- Prism
FEATURES
Analyze, graph and present your workComprehensive analysis and statisticsElegant graphing and visualizationsShare, view and discuss your projectsLatest product features and releasesPOPULAR USE CASES
- Enterprise
- Resources
- Support
- Pricing
Prism 11.2.0
Prism 11.2.0 was released on November 16, 2026.
Prism 11.2.0
Real experiments are messy. Animals are lost at follow-up, samples fail, and time points get skipped. Prism 11.2.0 is built for that reality, with more trustworthy statistics for imperfect data and new ways to calculate exactly what you need, right in your data table.
More accurate mixed-effects models. Mixed-effects model ANOVA now offers the Kenward-Roger and Satterthwaite methods for estimating degrees of freedom, the approaches statisticians and journal reviewers expect. When using a repeated measures or nested analysis with missing values or an unbalanced design, these methods ensure that your F tests and P values are more accurate, with no need to export to another program. A redesigned Repeated Measures tab puts the degrees-of-freedom and sphericity options side by side and helps you choose between them.
Calculations that used to need a spreadsheet. Window functions allow you to use a single formula to calculate group means, percentages of a group total, cumulative doses, moving averages, and within-group z-scores. New conditional functions such as MEANIF, SUMIF, and COUNTIF work the way spreadsheet users already know, and formulas like 2*MEAN(B) now just work. Use the Multiple Variables data table to populate a variable using formulas with these new capabilities. Results update automatically whenever your data change.
Faster graphs and sturdier analyses. Multiple variables graphs with many points or levels now create, update, and resize much more quickly. Cox regression converges on data that previously failed, and Baptista-Pike confidence intervals for odds ratios are more accurate.
And more. Easier level reordering, clearer labels for features included in other plans, a broad round of accessibility improvements for keyboard and screen reader users, and 70 other fixes.
Kenward-Roger and Satterthwaite Degrees of Freedom
Prism fits a mixed-effects model when repeated measures or nested data have missing values, and for nested t tests and nested one-way ANOVA. The F tests and P values from these models depend on the denominator degrees of freedom. The method Prism used in earlier versions, now labeled Legacy, is reliable only when the data are balanced and complete. Prism 11.2.0 adds two widely used methods that remain accurate when they are not.
- Kenward-Roger: Estimates the denominator degrees of freedom and also adjusts the standard errors of the fixed effects for small samples. This is the recommended method for most experiments.
- Satterthwaite: Estimates the denominator degrees of freedom without the small-sample adjustment to the standard errors.
- Legacy: Still available, with a warning that it is reliable only for balanced, complete data.
- Where it applies: Mixed-effects model fits in one-, two-, and three-way repeated measures ANOVA, nested t tests, and nested one-way ANOVA.
- Multiple comparisons: When Kenward-Roger or Satterthwaite is selected, the available follow-up tests are Bonferroni, Šídák, Holm-Šídák, the false discovery rate methods, and uncorrected Fisher's LSD.
- Reorganized Repeated Measures tab: The degrees-of-freedom methods and the Geisser-Greenhouse sphericity options are now on the same tab. Sphericity corrections are disabled when Kenward-Roger or Satterthwaite is selected, because these methods do not assume sphericity.
Why it matters: Missing values are common in repeated measures and nested experiments. A mixed-effects model can still analyze these data, but with the Legacy method the P values could be too small or too large when the design was unbalanced. Kenward-Roger and Satterthwaite are the methods statisticians and reviewers expect for these models, and Prism now uses them directly. You no longer need to export your data to another program to get trustworthy results.

Window Functions in Calculated Variables
Included with Prism Pro and Enterprise
Window functions calculate a summary statistic, such as a mean, sum, or count, over a set of related rows and report the result on each row. You write them in calculated variables on Multiple Variables data tables, and in custom transforms in Select and Transform, by adding clauses after a function. For example, MEAN(E) OVER B returns the mean of each treatment group on every row of that group.
- OVER: Splits the rows into groups and calculates the function separately within each group.
- WHERE: Chooses which rows contribute to the calculation, without removing any rows from the table.
- ORDER BY: Sets the order of the rows within each group.
- ROWS: Limits the window to rows near the current row, for running totals and moving averages. For example, SUM(F) OVER A ORDER BY C ROWS UNBOUNDED PRECEDING returns a cumulative sum for each subject in time order.
- Row-by-row statistics: CENTER, STANDARDIZE, ZSCORE, and RANK also accept window clauses, so you can calculate z-scores or ranks within each group.
