为了回应我们最近的居中徽标的文章之一与左对齐的徽标,一位读者推文:manbetx官网手机登陆“每个@nngroup时事通讯,总结:确切地说是所有其他网站已经完成的东西,或者您的用户将被混淆。”

虽然显然是一个夸张(我们在各种各样的主题上发布文章),但Tweet确实在我们的许多文章manbetx官网手机登陆和用户体验的主要原则之一中确定了一个大主题:一致性。一致性是原始的10可用性启发式和is a corollary of雅各布的互联网使用定律。令人失望的是,设计师必须遵循同样的殴打路径,我们讲述一致性又一次,因为源于基本的人类行为。在本文中,我们解释了这些原因最基本的原因:学习权力法。

学习研究和学习曲线

衡量人们如何学习任务或接口的最佳方式是运行alearning experiment。在一个学习实验中,人们进入实验室并做同样的任务multiple times。每次该人完成任务时,实验者都会记录一个或多个定量度量(通常,执行该任务的时间和错误的时间)。如果措施变得更好,人们从以前的经验中学到了,而且数字显示了多少。相同任务的重复可能或可能不通过不同的活动分开,有时参与者甚至在两次测量之间发送家庭,并要求在一天后,一周或一个月后回来。

One of the first rigorous learning experiments was described by Hermann Ebbinghaus in 1885 in his book on human memory. Since then, many other learning experiments have been reported in the psychology, human-factors, and HCI literature. All these experiments show that “practice makes perfect:” when people do the same task over and over again, they get better and faster. The chart below shows the results from one such study by David Ahlstrom and colleagues, who were investigatingpie menus和comparing them with other types of menus.

AhlStrom等人。的研究与参与者与8个不同的实践块相同的菜单界面进行交互 - 在每个块参与者中,在菜单中选择相同的6个项目,并且所获得的选择时间平均以获得该块的平均时间。学习曲线表明平均选择时间随着实践而减少。

This type of graph that plots the results from a learning experiment is a learning curve.学习曲线描述与时间的函数相同的人类行为的特定定量度量如何变化。在菜单实验中,感兴趣的衡量标准是任务时间 - 平均选择菜单中选项的时间。但措施可以从一个学习实验到另一个学习实验不同:它可以是您希望由于学习而改变的任何指标。例如,如果我们对UX教育感兴趣,我们可能会询问有关相同背景的人,以方便用户测试,并测量它们的促进错误。我们会给他们对这些错误的反馈,然后我们会要求他们第二次来促进用户测试。在几个这样的重复之后,我们将绘制每个测试所做的误差的平均误差。随着时间的推移,错误的图形将代表一个学习曲线。

The Power Law of Learning

在20世纪80年代,Allen Newell是一位着名的Carnegie Mellon认知科学家,分析了学习实验中报告的各种任务的反应时间,他指出,在所有这些研究中获得的学习曲线具有非常相似的形状:所谓的形状:所谓的权力法。权力定律有一个很好的数学属性:当您在日志日志刻度中绘制它们时,您可以获得一条直线。

The learning curve in Ahlstrom’s menu experiment is described by a power law; when plotted in log–log scale, it is well approximated by a straight line.

定义:学习权力法说(1)执行任务所花费的时间随着该任务的重复次数而减少;(2)减少遵循权力法的形状。

Newell主要集中在时间按时作为学习的定量衡量标准,但有证据表明电力法也适用于其他措施。

Analyzing a Learning Curve

虽然可以通过权力来描述学习曲线,但它们将不会被描述相同的权力法。

让我们假设我们有兴趣分析三个不同接口A,B和C的学习性,对于相同的任务(例如,在呼叫中心回答客户查询)。对于每个设计,我们要求参与者在不同,重复的试验中完成任务,我们衡量每次试用的任务时间。然后我们绘制与每次重复相对应的平均任务时间,我们获得了以下三个学习曲线,如下图中的一个学习曲线。

三个不同界面的三条学习曲线

Notice that in the first trial participants take roughly the same amount of time with all designs. But by the second repetition, design A is much faster than designs B or C. By the 3rd.repetition design A speeds up even more, and after the 4th重复反应时间到达高原:曲线变平,用户尽可能多地学习界面​​。预期没有更多的改进,额外的重复只会减少反应时间。我们可以这么说,设计a,learning is saturated after the 4threpetition(or that 4 is the saturation point for design A)

The learning curve for design C also flattens, but the plateau is reached later, by approximately the 10th或11.threpetition. So design C requires more practice to stabilize the performance. In other words, it takes people more trials to learn how to use design C than design A.

