it’s been a little quite around Onyx last year. We apologise and we hope that you all still enjoy Onyx.
The other day, my colleague Sarah asked me, how one could change the font color on path labels. I realised that we never implemented a color picker for that. So, now there are new menus for picking font colors of variable labels and path labels.
Also, we tidied up the style presets and added two new ones. One of them is the “cinema” style, which is sort of a tribute to Wes Anderson. Download the latest developer version to check it out.
Great news! It took us long enough but we finally have implemented chi^2 and derived fit statistics for missing data. Ωnyx now separately estimates the log-likehood of the saturated model when data have missing values. So, we finally can offer you CFI and RMSEA under missing data.
Also, we have one thing with which we beat all other players. Now, Ωnyx has awesome background images. In the main menu bar (top of the screen), choose settings and enjoy! The images were provided by http://deathtothestockphoto.com.
Today, we are proud to release another update to Ωnyx. The latest version allows you to obtain latent factor scores. You can find this new feature in the model’s context menu Estimation -> Obtain Latent / Missing Scores. Choosing this option will create a new data set with all observed variables and additional columns with estimates for all latent variables.
We are happy to release another update of Ωnyx with bug fixes and a couple of new features (see section Downloads for the latest version). Primarily, we fixed a couple of GUI-related problems with copy&paste and drag&drop, particularly so with definition and group variables. It seems that mean structures for lavaan scripts were not always correctly exported, so we added an explicit export of the mean structure when saturated means are selected in Ωnyx. Also, it seems we have “lost” the feature to hide variances from the graph, which now returned in the customize model menu.
As a new feature, you will find two new graph presets to play with: Mint and Camouflage (picture below):
Also, Ωnyx supports dotted and dashed path styles:
We would like to thank everybody for their great feedback that helped us improving Ωnyx a lot!
We are happy to announce a maintenance release of Ωnyx 1.0. We are grateful for your positive feedback, which allowed us removing a couple of bugs in the former version. Particularly, we must mention that there were problems reported concerning the counting of degrees of freedoms when a saturated means structure was chosen. This issue was fixed. Also, some issues concerned drag&drop with definition variables, fullscreen mode in Unix, issues with mean structures in the export to sem-package, and issues with reciprocal, asymmetric paths. All these were fixed.
For your convenience, we have added two new features. First, copy&paste now has two flavors: “Paste” and “paste and rename”. The former pastes copied variables and paths with the same name while “paste and rename” makes the pasted names of variables and paths unique. Furthermore, we have a new wizard that allows you to quickly create Latent Differential Equation models. See Steven Boker’s work for more details. You can find the wizard in the menu “Create new model”.
We are happy to announce the publication of Robin Beaumont’s book “Health Science Statistics using R and R Commander” (Scion Publishing; ISBN 9781907904318, which has a whole chapter on Ωnyx. We feel honored and are excited about the daily growing Ωnyx community out there. Thanks for being a part!
We are happy to share with you two new video tutorials recorded by Prof. Timothy Curby (website; twitter). The videos show how multiple group factor analysis and transactional models can be done in Ωnyx. They are freely available on youtube:
We are happy to announce the release of Ωnyx 1.0. We are grateful for your positive feedback throughout the development process.
Among the new features in this release are:
– improved optimization engine
– external runners allowing to run OpenMx and lavaan as backend
– redo after undo
– improved handling of means modeling (by default, Ωnyx assumes a saturated means model)
We are looking forward to your feedback on the new version and we wish you a happy modeling time!