Why it matters: Group means, percentages of a group total, normalization to a baseline, cumulative doses, and moving averages are everyday calculations. Before, they required restructuring the data, copying values by hand, or working in a spreadsheet and pasting the results back. Now a single formula calculates them in the data table, and the values update automatically when the data change. Window functions in Select and Transform custom transforms are available with all plans.
Conditional Aggregation Functions
Included with Prism Pro and Enterprise
New functions summarize the values of one variable using only the rows that meet a condition on another: MEANIF, SUMIF, MINIF, MAXIF, STDEVIF, STDERRIF, COUNTIF, and COUNTAIF. For example, SUMIF(F, C <= 14) returns the total of variable F for rows where C is 14 or less. These functions also accept window clauses, so SUMIF(F, C <= 14) OVER A returns that total separately for each subject.
Why it matters: Spreadsheet users already know this style of formula. It is now available in Prism calculated variables, so you can calculate quantities like the mean baseline value or the count of responders directly from your data. Conditional aggregation functions in Select and Transform custom transforms are available with all plans.
Arithmetic with Descriptive Statistics Functions
Included with Prism Pro and Enterprise
Formulas in calculated variables can now combine descriptive statistics functions with arithmetic. Formulas such as 2*MEAN(B), MEAN(B)+2, and MEAN(B*C) now calculate correctly, so you can normalize to a mean, express values as a fraction of a total, or summarize a product of two variables in one step. Formulas that combine descriptive statistics functions with arithmetic in Select and Transform custom transforms are available with all plans.
Complete Release Notes
Upcoming Changes
- (Mac) Prism 11.3.0 will require macOS 13 (Ventura) or later. Prism 11.2.0 continues to run on macOS 12 (Monterey), but if you use macOS 12, plan to update your operating system before upgrading to Prism 11.3.0.
New Features
Math and statistics
- Added the Kenward-Roger method for estimating denominator degrees of freedom in mixed-effects model ANOVA, giving more accurate F tests and P values when data are missing or unbalanced.
- Added the Satterthwaite method for estimating denominator degrees of freedom in mixed-effects model ANOVA, giving more accurate F tests and P values when data are missing or unbalanced.
- Added window function clauses (OVER, WHERE, ORDER BY, and ROWS) to aggregate functions in calculated variables, letting you calculate statistics within groups of rows, limit which rows contribute, and use sliding row windows for running totals and moving averages.
- Added conditional aggregation functions for calculated variables (MEANIF, SUMIF, MINIF, MAXIF, STDEVIF, STDERRIF, COUNTIF, and COUNTAIF), letting you summarize values from one variable based on a condition applied to another.
- Added support for arithmetic operations with descriptive statistics functions in in-table transforms, so formulas like 2*MEAN(B), MEAN(B)+2, and MEAN(B*C) now calculate correctly.
Other new features
- Added the ability to change the variable assigned to a graph object property, such as color, size, or grouping, directly from the Graph Inspector without opening the Format Graph dialog.
- Added a Size definition control to the Size tab of the Symbols section in the Graph Inspector, so you can change the variable assigned to symbol size directly, just as you already can for symbol color.
- Added stepwise level-reordering controls (Move up, Move down, Move to top, and Move to end) to the Format Data Table and Format Graph dialogs, so you can build a custom level order step by step, including from the keyboard.
- Added a Region selector for LabArchives sign-in in the Welcome dialog, so users outside the US can choose their region directly instead of editing a local configuration file.
Feature Improvements
Math and statistics
- Improved the numerical stability of Cox regression (Breslow, Efron, and Exact tie methods), so fitting now converges for data with large differences in predicted risk over time that previously failed or crashed.
- Improved the accuracy of Baptista-Pike confidence intervals for odds ratios in contingency tables with large counts, so Prism no longer falls back to the Gart adjusted logit method.
Reorganized ANOVA parameters dialogs
- Moved the Geisser-Greenhouse sphericity options to the Repeated Measures tab of the one-, two-, and three-way ANOVA parameters dialogs. These options are disabled when the Kenward-Roger or Satterthwaite method is selected.
- Refined the Repeated Measures tab of the ANOVA parameters dialogs: reordered the denominator degrees-of-freedom methods (Kenward-Roger, Satterthwaite, Legacy), strengthened the warning for the Legacy method, and clarified the "It depends" guidance.
- Updated the hint box on the first tab of the ANOVA parameters dialogs to reflect the selected degrees-of-freedom method and to summarize the options for random effects with zero or negative variance.
- Corrected the heading of the zero-variance section in the ANOVA parameters dialogs to refer to a random effect's variance being zero or negative, rather than the effect itself.