此外,设计更改进:学习曲线上的最高点和最低点之间的差异约为20s(22,用于重复15的重复15),而对于设计,该差异是大约19s。因此,设计C参与者并不像与设计A一样加速。

设计B还稍后比设计A(大约重复11)达到饱和度,但改善更大。更重要的是,学习界面之后的预期任务时间对于设计B而言,对于设计A(1S与2S)而言,较低。换句话说,设计B比设计A比设计更长,但一旦学到,人们就会更快地使用它。

A和C之间的选择很简单:一个是更好的every way, with an earlier saturation point, a greater speedup, and a superior task time once the interface has been learned. But the choice between A and B depends on whether in real life users will be exposed to enough repetitions to reach the saturation plateau. If, for instance, you expect people to use the interface every day as part of their work, then it makes sense to go with design B, because in the long run it will save more time. (During the first week of use, A will be better, but halfway through the second work week B becomes better, and then it stays better forever.) However, if your users will use the design occasionally, with large intervals of time between two different sessions, then design A will be better because it will help people learn the interface faster.

Let’s consider two enterprise software examples:

  • 员工目录:假设大多数人每天使用一次,我们应该更喜欢用户界面,如设计B用于此类应用程序。
  • 在费用报告中回收外国旅行的增值税:如果大多数人在每年出国的大多数商务旅行,我们都应该更喜欢设计a的用户界面。

改进和饱和点之间的比率表示学习曲线的斜率: if the curve drops a lot and fast, then it means that the interface is highly learnable. If the curve drops only a little, or it takes many trials to reach the saturation point, the interface is less learnable. So the term “steep learning curve” is actually a misnomer — in reality, steep learning curves are good. They mean that the improvement is substantial and that it happens fast.

Memory and the Power Law of Learning

Romans used to say that “repetition is the mother of learning” — the more we rehearse a piece of information, the more likely we’ll be to remember it. Not only that, but we’ll also be faster at retrieving it from memory. When applied to human memory, the power law of learning says that the time it takes to retrieve a piece of information from memory depends on how much we’ve used that information in the past, and this dependence follows a power law. Thus, we are fluent with concepts and patterns that we use every day in our work, yet high-school math (such as the definition of logarithms) or other facts that we don’t often encounter are hard to remember, because we haven’t used them often enough.

In an experiment reported by Peter Pirolli and John R. Anderson (1985), the time it took participants to recognize facts that they had studied decreased with the number of days they had practiced those facts. The curve follows a power law and reaches a saturation level approximately around day 12.

Items that are practiced a lot acquire我们的记忆中的高激活和are retrieved faster. Whenever we’re trying to solve a problem or recall a piece of information, the first things that come to mind are those items in our memory that have a raised activation. Let’s say you want to navigate to the homepage of a site. You may have encountered multiple solutions to this problem in the past — for example, clicking the logo or clicking alink. All these solutions will compete in a “race” in your memory and you will select the one that gets to the finish line first. But, based on the power law of learning, the one that wins the race is the one that’s been practiced most often. Of course, if the first winning solution doesn’t work, people will try the next best. But they will also start feeling annoyed and perceive the problem as harder, and the site as less usable.

The key implications of this research are as follows:

  • 学习权力律是真实的:在很长一段时间内已被证明是无数的实验(19th, 20th, and 21st几个世纪)。这是人类大脑的作品的方式,而且没有一厢情愿的思维或新的小工具将改变这一点。设计为它。
  • 学习不是二分法(作为一个简单的模型可能已经假设了),你要么知道某事,要么你不知道。实践的事情越多,最近的练习越多,所知越好。
  • 只需向用户展示教程或帮助屏幕就不足以让他们了解一些东西。
  • Doing something often is the way to strong learning.

一致性=无聊,旧界面?

