I have seen a lot of discussion on this topic but it all seems to be opinion based without any research backing up the assumptions.
It seems that there are times that it is particularly relevant to open links in new windows (providing help or interrupting processes) but what about the more ambiguous actions (i.e. links to external sites)?
Does anyone know of any (preferably recent) research on the affect links opening new windows has on users?
It's a heated topic to be sure, but the best insight comes from understanding the usage and intent of a user clicking a link and specifically how this relates to the work at hand and what is going on in the user's mind (i.e. what they are trying to achieve by clicking the link and how that relates to their workflow).
In general, if a link results in leaving the current focus of attention, it should open in a new tab. In contrast, if a link progresses the current locus of attention it should open in the current tab. This is often "simplified" (or bastardized) to the rule that if a link is to an external site it should open in a new tab, but this rule doesn't work as well in all cases, nor does the inverse where all links to the same site should open in the same tab.
Some examples and justification:
Google is a search engine. It's purpose is to find a page and navigate to it. This all happens on a single thread of thought—a single locus of attention. Even though the link is necessarily to an external site, it's still for the most part following a single natural linear progression, and perfectly captured by the linear back/forward navigational paradigm.
Similarly, Wikipedia is a wealth of information, and can often result in clicking through long chains of links to other Wikipedia pages. In general, this can also be considered a single thread of thought that may evolve and change, but is still largely linear.
On the other hand, (I know we all love them) but consider ads. Ads are almost always tangential thoughts. They represent something that you weren't expecting to want to navigate to, but were compelled to for whatever reason. This represents a branch in the user's original workflow. They didn't come to our page in order to click on an ad, therefore if the page did not open the ad's target in a new tab, we would be impeding the original workflow of the user. (Ironically, this right here is a perfect example of when a link should open in a new tab; see my argument below.)
When the focus of attention branches in such a way, the best user interface available is to parallelize the interaction by splitting the traditional navigational stack into a pair of independent stacks (i.e. a new tab or window). Of course each of these new windows represents its own linear progression.
The beauty of this is that when a tab maps exactly to a linear progression of a single train of thought, closing the tab becomes precisely analogous to dismissing that train of thought. Here's an example: suppose I'm reading a forum post wherein a user makes mention of some piece of information that is corroborated by a reference link. As a tangential thought, I may wish to verify their claim before continuing to read the rest of their post. Clicking on a reference should in my opinion always open in a new tab, because it quite clearly represents a branch in my original train of thought (which was to read the user's post).
When I'm done verifying the claim, I can simply close the tab because I'm done with that thought—with that focus of my attention, and lo and behold my parent train of thought reappears.
I can't say how frustrating it is when I click on a link that clearly represents a tangential thought only to have it clobber my previous train of thought by navigating in place. All of a sudden I now have to "back up" (through a completely unrelated history) to get back to what I was doing previously. What if I clicked around a bunch on that site? I may have to back up through three, four, five pages or more to get back to my previous train of thought.
(Things get even worse if I don't notice it happen, because it is very easy to close out that tab and completely lose my original place. All I can say is thank goodness for Cmd-Shift-T in that case.)
Now the tricky part, of course, is deciding if a link represents a tangential thought. The true answer is we simply can't know 100% of the time how a link relates to the user's train of thought, because we can't see into the mind of the user. However, in many cases we can make a pretty good guess, and as user interface designers we should design our websites in such a way as to make the best guess given the circumstances.
Should it always be up to the user to decide whether a link represents a tangential thought, and thus imposing the repsonsibility of command-clicking (or long-tapping—an operation that may not even be known by novice users) the link on the user, or should the website offer some convenience by trying to make a guess? Herein lies the real debate.
In my personal opinion, if there is a high probabilty that the link represents a tangential thought (such as with an ad or a reference), it should default to opening in a new tab. However, if it's unclear how the link relates to the reader's train of thought, the default should always fall to opening in the current tab, leaving the onus to fall to the user.
This question has been asked many times and many answers have been given, in this site and in many others. I've seen the answer change a bit during all those years, but mostly, all of them point to not opening new tabs/windows unless it is an external site or a document that is not intended as a web page, like a .pdf, .doc, .ppt, etc.
Some of the advancements in technology, like tabs on a browser, have given arguments to the proponents of opening links externally, the basic argument is that a tab is not as intrusive as a new window and is not blocking or hiding anything. On the other hand, privacy and control advocates state that it still their browser and no one should decide what to do for them. We can add to that last statement that UX concerned people say that the right option is to give users the option to decide what to do, at least on cases where the decision make sense.