Other feature improvements
- Improved the default layout of multiple variables categorical graphs, which now assign the categorical variable to the X axis and the continuous variable to the Y axis, so graphs like bar charts are vertical by default.
- Added the current license plan name to the About box and License Information dialog, so you can quickly see which Prism plan you're on (for example, Perpetual License or Legacy Subscription).
- Improved the positioning of the restriction pill on the Axis Variables tab of the Format Graph dialog by grouping the X and Y axis controls under a new "Variable assignment" heading, so the pill clearly applies to both axes.
- Improved how restricted graph objects appear in the Graph Inspector: all settings are now shown but disabled, with an UPGRADE pill on each section header, so unavailable features stay visible and discoverable.
- Continued the accessibility program with the next phase of keyboard, screen reader, and contrast improvements:
- Analyze Data dialog: full keyboard navigation of the analysis tree, search field, and "Use" dropdown, a visible keyboard focus, and screen readers now announce each control's name along with the selected item.
- Create New Graph and other dialogs: full keyboard navigation in the New Data Table and Graph dialogs, the Format Graph and Format Axes tabs, the Data tab of analysis parameters dialogs, and the Software Update dialog.
- Main document and Inspector: selected items in the "Available variables" and "Available functions" lists of the Data Inspector stay legible when the list loses focus.
- Welcome dialog: keyboard access to collapsible sections, tabs, and graph previews, a logical focus order with no keyboard traps, descriptive screen reader labels for preview images, and contrast for text and example images that now meets WCAG minimums.
- (Mac) VoiceOver now announces data table and results sheet cells, titles, and placeholders, and floating notes no longer block keyboard focus (press Control-F6 to move to a note and Control-F4 to return to the main window).
Performance Improvements
- Improved the performance of creating and updating multiple variables Strip, Bar, Box and Whisker, Violin, and Summary Symbol graphs with a large number of points or levels. Performance is now on par with legacy graphs.
- Improved the performance of resizing multiple variables Strip and Violin graphs with a large number of points or levels. Performance is now on par with legacy graphs.
- (Windows) Improved the performance of the Correlation analysis.
- (Windows) Improved the performance of creating and updating multiple variables Before-After Strip graphs with many points or levels. These graphs are now significantly faster than the corresponding legacy graphs.
- (Mac) Improved the performance of refreshing a Strip graph with large data sets or many levels after editing data table values. Prism 11.2.0 is faster than Prism 11.1.0 and earlier.
- (Mac) Improved the performance of creating and updating multiple variables Before-After Strip graphs with many points or levels. These graphs are now significantly faster than in Prism 11.1.0 and close to the speed of legacy graphs.
Compatibility Notes
- Mixed-effects analyses created in earlier versions of Prism keep their Legacy degrees-of-freedom results when the project opens. When you open the Parameters dialog and click OK, the analysis switches to the Kenward-Roger method and recalculates, so degrees of freedom and P values can change.
Analysis Bug Fixes
- Fixed an issue where changing the grouping factors on the Data tab of Multifactor ANOVA reset the model to main effects only. Interaction terms are now preserved when factors are added or removed.
- Fixed an issue where "+infinity" appeared instead of blank cells in the Multiple Comparisons results of repeated measures one-way ANOVA when fitting a full model with lognormal data and a non-positive definite variance-covariance matrix.
- Fixed an issue where the Multiple Comparisons results of a lognormal one-way ANOVA (Brown-Forsythe and Welch) were not recalculated to match the selected test when reopening projects created in Prism 10.6.1 through 11.0.2.
- Fixed an issue where a "Disk or memory error occurred" alert appeared in Two-way ANOVA, instead of recalculated results, when adding a data set before or between columns preselected for multiple comparisons.
- Fixed an issue where a misleading warning about missing time-point data appeared on an empty Two-way ANOVA results sheet, instead of results, when the source data had some missing values.
- Fixed an issue where the floating note explaining that Two-way ANOVA requires N values, not only mean and SD or SEM, did not appear on the empty results sheet when some N values were missing from the data table.
- Fixed an issue where the wrong floating note appeared on the One-way ANOVA results sheet when switching from a normal to a nonparametric test with missing values in the data table.
- Fixed an issue where the wrong floating note ("You have not entered the right kind of data") appeared on a multiple variables nonlinear regression results sheet after clearing the analyzed variable, instead of the note that the model is out of date.
- (Mac) Fixed a crash in Two-way ANOVA when adding a data set before or between columns preselected for multiple comparisons. The results are now recalculated.
- (Mac) Fixed an issue where a custom default confidence level, set with "Make options on this tab be the default", was not applied to new t tests, which still used 95%.
