Q vs Displayr: Differences and Similarities

From Q
Jump to: navigation, search

Q and Displayr are the two main products developed by the company, Displayr.

What the products have in common

  • The basic computational engine is identical.
  • All statistical analyses are identical.
  • All chart types and visualizations are identical.
  • Each is 100% compatible with the other.

Intended uses of the products

Displayr Q
  • Data science: novices through to experts
  • General-purpose statistics
  • Online reporting/dashboards
Analysis of surveys by survey research specialists (e.g., market researchers, pollsters)

Differences in features/capabilities

The basic user interface of the two products differs. In Displayr, the user creates a document, and customizes the content of the pages. In Q, the user creates tables and other analyses, with one analysis per page.

Things that Displayr can do that Q cannot do Things that Q can do that Displayr cannot do
  • Lay out multiple outputs on a single, PowerPoint like page (e.g., two tables, a chart, and text)
  • Publish data as an online report/dashboard
  • Automatically update data (e.g., every hour)
  • Multiple people working simultaneously on the same data
  • Accessible from a web-browser (Q is a Windows app)
  • When applying filters, an OR operation is used where variables are grouped in the same Question (known as a Variable Set in Displayr). In Q, when applying filters they are always an AND operation when selecting multiple filters on the Tables tab, although OR can be specified at the time of creating the filter.
  • Choice Modelling Simulator that can be accessed online
  • QScripts
  • Ready-Made Formulas
  • Rules. These still work in Displayr, but cannot be edited.
  • A graphical user interface for creating complicated filters (complicated filters Filtering is 'OR' where categories have been grouped into the same Question (known as a Variable Set in Displayr).
  • Large-scale latent class and mixture model models (Displayr has some latent class features, but cannot run analyses that take more than 2 minutes to compute)
  • Bigger Data. Longer-term, Displayr will be better with large data files than Q, but for the moment files of more than 50MB can cause problems in Displayr
  • Create weights through a dialog box (in Displayr, this requires writing code from scratch except for simple weights of one variable).
  • Detailed fine-tuning of statistical assumptions (e.g., specifying the overall significance level).
  • A graphical user interface for creating complicated filters (see Binary Variable). This is done by writing code in R or JavaScript when using Displayr.

Optimal workflow for using Q and Displayr

One-off projects

Where there is a need to use Q and Displayr for a one-off project (e.g., analyzing the data in Q and then creating a dashboard), the basic workflow is to do all the analysis in Q, and then switch to Displayr to layout and publish the dashboard.

Ongoing projects (e.g., trackers)

Where there is a need to use both Q and Displayr as a part of an ongoing project (e.g., a tracking study that gets updated quarterly), the optimal way to do this is to:

  1. If it is a large market research project with extensive data cleaning and tidying requirements and you have a team that already knows how to use Q, perform the initial setup in Q (as it has more features for data cleaning and setup). However, do not create any reporting at this stage. Note that if you are not already using Q it is better to start the project in Displayr, as it is easier for the users to learn one new product (i.e., Displayr, rather than learning Q and Displayr at the same time), and it supports multiple users working simultaneously.
  2. Upload the QPack to Displayr and, from that point onward, treat Displayr as the "host" (i.e., update data from within Displayr).
  3. Use Q for special-purpose analyses on an as-needs basis, as follows:
    • Advise users that the document is being maintained, and that they should not make any modifications. This is because any changes made in Displayr after it is downloaded, will be over-written when the QPack is uploaded.
    • Perform the required work in Q. For example:
      • Adding rules to tables (e.g., applying Rules for hiding data with small numbers of rows/columns)
      • Creating complex filters or weights
    • Upload the QPack to Displayr, replacing the current document.

There are a few reasons for treating Displayr as the "host":

  • Displayr documents can be edited by multiple people at the same time.
  • Displayr documents can be exported as dashboards, permitting them to be viewed by people that do not have licenses.
  • Any work done in Q can be viewed in Displayr. By contrast, any pages created in Displayr will be split up and presented as separate outputs in Q.
  • Displayr can automatically update with data and republish documents.