Sunday, August 12, 2012

How are libraries designing their search boxes? (I)


Today academic libraries offer users a wealth of search options including

1) Classic catalogues
2) Next Generation Catalogues
3) Discovery services
4) WorldCat/Consortium catalogues
5) Google Scholar supported with library links
6) Individual databases
7) A-Z journal/database lists

and if one considers that the user might be searching for the library website pages for policy related questions

 8) Site searches

Assuming one believes that the portal should be search centric (there are dissenters who think there is no need for a embedded search box on the main page but instead links off the main page to various search tools is sufficient) so there should be a search box on the homepage, the question is how should we design our search boxes?

One search box vs multi-tab vs multiple search boxes vs no search box

There seems to be four main options

 1) One search box (but with options via text links, radio buttons, drop-downs)
 2) A multi-search tab box
 3) Multiple search boxes on home page
 4) Where there are no search boxes at all on the main page.

What follows is what I saw on library pages that are using Summon. How I selected them is mentioned here.

Type of Design No of libraries %
One Search Box  44 35        
Multiple Search Tab 64 (including 4 "Vertical style"  51
Multiple boxes on Search page  7 6
No search box 11 8
Total 126  100

I removed a couple of cases, I couldn't confirm if they were using Summon (perhaps as a beta). Overall a slight majority have multiple search tabs, while a not insignificant minority just went with a One search box.

A very small number (7) had multiple boxes on the home page and slightly more had no search boxes at all. Since these are in the vast minority, I won't analyse them further.

It's also somewhat interesting to compare with this 2012 study  “Discovering” what's changed: a revisit of the OPACs of 260 academic libraries , though that study was on the state of affairs between September of 2009 and July of 2010.

Also sampling was different, I sampled libraries that self identified themselves as using Summon on lib-webcats, while the study random sampled from Peterson's Four-Year Colleges, 2009 so it included libraries with no web scale discovery product.

The main finding in that paper that 96% libraries with Discovery services still provide access to legacy catalogues remains true. Of the 51% of libraries in my sample using multi-search boxes, pretty much all of them provide a tab for the classic catalogue.

Of the libraries with one search boxes, the majority do provide a text link to the classic catalogue.

But let's look at each category in turn.

1. One Search box
1a.One Search box - minimalist design with text links
1b.One Search box - more complicated design with radio buttons, check boxes and pull-downs

2. Multiple tab

1. One search box

As I was looking at Summon libraries, when the choice was made to keep with one search box, the option was usually Summon obviously. Such a design rather than going with tabbed approach is probably based on the observation (though statistics might prove otherwise) that most users would just stick with the default anyway?

It make sense that such a approach would go with a minimalist style and the ones in this section generally have a clean one search box with perhaps one or two text links.

1a. One search box - minimalist approach with text links

Most libraries that adopt Summon usually include at least 2 links, a "About" link and a "Advanced search" link, however some have adopted a even more barebones approach.

University of Michigan - Dearborn has probably one of the simplest designs possible.

Just a simple "Go" button , no other links.

I am not sure what/if any association it has with the University of Michigan which has a very sophisticated build of Summon using the API + Drupal.

The University of Canterbury libraries just has a single search box with no additional options. The "How do I Find?" option on the right exposes help files on LibGuides.

Do note that further down the page there are links to catalogue (under Resources heading) but one wonders how many people actually see that as opposed to generally eyes being drawn to the search box.

Bahrain Polytechnic Library Learning Centre has a even simpler Search box for Summon with a one search box and the button "Go" that resides on a LibGuide they apparently use as a library page.

The James Cook University Library does not have a link to advanced search but there is a "About" link.

The Edith Cowan University also has a single search box with a text link to advanced search.

Here's a fairly typical example of a Summon - One search box with advanced search link and about link. Sheffield Hallam University Library.

Here's another from University of Wollongong.

Murdoch University has a attractive looking one search box with links to advanced search and about. Unlike most libraries clicking on the advanced search does not send you to the default Summon advanced page but they have taken additional effort to create their own advanced search form with the attractive banner. Very nice work.

Grand Valley State (GVSU) one of the pioneer libraries also has one of the simplest designs. A one box search to Summon, with links to the classic catalogue & databases. Clicking on Databases gave you listings of database including A-Z name list, by subject and by format.

