Analyse-it blog The latest news on new features and software releases.

19-Jun-2008 Analyse-it 3.0 now in development

One of the primary reasons we launched the blog is to let you know what we’re currently working on, and give you to opportunity to feedback and influence the development.

At the moment we're spending most of our time developing version 3.0 of the Analyse-it Standard edition. The improvements are based on what you've asked us to include, and through insights gleaned from recent customer surveys. Improvements and new features will include:

  • Test advisor to help choose the most appropriate statistical test – handy, if your statistics knowledge is a bit rusty.
  • New and improved statistical tests – including logistic regression, improved ANOVA with post-hoc tests, new regression fits.
  • Simpler dataset setup – without the need to set measurement scales before running a test.
  • Transform variables, and easily choose groups to analyse, when running a test.
  • Clearer hypotheses and automatic interpretation of p-values.
  • Improved help with step-by-step instructions for each test.

Version 3.0 will be a major upgrade and will be released later this year. As per our upgrade policy, you'll get the upgrade free if you purchased your licence within 2-years of the release date – so if you're thinking of buying a licence now you will get the upgrade free. If you don't qualify for the free upgrade you'll be able to buy the upgrade for a small cost, necessary to cover development costs. We'll announce final pricing and upgrade pricing nearer release.

In the next few weeks we’ll post more about the new features, including screenshots, so you get a chance to comment on individual features. We'll then incorporate any requests and suggestions into the final version before release.

Start now by posting a comment to let us know what you’d like to see included in Analyse-it 3.0.

Comments

I would be happy to participate in any planned product testing. I've been an Analyze-it user since 2001. Best of luck. Stephen Gervais
Thanks for the offer to beta test the new version Stephen.

As we did with Analyse-it v2, we'll release Analyse-it v3 to a few PC savvy beta testers first. We test Analyse-it on all versions of Excel and all versions of Windows, but this stage lets us quickly identify and resolve any issues found "out in the wild" on the myriad of PC configurations out there. Limiting this phase to users with the most PC expertise means we can fix the issues quickly with users we can converse with in technical/programming oriented language.

After the first stage we'll expand the beta test programme to a wider audience to check for any minor cosmetic issues, presentation issues, get feedback on the new features, help, etc..

When v3 is ready to go to beta we'll post an appeal for testers to this blog. Stay tuned!
Comparing Analyse-it with the beloved but know defunct StatView software, I miss any survival analysis tools, such as Kaplan-Meier-plots, and logistic regression of survival data.
Thanks for the suggestions Friedrich.

Good news! Logistic regression will be included in version 3.0 of the Analyse-it Standard Edition, due for release in the next 2-3 months. And Kaplan-Meier plots will be available toward the end of this year.
It could be great if weighted (numbers, square root of the numbers and/or other weighting factors) lineair and polynomial regressions could be possible. Also calculation of the standard error of an estimated Y for a given value of X (both for linear and polynomial regression) would be appreciated.
Analyse-it seems to be efficiently set up to perform multiple tests on a sample. It is not well structured to do one test on many samples. For example 'Describe It' provides all sorts of descriptive stats on one sample. Determining the normal destribution when you have 100 samples is a real test of patience and perserverance. Hopefully the new version will make task like this easier.
Hi George,

A few customers have asked for this feature before so you'll be pleased to know it's something we're looking to include in Analyse-it 3.0.

We're thinking of changing Describe (descriptive statistics) so you can select multiple variables at once, then Analyse-it will generate multiple descriptive statistics reports -- one for each variable. That would save you manually having to create the analysis reports one at a time, for each individual variable.
Thanks for your suggestions GVN.

We haven't considered weighted regression so far, but will definitely look into including it. However, we are planning to include the ability to easily calculate predicted Y value -- with a confidence interval and standard error -- for given X's.

For what application are you interested in using weighted linear/polynomial regression?
Calculus of predicted Y with SE and CI (not only in a figure) is interesting for regressed median marker estimations during pregnancy,with X= gestational time (days or weeks) and estimated Y= regressed median marker concentration (hormones). These estimated values can be used in Down syndrome screening for calculus of the so-called MoM (multiples of median) values.
The post is very nice. The blog is good, so our knowledge of the program will increase. Thank you, this is a good service.
Thanks for the suggestions Guido, and your appreciative comments of the blog. So far it seems Analyse-it users are finding the blog very useful, and we're getting good feedback through the comment facility.

You'll be pleased to hear that Analyse-it 3.0 will provide you with predicted Y, SE and CI's for X values you choose. A table will show the statistics for each X, so you can see the numerical values rather than trying to interpret/guess them from the lines on the scatter plot.
I would like to know if Analyse-it will include the Clarke diagram to split the correlation area between a reference method and a second method into 5 zones according to the clinical weight of possible differences.
Thank you for your help
We don't have any plans to include the feature you describe in the Standard edition of Analyse-it v3.0.

I have seen studies of the type you describe quite often in the medical statistics books such as Altman & Bland and Armitage & Berry. We will give it some more thought, as it may be something we can include in a future edition of Analyse-it.

Comments are now closed.