- (Mac) Fixed an issue where the "Diagnostics for residuals" option stayed enabled for Welch's t test, which does not support it, and selecting it produced no results. The option is now disabled when Welch's test is selected.
Graphing Bug Fixes
- Fixed cropped values in the "Number at risk" table on survival graphs when a larger, non-default font size was used. Values now display centered under the X axis label instead of being truncated with the group-title separator hidden.
- Fixed an issue where graph titles shifted down and overlapped the plotting area when reopening projects from earlier Prism versions in which the title's distance from the top of the graph was set very small.
- (Windows) Fixed an issue where the global info sheet was missing when a graph was sent to Microsoft Office as an embedded object and then opened from Word or PowerPoint.
- (Windows) Fixed an issue where Arabic text used as axis labels was displayed in reverse order.
- (Windows) Fixed an issue where confidence and prediction bands showed unwanted zig-zags on graphs with a segmented X axis that had a gap between the segments.
- (Windows) Fixed an issue where X axis labels in Japanese were misaligned.
- (Windows) Fixed an issue where bar annotations appeared in the top-left corner of certain grouped graphs and remained after the bars were removed. Annotations now display in the correct position and are removed with the bars.
- (Windows) Fixed an issue where subscript and superscript formatting in the X axis title of XY graphs reverted to normal text after saving and reopening the file, or after any change to the graph.
- (Mac) Fixed an issue where the line of identity on XY graphs was drawn above the data symbols after clicking OK in the Format Graph dialog. It now stays below the symbols, as it does after clicking Apply.
- (Mac) Fixed an issue where lines with text, such as pairwise comparison annotations, moved to the top-left corner of the graph after applying formatting with Magic.
- (Mac) Fixed an issue where the show/hide (eye) icon was missing in the Inspector for Dendrogram, Heatmap, Confidence Ellipse, and Convex Hull graph objects.
- (Mac) Fixed an issue where the wrong contextual menu appeared for separate categorical size and color legends on multiple variables graphs.
Other Bug Fixes
- Fixed a crash that could occur when saving a project in .pzfx or .pzf format if a categorical variable contained excluded cells.
- Fixed an issue where a calculated variable showed category index numbers instead of the referenced variable's values after switching it from categorical to continuous.
- Fixed an issue where non-integer values in a calculated variable showed an excessive number of decimal places (floating-point artifacts) after switching the variable from continuous to categorical.
- Fixed an issue where the contextual menu for graph objects showed commands that do not apply, such as symbol-only commands for an object with no symbols. These items are now hidden, and the menu is consistent across Windows and Mac.
- Fixed an issue where an error message appeared too soon while typing a calculated variable formula in the Data Inspector. The error now appears only when you confirm the formula.
- Fixed an issue where the floating note did not appear on a multiple variables heat map when the continuous variable used as the metric was deleted.
- (Windows) Fixed a data-loss issue where closing an edited .prism file after switching between sheets many times could skip the "Do you want to save changes?" prompt, discarding the changes and removing the backup file.
- (Windows) Fixed a data-loss issue where data table values were replaced with zeros or blanks after opening and saving the automatically saved copy of a project that had been closed without saving after excluding values.
- (Windows) Fixed a crash that could occur when editing a graph's appearance in the Format Graph dialog in files created with older Prism versions.
- (Windows) Fixed an issue where calculated variable values were not saved to the .prism file after unfreezing a frozen data table and saving, which made the column appear empty when the project was opened in Prism 10.6.1 and earlier.
- (Windows) Fixed an issue where Prism activated with a serial number could repeatedly prompt for the serial number and activation code on startup when the license server's response was delayed.
- (Windows) Fixed an issue where closing the reactivation window without reactivating a deactivated license and clicking "Continue Activation" displayed a misleading "Remind Me Later" alert instead of prompting to quit.
- (Windows) Fixed an issue where the Prism activation window appeared corrupted after moving it between displays with different scaling settings.
- (Windows) Fixed an issue where activating Prism from Viewer Mode did not automatically sign users in to their Prism Cloud account.
- (Windows) Fixed an issue where the "Share..." and "Open in Prism Cloud..." commands did nothing after reopening a project published to Prism Cloud, before any other cloud operation. They now open the relevant page in your browser.
- (Windows) Fixed an issue where the Prism Cloud submenu in the File menu offered commands that do not apply when the open project was published by another user to a workspace the current user cannot access.
- (Windows) Fixed an issue where reordering sheets in the Gallery view, by dragging or with the Reorder Sheets dialog, did not update the sheet order in the Navigator.