While GVSU is typical of one search box libraries in that it has a list of text links to other options it is somewhat unusual in that it doesn't have a text link to the advanced search option to Summon, nor a "What is Summon?" or help type link. Either option is pretty popular on many libraries.

Matthew Reidsma at ALA 2012, gave a talk "GVSU and Summon at Three Years" [video] traces the evolution of the GVSU website from a tabbed interface to eventually a one search box and explains why.

1b. One search box - more complicated designs with radio buttons , check boxes and pull-down menus.

One single search box with text links to other options is all very well, but what if you want to allow users to pre-filter their search or change modes without been sent to another page? That's where radio buttons and checkboxes come in.

If you include radio buttons you need to consider

  • Number of radio buttons 
  • Type of search
For type of search you need to consider that you are not limited to Summon searches (with the appropriate refinements) but you could send the user to any other library search typically the classic catalogue.

Most libraries do a maximum of 3 radio buttons. 

Victoria University Library does 3 options , everything, book/ebook and journal articles. All three options use Summon only. 

The three options to search "Everything", "books", "journal articles" seems to be a fairly popular design used by Okalahoma State University Library , Auckland University of Technology, New Zealand, Lincoln University, Charles Darwin University Library.

Some libraries go even beyond these 3 radio button options. On top of offering the 3 radio box options, you can use check-boxes to exclude categories of items.

For example University of Tasmania, Australia (UTAS)  , offers not just the 3 radio buttons, but also a check box to exclude book reviews by default.

The University of Waikato  offers the same but excludes even more , newspaper articles + book reviews. The idea here is that in a typical Summon search, newspapers and book reviews may swamp out other results so by default it's best to exclude them. This check box UI is the most transparent way to tell users what is happening before they click search.

Queensland University of Technology (above) goes even further with 3 radio buttons and two check boxes.

Both Online only (some people want access to online full-text only) and in the library catalogue options are useful, though I admit my mind boggles a bit with the potential interaction between the 2 check boxes and the selected radio button.

Most libraries stick to 3 radio buttons but Griffith University offers five radio buttons (below).

The two additional ones are Journals (as opposed to journal articles) and Databases. All 5 options are using Summon as opposed to sending the user to the classic catalogue or E-Journal A-Z lists etc. This surprises me because while there are values in the content type  for Journals (or rather Journals/Ejournals), I didn't realize there was one for Databases or you could add content types!

In fact, I probably going to take a closer look at this library, they seem to have fixed some issues I identified in our version of Summon.

So far the libraries using radio buttons only switch modes between Summon , but of course one can design the search box to allow searching using non-Summon systems.

University of Surrey for example allows you to change modes to search the library catalogue. No doubt this is to allow quick access for users who prefer traditional library catalogue.

University of Sydney  goes one step further and allows searching of the library website.

The above library probably goes the most even radio button leads to a totally different search option including catalogue, library website, google scholar and Summon. I am unsure what Shrops. Health Libraries searches.

Besides check-boxes and radio buttons, a lesser option used is pull-down menus. One example is Brown University Library.

Brown University Library has a pull down menu with many options. Their system seems to be highly customized using a combination of VuFind and Summon to power the Article search portion and of course they also give you the option to search the library website.

It is perhaps important to note even the "Everything" options uses I believe a customizated VuFind implementation to create a "bento-style" results list in particular one popularized by Villanova University

I have blogged about such implementations here  , and I am beginning to think this is a much better approach compared to making users pre-select options using check-boxes, radio buttons, pull-down menus or do post search filtering by facets.

Dartmouth has a somewhat unique design. It is a strictly one box search but when you click on the "More Options" button you get a tabbed search box appearing as seen above.

2. A multi-tab search option 

You may have noticed that some of the single search box libraries have pretty complicated designs with lots of text links, radio buttons, check boxes and pull down menus. At this stage you might think why not a multi-tab search box? This is exactly what the following group of libraries have done.

There was also a study showing that a tab approach compared to pull-down menus tends to encourage users to switch from the default option.

I did an extensive analysis of the 60 libraries using multi-tab boxes recording

i) Number of tabs
ii) Order of each tab and exact label used
iii) "Normalised" the labels

 Here's a snapshot of a subset of  (ii)

The colors correspondence to a general common search functionality

Below shows a "normalized" version, where I changed the labels.

Obviously, this allows me to do a wealth of analysis but this blog post is already very long, so I will just focus on 2 boxes, the Summon search box and the Catalogue search box. The next blog post will focus on other areas.