In some places you may find references to articles, blog posts, references to studies or strong opinions, like Nielsen's, etc. But no study, by itself seems to be found.
Studies about the subject:
There is only one study that I have found that mentions something related to this subject, 2006-04-13 - Eelco Herder - Ph.D. Thesis - Forward, Back and Home Again Analyzing User Behavior on the Web.. Specifically in section 6 of that study, we can read that most of users on the studied group used new windows and tabs for different tasks, originating that task on a search engine or a specific page. This by itself, doesn't help the question asked here, but the analysis of the situation, tells us that users prefer to open some links on new windows or tabs when they think it's important to compare information, read carefully the linked page, keep the original page on sight or keep the navigation history intact to be able to go back to a previous point.
The study also mentions that even when users wanted to do that, it also loads the brain with extra work, having to remember how they got to that tab/window, having to keep the navigation history, including the branching, on their minds and having to deal with the little help provided by page titles to do that.
It's an interesting study about certain behaviours, not just this subject. The studied group is small and they consider themselves as advanced users, so, as usual, any result has to be taken with proper judgement and be adapted to each situation's circumstances.
Other very common references, but not studies:
1997-10-15 - When to Open Web-Based Applications in a New Window on Nielsen Norman Group. You may also find this article referenced in many sites but in two ways, the old Nielsen usability site, (useit) or the modern Nielsen Group. It's the same article. This article is the most referenced, it doesn't show any data, but we always assume that all the articles from Nielsen are written considering the experience gathered on his multiple usability studies. By the time this article was written, there where no tabs, so all the references and experiences it's based on imply windows. This article gives what is the most common opinion; open links on the same window unless it's something that is not intended for a web browser.
But it's important to realize that this article seems to be based on a very specific situation that years, people and applications have expanded, Nielsen is talking, basically, about applets, their behaviour was very similar to what you would expect from a Flash application, or embedding a word document on a browser now, but with some differences, like applets where used, sparingly, for very specific and specialized tasks, never for amusement or just design, they did look like the web page they where on and the load on the browser and client system was significative. People usually miss that fact, which is relevant and is easy to realize if you are old enough or if you read the linked reference, Navigating the Applet–Browser Divide; in which you can read an addendum from 2007 reminding us about the specific circumstances of the study, but also, telling us that the study and conclusions are still valid.
2000-11-06 - do not change the current window without informing the user on Techniques for Web Content Accessibility Guidelines 1.0 on the W3C. Not a study, but a basic set of recommendations on how to use related mechanisms.
2005-08-29 - Open New Windows for PDF and other Non-Web Documents on Nielsen Norman Group. This article is also cited many times and is much more related to modern times and circumstances, it states what is the most common idea, use new windows/tabs for external documents. The article is focused on documents and media related to specific applications; like .pdf documents on a .pdf reader, .doc documents on a word visor/editor, etc. I addresses the UX aspect of the situation, stating that a different experience has to be on it's specific environment. Again, we assume the base for this article is extensive study, he cites one study on 42 intranets, which you can buy on his site.
2011-01-01 - Top 10 Mistakes in Web Design on Nielsen Norman Group. This updated list from Jakob Nielsen mentions as item #2 having .pdf, and that implies other similar documents, as elements on the browser, recommending to download them, which is related to opening a new tab/browser since another option, instead of having the same or different window/tab is to offer downloadable material. Also, as #9, is a more direct reference to the subject.
2012-31-01 - Why External Links Should Open in New Tabs on UXMovement. This article doesn't claim to be based on any studies, the author supports the idea of opening external links on a new tab, as you could infer from the title. The comments section, is a mixture of all the know opinions, but the vast majority consider that a wrong procedure, the argument range from computer illiteracy, lack of control and disabilities (or the software used to help people with).
2013-09-05 - G200: Opening new windows and tabs from a link only when necessary on Techniques for WCAG 2.0 from the W3C. Not a study but a recommendation. The original version is older than 2013, but I mention the current edition. All the processes on the W3C are lengthy and bureaucratic, but most of the people involved have years of experience on different areas, so if something has reached the level of recommendation, there must be a good reason for it, we may not agree, but we almost positively can say that some studies are involved.
????-02-01 - Beware of opening links in a new window on webcredible. The author of this blog post, who should be good at usability, communication and UX professional, doesn't understand the concept of dates, so we don't really know where the article was written, but considering that he only speaks about new windows, it should have been around 2008. The post gives good usability reasoning for not using new windows/tabs as well as the usual recommendations for when to do it. The only reference they mention is the 1997 article from Jakob Nielsen that most people use.