- (Windows) Fixed an issue where the Fill pattern checkbox in the Format Graph dialog did not match the actual state of each data set on XY graphs. Unchecking it for all data sets or for one now updates the graph consistently.
- (Windows) Fixed an issue where double-clicking a column title did not open the Column Titles tab of the Format Data Table dialog.
- (Windows) Fixed an issue where the UNINSTALLPRISM10=1 option for silent installation did not remove Prism 10 when installing Prism 11.
- (Windows) Fixed an issue where Prism 11 graphs pasted into Microsoft Excel were identified as Prism 10 objects in the Excel formula bar.
- (Windows) Fixed an issue where the data table unexpectedly scrolled to the first variable when switching to Formula Editing Mode for a calculated variable while an entire variable was selected.
- (Windows) Fixed an issue where the Group dropdowns in the Format Graph dialog did not show variable levels for Convex Hulls or Confidence Ellipses after switching the fill or border color from "Single color" to "Variable (by group)".
- (Windows) Fixed an issue where the axis label dropdown menu appeared truncated on the Labels subtab of the Format Graph dialog for a heat map when "Variable values" was selected for the Y axis labels.
- (Windows) Fixed an issue where the "Color for non-numeric values" checkbox label was truncated in the Heatmap section of the Inspector when a heat map used a custom color scheme.
- (Windows) Fixed an issue where the separator lines between dropdowns in the Create New Graph and Change Graph Type dialogs flickered when hovering over them.
- (Windows) Fixed an issue where an extra "Preview" label appeared in the Create New Graph dialog for survival graphs.
- (Windows) Fixed an issue where controls on the Error Bars view of the Format Graph dialog shifted position when changing the Stroke Color or Type dropdown.
- (Windows) Fixed an issue where the Equivalents section of the NORMDIST function description was truncated when the Prism window was not maximized.
- (Mac) Fixed an issue where graphs exported to a network folder over SMB appeared dimmed and then disappeared from the folder.
- (Mac) Fixed an issue where setting color and transparency in the system Colors panel with the color wheel, sliders, image palettes, or pencils had no effect on macOS 27.
- (Mac) Fixed an issue where the Select All control in the top-left corner of tables in several dialogs, such as Nonlinear Regression initial values and Format Axes additional ticks, did nothing when clicked on macOS 27.
- (Mac) Fixed an issue where text after a slash (/) in an Info sheet constant was truncated after saving and reopening a project in .pzf or .pzfx format.
- (Mac) Fixed an issue where excluded values disappeared from a calculated variable when duplicating a Multiple Variables data table.
- (Mac) Fixed an issue where the Format option on the X axis tab of the Format Axes dialog was unavailable for interleaved, separated, and stacked bar graphs, which prevented using row or column titles as X axis labels.
- (Mac) Fixed an issue where the "Set all scaling factors to" field in the Equalize Scaling Factors dialog on a layout sheet did not accept a custom value.
- (Mac) Fixed an issue where cell background colors stayed at the same row and column positions after inserting or deleting rows or columns in a data table. The highlighting now moves with the cells.
- (Mac) Fixed an issue where pressing Shift-Command-Down Arrow or Shift-Command-Up Arrow in data or results tables, when the first or last cell in a column was already selected, unexpectedly added adjacent columns to the selection.
- (Mac) Fixed an issue where Shift-Command-Right Arrow did not extend the selection to the remaining data sets in a row when the last cell containing data was already selected.
- (Mac) Fixed an issue where Undo and Redo were unavailable when editing a calculated variable's formula in Transform Definition Mode.
- (Mac) Fixed an issue where Transform Definition Mode did not start when pressing = with the insertion point placed before the name or value of an existing variable.
- (Mac) Fixed an issue where the list of recently used functions in the Data Inspector did not update immediately after a function was used.
- (Mac) Fixed an issue where the Backspace key deleted cell values, column titles, and variable names in a frozen data table. Frozen data tables can no longer be changed from the keyboard.
- (Mac) Fixed an issue where a column title was not imported from a text file when it was placed in a subcolumn other than the first.
- (Mac) Fixed an issue where the Symbols tab of the Format Graph dialog showed a checkbox labeled "Color for non-numeric values" instead of "Use as color for non-numeric values".
- (Mac) Fixed an issue where the floating note appeared twice on data tables linked to restricted multiple variables analyses or to graphs with restricted graph objects.
- (Mac) Fixed an issue where the floating note on data tables and graphs with a custom level order showed the wrong variable name and number of affected variables when the project was opened with a Prism plan that does not include custom level order.
Analyze, graph and present your scientific work easily with GraphPad Prism. No coding required.