2a. Position and frequency of Summon use in multi-search boxes

Of the 60 libraries, where do they place Summon?

Summon - Position in tab rowNo of libraries

Here I count something as "Summon" Tab only if it does a broadly unrestricted search of Summon (options to exclude/include relatively minor domains like newspaper articles, book reviews is allowed).

For example, Buffalo State College's "book" tab, which actually uses Summon for a restricted search is not counted as "Summon" (In fact all 3 tabs for this library "Everything", "Articles", "books" uses Summon, but only the first counts as "Summon" for above table). 

Similarly, University of Southern California uses Summon heavily, where the "Everything", "Books", "E-books", "Articles" actually use Summon but scoped to different areas. But only the first counts as "Summon" above.  

Of the remaining 5 libraries that don't have a unrestricted Summon tab they either have a blended/bento style search where Summon could be one of the bento boxes and/or use Summon in a scoped search for articles or other areas. For example College of William and Mary has an articles tab that is really Summon scoped (the default Catalogue appears to be VuFind?), while some use a bento style search either alone or together with Summon.

In any case even though Summon is listed as the first tab in 55 libraries oddly enough this doesn't necessarily mean 55 libraries use Summon as a default.

First there are cases like UNC  while the Summon Tab is listed first (labeled Article+). the default tab is Catalog. 

Then conversely there are cases where the Summon Tab is not listed first but are actually default. 

Drexel University Library lists the Summon Tab second but that's the first option you see when you visit the page. Below shows Drexel University's "Article & More" as default.

Florida Institute of Technology is even more extreme, it lists Summon last out of 6 options but a user visiting the page will by default be at the Summon tab.

But in general as expected most libraries with multi-search tabs, use Summon as the default tab due to the desire to expose a new asset. Arguably the main decision as the default tab is between 

  • Summon (Vanilla version - broadly unrestricted)
  • The Catalogue (whether classic or next generation like VuFind, Encore, Primo)
  • A bento style box

The third requires some technical expertise so the default is usually between Summon or The Catalogue.

2b. Position and frequency of catalogue use in multi-search boxes

Catalogue - Position in tab rowNo of libraries

As shown above (but taking into account the fact that a handful of libraries default not to the first tab), the most common position for the "catalogue" is 2nd. In total there are almost as many libraris with a "Catalogue" tab as a "Summon" tab but most common position is in 2nd tab.

Similar to Summon, I only count broad generally unrestricted searches of the library catalogue (whether classic catalogue or next generation) as "Catalogue".

There are examples of tabs labelled "Videos", (another one here) , Music & Video , DVD & Video , CDs/DVDs , which are restricted search of the library catalogue for such material which are excluded from the statistics above.

The existence of such tabs suggest either a) Such searches are popular or more likely b) Summon isn't particularly good at filtering down to such items using facets?

Taking all these qualifiers into account , the fact that almost every library has a catalogue tab  , supports the findings at “Discovering” what's changed: a revisit of the OPACs of 260 academic libraries that despite having a Discovery service, libraries's still find the need to retain the catalogue option though mostly as a secondary option.

Whether this is out of a need to appease users who are used to the library catalogue interface or because of weaknesses in Summon that make known item searching difficult is something the statistics can't answer. 

This blog post is getting really long, so I will reserve further analysis of the tabs and labels for a future blog post.

2c. Summon tabs in multi-search tabs

A study of the 55 Summon tabs doesn' reveal much, the styles generally similar to what is found at the Single Search box examples above. One would think because such libraries have multiple tabs as opposed to a single box, the Summon tab could be simpler with less options since the options could be covered in another tab.

But that doesn't always seem to be a case. Some examples

2d. Vertical Tabs?

The analysis above about multiple tab search was done on 60 libraries. In addition, there are 4 other libraries that have a multiple tab like interface but the options are listed vertically rather than horizontally. See below.


This was a very long blog post showing different styles of search box, it's unclear to me how useful it was as most of the "Findings" are not unexpected but in the end, you will have to do your own usability tests , log analysis to determine which is best for your users.  E.g University of Texas Libraries , Open University UK, NCSU seems to be blogging , writing papers about their studies of user behavior.

BTW If you want to keep up with articles, blog posts, videos etc on web scale discovery, do consider subscribing to my custom magazine curated by me on Flipboard.

blog comments powered by Disqus

Share this!

Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.
Related Posts Plugin for WordPress, Blogger...