reading/scanning - tables the facts

21 Sep 2005 - 11:29am
9 years ago
4 replies
552 reads
Ola Thörn
2005

Many kinds of information is best presented in tables.
For example information that we want to compare on
different attibutes e.g. financial information.

But also information about e.g. songs in an mp3player
is good to present in a table allowing sorting on
different attributes.

My question is - what is the ultimate design (shading,
horisontal, vertical lines, no lines)of a table from a
reading point of view.

I remember that Tufte suggested that vertical lines
between columns was inappropirate - since "lines" was
formed by the info in the columns (like below).

48 87 96
48 87 96
48 87 96

However I think this conclusion was based on analysis
rather than empiria.

Does anyone have any hardcore (empirical) facts from
perceptual/cognitive/human factors psychology?

For example a study where reading speed has been
compared on different table designs. Perhaps any
eye-tracking study?

Any differences between tables in print and on screen?

Regards / Ola Thörn

Vänligen Ola

Comments

22 Sep 2005 - 6:38am
Ben Hunt
2004

I don't have any hard data, but common sense would say that Tufte is correct
- to a point.

For small tables, like your example, the theory holds true, and the columns
& rows are quite readable.

The effect breaks down when you have bigger tables of data (say more than 12
rows/cols), because then you get the problem of scanning across from a row-
or column-header to a particular matching column or row. In that situation,
clear row dividers help to some extent.

What helps more is alternating tones on rows. (This effectively triples the
width of the bounds within which you're scrolling your eyes, because you
have a central "wanted" strip, bounded by two "not wanted" strips.)

I've done designs that use alternating tones on both rows *and* columns, to
make it easier to scroll up & down or left & right, but the tonal changes
need to be kept quite subtle to avoid garishness.

- Ben

-----Original Message-----

Many kinds of information is best presented in tables.
For example information that we want to compare on different attibutes e.g.
financial information.

But also information about e.g. songs in an mp3player is good to present in
a table allowing sorting on different attributes.

My question is - what is the ultimate design (shading, horisontal, vertical
lines, no lines)of a table from a reading point of view.

I remember that Tufte suggested that vertical lines between columns was
inappropirate - since "lines" was formed by the info in the columns (like
below).

48 87 96
48 87 96
48 87 96

However I think this conclusion was based on analysis rather than empiria.

Does anyone have any hardcore (empirical) facts from
perceptual/cognitive/human factors psychology?

For example a study where reading speed has been compared on different table
designs. Perhaps any eye-tracking study?

Any differences between tables in print and on screen?

Regards / Ola Thörn

22 Sep 2005 - 8:08am
Suresh JV
2004

In case of large tables, Grouping 5-7 rows in addition to alternate
Row colors would aid horizonatal scanning.
--------------------------
Dsds 45 744
Asdf 12 321
Wqer 32 987

Pwe 89 987
Vbn 90 564
--------------------------
There is some data on table view usability somewhere which even
Says that people want charts but actually need/use table view.

There are no scroll bars in printed tables. ;)

Regards,
Suresh JV.

22 Sep 2005 - 12:13pm
Rajesh Sidharthan
2005

Unlike print, you can also rely on interactive mechanisms to help scanning.

* Highlight entire row if the user clicks on it.
* Row selection on tab or down arrow...

</raj>

Ben Hunt wrote:

>[Please voluntarily trim replies to include only relevant quoted material.]
>
>I don't have any hard data, but common sense would say that Tufte is correct
>- to a point.
>
>For small tables, like your example, the theory holds true, and the columns
>& rows are quite readable.
>
>The effect breaks down when you have bigger tables of data (say more than 12
>rows/cols), because then you get the problem of scanning across from a row-
>or column-header to a particular matching column or row. In that situation,
>clear row dividers help to some extent.
>
>What helps more is alternating tones on rows. (This effectively triples the
>width of the bounds within which you're scrolling your eyes, because you
>have a central "wanted" strip, bounded by two "not wanted" strips.)
>
>I've done designs that use alternating tones on both rows *and* columns, to
>make it easier to scroll up & down or left & right, but the tonal changes
>need to be kept quite subtle to avoid garishness.
>
>- Ben
>
>
>-----Original Message-----
>
>Many kinds of information is best presented in tables.
>For example information that we want to compare on different attibutes e.g.
>financial information.
>
>But also information about e.g. songs in an mp3player is good to present in
>a table allowing sorting on different attributes.
>
>My question is - what is the ultimate design (shading, horisontal, vertical
>lines, no lines)of a table from a reading point of view.
>
>I remember that Tufte suggested that vertical lines between columns was
>inappropirate - since "lines" was formed by the info in the columns (like
>below).
>
>48 87 96
>48 87 96
>48 87 96
>
>However I think this conclusion was based on analysis rather than empiria.
>
>Does anyone have any hardcore (empirical) facts from
>perceptual/cognitive/human factors psychology?
>
>For example a study where reading speed has been compared on different table
>designs. Perhaps any eye-tracking study?
>
>Any differences between tables in print and on screen?
>
>Regards / Ola Thörn
>
>
>
>_______________________________________________
>Welcome to the Interaction Design Group!
>To post to this list ....... discuss at ixdg.org
>(Un)Subscription Options ... http://discuss.ixdg.org/
>Announcements List ......... http://subscribe-announce.ixdg.org/
>Questions .................. lists at ixdg.org
>Home ....................... http://ixdg.org/
>
>

22 Sep 2005 - 12:41pm
david gee
2004

Ben Hunt wrote:

>What helps more is alternating tones on rows. (This effectively triples the
>width of the bounds within which you're scrolling your eyes, because you
>have a central "wanted" strip, bounded by two "not wanted" strips.)
>
>
>
I really like using alternating tones on rows, while keeping it as
subtle as possible. I think the original poster pointed out that Tufte
recommends against demarcations between columns, as the demarcation
should be formed by the columns. In practice, some grids call for
centering the text in some columns, which can throw this off.

One thing I've done in the past for web-app datagrids is to use a
javascript control which highlights the relevant row and column on
mouseover, creating an intersection of all the relevant data. I've found
that this is really only necessary for large grids with many numeric
columns, and again subtlety is key. Unfortunately, HTML and CSS have
very limited support for columns, and the javascript to do this can be
pretty intensive, so it's not a very scalable solution.

david

Syndicate content Get the feed