same value in all rows
- (Windows) Fixed an issue causing Prism to freeze after clicking the OK button in the "Parameters: Fit Spline/LOWESS" dialog if the number of rows with data was very large (~500,000 or more) and the "Number of segments" edit field was empty
- (Mac) Fixed the issue in which Prism would crash upon switching to "ROC Curve" results sheet if analyzed data sets contained a large number of values (~500,000 or more)
- (Mac) Fixed the issue in which "Confidence bands" and "Prediction bands" radio buttons unexpectedly appeared enabled in the "Confidence" tab after changing the fitting method from "Least squares regression" to another option
Graphing Bug Fixes
- Set the color used for data points on some automatically generated graphs to semitransparent so overlapping residuals are noticeable
- Fixed the issue in which connecting lines were drawn incorrectly on the graph if two or more points which were located side by side appeared outside the axis limit
- Fixed the issue when white spaces unexpectedly appeared on a grouped graph with bars if you choose to remove spaces using the spacing option in the 'Graph Settings' tab in the 'Format Graph' dialog
- Fixed the issue when data was not shown as stacked bars after changing the minimum range value for the Y axis
- Fixed the issue when Prism incorrectly plots "Points & connecting line" XY graph when there are missing values in a source data table
- Fixed the issue when plotted data unexpectedly changes at the end of the graph if data had negative values on Y-axis and Y-axis minimum range was 0 or higher
- Fixed the issue when part of values appeared outside of the range on the Heatmap graph after changing Paired t test to Unpaired t test if the source data table contained one row with replicates
- Fixed the issue when labels' orientation resets to below and vertical after re-opening Prism Win project in Prism 8 Mac
- Fixed the issue in which plotted values unexpectedly disappeared from a Column bar graph after adding symbols to it
- Fixed the issue when a long multi-row Y axis title on a graph near the page edge was unexpectedly divided into several rows
- (Windows) Fixed the issue in which only one data point on a nested graph was formatted after choosing to format all points in a data table subcolumn
- (Windows) Fixed the issue when connecting lines at the edge of an axis range on a "Points & connecting line" graph were thinner than other connecting lines.
- (Windows) Fixed the issue where connecting lines appeared on top of symbols, even when the option to display the line underneath the symbols is selected
- (Windows) Fixed the issue when ticks were incorrectly directed after changing label position for heatmap graph
- (Windows) Fixed the issue when plotted values did not appear on 'Interleaved symbols', 'Separated symbols' and 'Superimposed symbols' graphs with bars from the Grouped graph family
- (Windows) Fixed the issue when the default location of legends was changed after making the multiple selected legends larger and than smaller
- (Windows) Fixed the issue when lines connecting points on the edge of the Y-axis range on a "Points & connecting line" graph became unexpectedly thinner than other connecting lines
- (Mac) Fixed the issue in which points on a before-after graph were being labeled with row titles in a Layout with "Label each point with its row title" option enabled, while the labels were not present on the original graph
- (Mac) Fixed an unexpected artifact that appeared on a graph after switching from edit mode to view mode from the graph title or axis title if zoomed in on the graph
- (Mac) Fixed the issue in which mean/median did not appear on a Scatter graph for newly added data sets from teh data table with another format of Y values
Other Bug Fixes
- Fixed the issue in which Prism unexpectedly crashed after performing cut operations in Microsoft PowerPoint documents containing linked Prism graph objects
- Fixed the issue when “Fill Pattern” submenu in the “Format This Bar” section of the contextual menu for scatter plot with bar graph would appear active, if “Fill pattern” checkbox in the Format Graph dialog is unchecked
- Added "0" to label the axis in the "Residuals plot" and 'Homoscedasticity plot" thumbnails with graphs in the 'Residuals' tab in Parameters dialogs
- Fixed the issue when dividers (/) were unexpectedly removed from the date entered into custom fields of an Info sheet upon reopening a project
- Fixed the link to a help page from the interpret button for Normality tests
- Disabled the Error drop-down in Change Graph Type dialog to be consistent with same drop-downs in the Format Graph dialog when there are entered limits or plus/minus values
- Fixed the issue when text size and graphs inserted to the layout were resized unexpectedly after opening Prism 7 file in Prism 8
- Fixed the issue when the link between graph sheet name and data set title was lost after creating a graph from a data set and then deleting column titles in source data sets
- Fixed the issue when Prism pasted data incorrectly after checking 'After rows, start a new column' checkbox in the 'Import and Paste Special' dialog
- Fixed the URL linked to the "Learn more" button in Welcome dialog for Prism Beta
- Fixed the issue when the graph title of Parts of Whole graph did not adopt the name of the renamed data table if the corresponding data set does not contain column title
- Fixed the URL for the hint on the Column tab of the Welcome dialog
- Fixed the issue when the "Help" section and the "GraphPad Prism 8" update notification toolbar icon were gone on a resolution of 1280x1024 and lower
- Fixed the issue in which a new Contingency/Parts of Whole/Multiple variable/Nested project was not created after double-clicking "Enter or import data into new table" on the Welcome dialog
- Fixed an issue in which the Help page for the "Factor Names" tab of the Three-way ANOVA parameters dialog was missing
- Fixed an issue in which the Help page for the "Residuals" tab of the One-way ANOVA parameters dialog was missing
- (Windows) Fixed the issue in which Prism unexpectedly crashed when deleting data tables when the delete command was previously undone
- (Windows) Fixed the issue in which Prism unexpectedly crashed while double-clicking on a linked graph in Microsoft PowerPoint while the original graph in Prism was deleted
- (Windows) Fixed the issue in which the old "Dose-response - Special" folder was present in the Parameters: Nonlinear Regression dialog
- (Windows) Fixed the URL linked to the "Learn more" button in Welcome dialog
- (Windows) Made the pasting of values from green results to data tables be smarter so that decimal places are automatically determined in a target table
- (Windows) Fixed the issue when numerous alerts about disconnected drives unexpectedly appeared when Prism was running in the background with the project file opened from a mapped drive
- (Windows) Fixed the issue when excluded row title in a data table had no asterisk indicating it is excluded
- (Windows) Fixed crash on switching to gallery mode after undoing the deletion of sheets
- (Windows) Fixed the issue when there was no difference between 1/2 and 1/4 thickness set for symbol border on a graph sent to Microsoft Office
- (Windows) Fixed the issue in which zooming did not work properly for 'orphaned' results sheets
- (Windows) Fixed the issue in which the "Merge" command didn't work while the "Results" section was empty in the Navigator
- (Mac) Fixed the inability to insert European characters from the "Character Palette" dialog into "Info Notes"
- (Mac) Fixed the issue in which row titles in a data table became unavailable for adding text
- (Mac) Fixed the issue in which special characters were lost after saving and reopening a project in PZF format
- (Mac) Fixed the issue in which custom subcolumn titles were lost after performing "Undo" operation
- (Mac) Fixed a typo in the "Parameters: Multiple Linear Regression" dialog on the "Diagnostics" tab - "How to quantify goodness-of-fit?"
- (Mac) Fixed inability to save a project after copying row and column titles to a different data table
- (Mac) Fixed inability to save a project with Two-way ANOVA analysis after deleting data set and performing "Undo" operation
- (Mac) Fixed truncated text on the "Random error" tab of "Parameters: Simulate XY data" dialog
- (Mac) Fixed a typo in "Wilcoxon. Also compute the CI of the median paired differences" checkbox in "Parameters: t Tests" dialog
- (Mac) Fixed a typo in "Parameters: Nonlinear Regression" dialog on the "Compare" tab - "Extra sum-of-squares F Test"