Pick One – Multi, Pick Any, Pick Any – Grid, Date, Number and Number – Multi questions can now be selected as splitting questions. The multiple response questions are split and each variable is analysed separately. The numeric variables are split into five approximately-equally-sized categories.

More sophisticated methods are now used in most statistical tests that are performed when data is weighted.  In particular, Taylor series linearization is used when computing the standard errors.   This results in a change in automatic statistical tests (i.e., the tests that determine whether cells on tables are highlighted as being significant or not).  The Q4.1 Reference Manual contains more information on this.  The key benefit of this change to the testing is that tests are more powerful when the weight is correlated with one or more of the variables used to construct the table (i.e., selected in the blue and brown drop-down menus).  There is an additional benefit for agencies that work with government clients: the testing will generally be more consistent with the way that testing is conducted by government statistical agencies.

There have been two additional changes to statistical testing.   It is now possible to specify a minimum sample size used in testing (in Edit | Project Options... | Statistical Assumptions).  And, corrections for multiple comparisons now ignore any cells in a table where no p-value could be computed (most commonly, this is for columns containing no data).

In addition to the standard automatic tests of significance, it is now possible to show comparisons between different columns on a table.  Select Statistics – Cells and Column Comparisons to see letters showing which columns are significantly different.  As with the standard tests, the False discovery rate is automatically applied.

Sample sizes can be placed on the bottom of columns or on the right of rows by selecting Statistics – Below and Column N or Statistics – Right and Row N.  The Row N and the Column N are now computed based on all the non-missing data in that row or column (previously, with multiple response questions which did not add up to 100%, the Row N and Column N reflected the total number of respondents with data).  This change has been requested by many users.

The reporting of Base n on Date questions has been modified to be consistent with the presentation with other styles of questions.  Previously, Base n referred to the total number of observations for the time period being displayed; now, it refers to the total number of observations in the data (Column n and Row n will reveal the number of observations in the time period).

Where a table involves a Date question, you can now use moving averages and local polynomial regression to smooth the data.  The Q4.1 Reference Manual contains a new section which describes how this is done and which also highlights other tools in Q for analysing tracking studies.

Q4.1 now supports 64-bit processors, which are becoming the norm when you purchase a new PC or laptop.  The advantages of 64-bit support for Q includes:

• No out of memory problems with big tables or big data sets, compared to a 32-bit limit of around a million cases.
• Q can use as much RAM as your computer has, compared to a 32-bit limit of around 2GB.

To take advantage of Q's 64-bit support, you must be running a 64-bit edition of Windows Vista or Windows 7.  Here is a guide from Microsoft to check whether you are already using a 64-bit edition of Windows.

The only disadvantage of 64-bit support is that the statistics for Experiment tables and the results of Segmentation analysis differ from what you see on a 32-bit edition of Q by around 1%.  These differences occur because 64-bit Q can perform calculations with more precision than 32-bit, meaning that Q effectively has access to more decimal places, and this can cause some differences with Experiments or Segmentation because of the many complex calculations involved in computing their results.

Another trend in computing is the uptake of dual-core and quad-core processors.  Recognising this trend, we have initially made Q's most expensive calculations support multiple cores: Experiment and Ranking statistics.  This means when you view an Experiment question in a table, or compute its Segments, Q can be up to twice as fast if you have a dual-core processor, or up to four times as fast if you have a quad-core processor.

Q5 will have even more areas targeted for multi-core support - please tell us if there is a particular area of Q that is slow for you and we will make it part of the plan for Q5.

Different versions of a question, such as where a sample has been split or when question wording has changed from one wave of a tracker to another, can now be automatically merged by selecting the questions in the Variables and Questions tab and selecting Insert Ready-Made Formula(s) and Merge Questions.  The method is designed to deal with situations where code frames differ between the two questions and even to deal with changes from Pick One to Pick Any questions or vice versa.

Projects can now be merged. This is done by opening Q twice and opening the projects to be merged in each of the Q windows. Variables and entire questions can be dragged from the Variables and Questions tab in one project to another. Tables can be dragged from the report tree in one project to another. You can even drag tables and variables between projects with different sample; Q will automatically update them to take the new sample into account.