[????-??-?? - section 10:12 Indicate Internal vs. External Links on usability.gov]. This is just a recommendation, but in the references section mentions some studies that may have relevant information. If somebody can find them and get permission to publish them or at least part of it, we may have more references.
Some other references or related material. Still not research, just articles and opinions based on the same as above.
I'm a proponent of it's ok to open outgoing links in a new window, but I can't find any arguments/data one way or the other within the last 4 years that doesn't reference 4 year old dogma. In the absence of recent analysis I rely on "what do the experts do" and this is what I've found:
sensible.com (Steve Krug) - generally doesn't open outgoing links in new window but link to amazon.com to buy books does open in a new window/tab.
cooper.com (Alan Cooper) - does open outgoing links in new tab/window.
www.nngroup.com - does not open links in new window.
google.com - does not open links in new window.
news.google.com, twitter.com, most emails, most forums - does open outgoing links in new window/tab.
wikipedia.org, stackexchange.com - does not open outgoing links in new window.
edit: found a recent analysis (courtesy of 3nafish): http://uxmovement.com/navigation/why-external-links-should-open-in-new-tabs/ and I agree completely.
I previously worked for a company sent out roughly 1 million subscriber emails an hour to people subscribed to bulletins, should those users choose to respond to those emails they would click a link to go to our web site where they would continue on their journey.
After much research, as every response was effectively money in the bank we came to one simple conclusion: In code ... never!
And here's why:
Where something renders (same window, new tab, new window) should always be a contextual choice the user should make allowing them to actively make the choice opens up their options and doesn't enforce a particular way of working.
Should you wish to render a dialog on your web page that should be a scripted dialog that renders as part of the page.
The net result of these decisions was a roughly 5% increase in responses on an already high response rate when no other factor changed (A/B testing).
Side Note:
You can do this testing yourself to prove the concept if you wish, our testing covered roughly 2 million unique users over a 6 month period.
It's debatable. I think it's courteous to open a new tab for an external link, but it doesn't make sense on mobile.
We're looking for long answers that provide some explanation and context. Don't just give a one-line answer; explain why your answer is right, ideally with citations. Answers that don't include explanations may be removed.
On today's web, never.
Once, opening a link with href = blank was a legitimate way to present a modal window for a tangential task, keeping the host page in view so the user could easily return to their original workflow once they were done. Granted, the fact it broke the back button probably negated the benefits, but there was still a case for it nevertheless.
Fast forward to 2013 and things have changed: modern desktop user agents open these links in new tabs, and users have started using mobile browsers that hide away tab management for the sake of screen estate. Because the whole page changes, desktop users are prone to miss tab changes due to change blindness, and because the tabs are tucked away on mobile, mobile users find it awkward to shift tabs even when they know they need to.
So don't do it in future. If you need a modal panel, try to use a lightbox solution instead.
I think that the key word is permalink or something along this lines that reflects the same concept.
Talking about this, as a "flow" without considering the possibility that the URL alone sometimes it's not enough to re-create the same content - or the same experience if you will - it's probably pointless, at least it's pointless to me.
As an example consider the typical web layout based on infinite scrolling, which is a popular solutions for blogs or pages that are usually full of content; now consider that you reach a point during the scrolling, if you open a link in the same window, you are likely to never reach that position again when you will go back to that very same page. In this situation opening a link in the same window will result in a loss of information that it's likely to be unrecoverable and will break the "flow" .
A notable exception, as someone else pointed out, is Google, if you use the image searching engine, you will notice that it's using an infinite scroll layout, but you can open a link and go back at that very same point thanks to the information packed in the URL, in other words, you got a permalink that doesn't just takes you at a really specific website or page, it also contains enough information to recover your initial experience.
Another exception is about how the site works and the technologies associated with it, for example if you are using flash based content in your website, you are likely to being unable to recover the very same experience unless you really try to code and elaborate on how to store the valuable information .
And even if you can do that, you probably have a good reason to don't do that, Youtube does offer this option, if you use the URL as a permalink you will be able to reach the very same video, but at the same time you can also infer a time coordinates to the very same URL; at the end Youtube leaves this option to you, how to pack your own permalink it can even be considered a "software feature", but it's clear that the default behaviour is conservative in a way that the URL will always bring you to the same page, with the same content, but not really the same experience.
For me, the only legitimate reason to open a new window if there's a need to have two windows open at once.
but what about the more ambiguous actions (i.e. links to external sites)?
I haven't heard any good arguments for that.
So barring a good argument to change default web behavior, I tend to encourage people to not mess with the default web behavior.