我们已经看到,每次重复都可以帮助用户练习概念或动作。因此,通过与其他网站一致,您将使用户再次重复一个高常见的UI元素,并且您还将在所有这些其他网站上获取实践的好处。记住Jakob的法律:您的用户将大部分时间花在其他网站上。

学习曲线对于两个不同的接口:当介绍新接口时,旧接口已经处于饱和级别(用于新接口的重复1对应于旧接口的重复5)。用户将需要更多的时间和良好的意志,使新的次优界面比继续使用旧的界面。

As shown in the graph above, when you are creating a new design pattern that goes against an old, familiar one (e.g., logo on theright页面而不是左侧,horizontal scrolling而不是桌面上的垂直滚动,汉堡菜单而不是导航栏在桌面上),新模式的学习曲线将在第一次重复的陡峭,高部分,而竞争旧替代方案的学习曲线将已经达到饱和。对于您的新设计来说,它将需要多少数重复,也可以达到饱和,也许比竞争对手更好。除非您的用户被俘虏,您可以强迫他们练习,除非他们会放弃并进入其他地方,而不是努力使用设计:users hate change

所以这意味着没有希望创新,对吧?我们注定要在右上角有搜索框,logo on the left和酒吧的导航?

任何类型的创新都会为用户和设计师提供费用。对于用户来说,因为它将是一个新的模式,他们必须学习,并将它们带到一个不介入的慢路上。对于设计师来说,因为他们必须提供额外的脚手架,例如上下文尖端progressive disclosure帮助用户浏览新路径。成本啊f implementing these tools can be significant. Think twice at what you are trying to achieve — is it worth departing from the well-beaten path? Does it make sense to innovate or will you be just as well served by a traditional design?

It also means that innovation is easier push when you have a captive audience or when the感知价值of a品牌比使用新设计模式的成本大得多。That’s why traditionally, big companies with a large user base (think Apple and iOS or Google and Android, to a lesser extent) can afford to innovate — because people who are already using these platforms will have to put up with the new interface (especially if the company is pushing updates aggressively, like Apple does with iOS, or the innovation happens in an enterprise, where users don’t have a choice to go back to an older version of the interface).

It’s also easier to innovate if your users will experience the new interface very often, perhaps several times a day. That means that people will get faster to the saturation part of the learning curve because they will have quite a few opportunities to practice. (Yet, if the saturation point is too far in the future, people may actually never get there.Windows 8.是一个实时证明:微软最终更改了设计,而不是等待用户到达饱和度高。)

Innovation can also happen if designers adopt it en masse and create a new standard. If all websites rebelled tomorrow and started placing the logo in the top right corner, then users would get the repetitions needed to reach saturation relatively fast, everywhere. Usually, this process takes time, but it did happen with design elements such as the swipe-to-delete gesture in iOS or the汉堡包菜单在移动

我们可以制作一个简单的决策树,以便在传统设计已经确定的情况下引入一个异常的用户界面:

  1. 新设计是否会表现很多一旦用户有“降临”的学习曲线,比旧的更好?如果没有,甚至没有尝试。
  2. Is it credible that users will be willing to try the new design again and again, until they have learned it well enough to realize those long-term benefits? If people are likely to give up (e.g., leave a website for a competing, familiar design), then don’t introduce the new design.
  3. Can you speed up learning, either by exposing users to the new design more often or by making it easier to learn? If yes, you will increase the proportion of users who will be willing to embrace the new design.

所以是的,一致性是设计创新的诅咒。如果您相信,一旦您的用户学习了界面,他们将节省时间Quo,那么它可能值得尝试。但请记住,创新的道路是迂回且昂贵的,如果您的用户没有许多学习机会,他们可能永远不会达到最佳性能平台,只有在学习发生后才访问。

了解有关人类记忆和研讨会的学习权力法的更多信息The Human Mind and Usability

参考

David Ahlstrom, Andy Cockburn, Carl Gutwin, Pourang Irani (2010). Why It’s Quick to Be Square: Modelling New and Existing Hierarchical Menu Designs. CHI 2010.

赫尔曼埃宾豪斯(1885年)。记忆:对实验性心理的贡献New York: Dover.

Allen Newell, Paul Rosenbloom (1980).Mechanisms of skill acquisition and the law of practiceTechnical Report. School of Computer Science, Carnegie Mellon University.

Peter Pirolli, John R. Anderson, J. R. (1985).实践中实践的作用检索实验心理学杂志:学习,记忆和认知,11,136-153。