Decision Tree vs Decision Table

11 Aug 2011 - 6:07pm
3 years ago
6 replies
5087 reads
Chris Rink
2009

I am currently working on a project where we need a method of creating conditional logic and execution. We are currently stuck between two types of implementations.

The first is more of a decision tree implementation. Each condition has a yes and a no clause and the statements can be nested. (decision-tree.png)

The second is more of a decision table. There is a set of rules that contain a condition and action combination. Each rule is evaluated sequentially and the first one that applies is used. There is always a default rule in this case. This is commonly seen as a nested if then else. (decision-table.png)

Has anyone done any usability testing on this type of condition building before? If so, what was your experiance?

AttachmentSize
decision-tree.png35.63 KB
decision-table.png100.36 KB

Comments

11 Aug 2011 - 9:05pm
GeoffWill
2010

Chris

I did usability and design work on a decision oriented rules kind of UI for MS CRM (now dynamics). We did two. The table version did not do well. This had the conditionality down the left side and expressions were formed within the row with various list boxes that contained alternative terms.

A better ui was a kind of tree with rules built into the branches. This was for determining what kind of people and equipment and something else were needed for particular situations in a medical UI. This worked better, easier to visualize.

Hope this helps, it was years ago. I think the table lacked the continuity that the tree and phrasing structure provided.

geoff

-----Original Message----- From: Chris Rink Sent: Thursday, August 11, 2011 6:19 PM To: gwbando@msn.com Subject: [IxDA] Decision Tree vs Decision Table

I am currently working on a project where we need a method of creating conditional logic and execution. We are currently stuck between two types of implementations.

The first is more of a decision tree implementation. Each condition has a yes and a no clause and the statements can be nested. (decision-tree.png)

The second is more of a decision table. There is a set of rules that contain a condition and action combination. Each rule is evaluated sequentially and the first one that applies is used. There is always a default rule in this case. This is commonly seen as a nested if then else. (decision-table.png)

Has anyone done any usability testing on this type of condition building before? If so, what was your experiance?

12 Aug 2011 - 5:05am
William Hudson
2009

Chris -

No actual usability testing, but devices like logic and protocol analysers have been using on-screen decision trees for years. Naturally, it depends a lot on who your audience is how complex these can become and still be intelligible.

Sometimes the decision trees are vertically formatted, more like a decision outline or the nesting of conditionals in a program listing.

I wrote a parser and visual display using a spreadsheet-style approach many years ago with nesting of conditionals. Users could select the arguments from drop-downs. The project didn't include usability evaluation (sigh) but we didn't receive too many complaints.

Regards,

William Hudson Syntagm Ltd User Experience Strategist UK 01235-522859 World +44-1235-522859 US Toll Free 1-866-SYNTAGM mailto:william.hudson@syntagm.co.uk http://www.syntagm.co.uk skype:williamhudsonskype twitter:SyntagmUCD

Syntagm is a limited company registered in England and Wales (1985). Registered number: 1895345. Registered office: 10 Oxford Road, Abingdon OX14 2DS.

UX and UCD courses in London and Brussels www.syntagm.co.uk/design/schedule.shtml

12 Aug 2011 - 4:52pm
Chris Rink
2009

First off, thank you to Geoff and William for replying. Your experiance was very helpful.

To understand this a bit more, I did a quick usability test. I tested perception of the decision tree and decision table conditions statements as well as a test to find a particular condition. The results were interesting.

  1.  In general, people liked the look of the nested tree more than the un-nested table.
  2. However, in general people felt that the nested tree was a bit harder to understand than the table.
  3. Number two was backed up by the “quiz”: people were twice as likely to make a mistake interpreting the tree as they were interpreting the table.

 

This appears to be a general flaw with any nested condition based UI. When created nested conditions, the order of the condition causes abiguity of when the condition will apply. 

Thanks again.

15 Aug 2011 - 7:33am
Brian Mila
2009

Chris,

I would love to know a little more about your test setup and your users.  I've got a similar design problem, perhaps we could brainstorm on some new interface ideas?

Brian
brian dot mila at trizetto dot com

15 Aug 2011 - 2:29pm
Chris Rink
2009

The test setup was this.

Participants: A group of Sales Engineers that would commonly use this type of interface. These are technical people that build proof of concept applications in order to sell to a customer.

Test Methodology: Survey using Survey Monkey. We had two approaches that we asked the participants to evaluate.

Questions:

 

  1. Given a simple condition scenario. Do you prefer design A, design B or do you have no preference.
  2. Given a complex condition scenario. Do you prefer design A, design B or do you have no preference.
  3. For design A in the complex condition scenario, what action cooresponds to condition X?
  4. For design A in the complex condition scenario, rate (1-5) the understandability of the design A.
  5. For design B in the complex condition scenario, what action cooresponds to condition X?
  6. For design A in the complex condition scenario, rate (1-5) the understandability of the design B.

 

I also had some qualitative questions and requests for additonal comments but those weren't as relavent for the decision.

Notes: Another question that I was thinking of doing but didn't was:

For design A in the complex condition scenario, what condtion cooresponds to action X?

Hope this helps,

Chris

 

15 Aug 2011 - 5:05pm
a2slbailey
2010

Thanks for sharing your insights and methodology. This has been an interesting discussion so far, however I think it's important to distinguish that this was not a test of the usability of the 2 designs, it was a survey of potential user's perceptions and preferences. While perceptions and preferences can be valuable inputs into design, they shouldn't be confused with usability or with the results that will be achieved by conducting usability tests using prototypes.

sb

Syndicate content Get the feed