Usa-blog-bility

Navigation and Its Role in the Information Architecture

Navigation and the Role it Plays in the Information Architecture

I thought it might be a good idea to review the role or function of the navigation menu as it relates to a Web site’s information architecture (IA). I have seen Web sites that don’t create a clear relationship between the navigation and the IA.

 

The navigation menu is a result of the Web site’s information architecture (IA). The IA represents how you categorized your site’s content. An analogy that may make sense to all of us is the structure of a house.

 

The rooms in your house are (hopefully) structured some way that makes sense.  Except for closets or master baths, most rooms are accessible from some main hallway. You don’t have to find the room that has the only entrance to another room in the house. Or you don’t find a room because you happened on a hidden doorway in another room to this room you had not idea even existed.

What the Navigation Is

The navigation menu is the map of your content. The navigation menu informs the user of what types of content your site has to offer. It helps them to find the content they want to find. Well designed and labeled navigation menus must represent your user’s mental model. So clearly navigation/category labels are important.  

What the Navigation Menu Isn’t

The major issues I see on sites are the following:

· Navigation menus that link to other web sites. The navigation menu should never lead to another Web site. Again, it is the content of your web site and not someone else’s site.

· Floating pages that are not anchored to a navigation category. There should never be any floating pages and all pages must be included in a content category.

· Site designs that drive the navigation menu. The navigation menu schema should drive the site design. Just because you want to use horizontal tabs doesn’t mean you should when another schema works better.

· Segmented web sites. Site segmentation by user type makes it difficult for the user to identify what content is in that category. Labeling a category for Medical Professionals doesn’t inform the user about the content. It is often much better to segment your category page than your navigation.

 

Creating an IA and resulting navigation menu is the most important piece of your Web site. Certainly I am not negating the importance of an aesthetic design. However, having a visually appealing site that has poor navigation, results in poor usability and frustrated users.  

Form Field Alignment

Lately I have noticed a change in the alignment of labels to their corresponding field on a form. In the past, fields typically were left aligned to the form edge. This was considered a standard or minimally good practice. The latest trend seems to be the right alignment of the label to the field resulting in a jagged edge.

Right aligning the label certainly makes it easier to keep it adjacent to the input field. This typically occurs when the labels are long. You don’t have to be concerned about the length of the labels for any of the fields. As a result your labels are easier to tightly align to the fields. There won’t be a large gap between the two as compared to left alignment.

But that isn’t an excuse to have long field labels. Clear, concise field labels are important especially for user comprehension. And right aligning the field label to the input field makes it difficult for users to quickly scan the fields and find the required fields. This alignment design also causes visual fatigue.

Left aligning fields to the form requires you to create short, concise labels. But it also makes it easier for users to quickly scan the form and identify the required fields. It may require a bigger effort to create short field labels that make sense to the user but in the end it is the best choice.

I thought it would be nice to see an example of both left and right aligned labels on the same form and the visual path that each requires. The following image shows the field labels right aligned and adjacent to the input fields. Notice the ragged edge of the user’s visual flow pattern. Another issue is the required field mark is also right aligned to the fields and it is difficult to scan the form and decide which fields are required. The Submit button is also left aligned to the bottom of the form and that isn’t the last place the user looks.

 

Above is the same form with the fields left aligned to the form edge rather than right aligned to the field. Note the straight visual track and how much easier it is to quickly scan the form for the required fields. The Submit button is aligned to the right at the bottom because that is the last place the user will look and right under the last field they need to complete.

 If you are considering right aligning the fields because of long field labels, then you should reconsider. Try renaming your field labels if it is driving the gap between the label and the corresponding field. Long field labels can overwhelm the user. If it appears that the label is confusing, add an example (as shown above) to the end of the field.  Following this standard will improve the usability of your forms. 

Author's Bio

Susanne Furman has been a usability engineer for almost a decade. She earned a Ph.D. in Applied Experimental Psychology, Human Factors from George Mason University. She manages the usability program and Usability.gov at the U.S. Department of Health & Human Services.

The Picture Says It All or Does It Say Too Much

I am sure that we have all seen Web sites that use images to benefit the user experience and of course a lot that don’t. Graphics often help when they improve the user’s experience but sometimes images just take up space and lead to more confusion.

 

Jared Spool and associates have classified graphics into three types: navigation graphics, content graphics, and ornamental graphics. He believes that well-done navigation and content graphics contribute to the user’s experience; unfortunately it is a lot harder to see the value of ornamental graphics. 

 

Navigation Graphics

So what is a Navigation graphic? Well it is probably easier to give a visual example then try to explain what it is. Let’s say you go to one of the online electronics retailers and you want to purchase a digital camera. If the link was labeled sleek and slim or Digital SLR, you may be confused about what camera you are interested in viewing. But if I were to add an image example of what a sleek and slim or Digital SLR with the link, it would help you to narrow down the camera selection.

 

Content Graphics

Content graphics provide additional information to a written description. So let’s say I want to sell my 100 year old Victorian row home located on Capitol Hill. I could write a nice description of the house and describe the large kitchen and the landscaped back yard. But how appealing is that? You would get a better idea of both the descriptions if I included nice pictures of the house and yard.

You may have some specific questions about the location of my house. You may want to know where exactly is it located on Capitol Hill and what is it close to. Can you walk to the metro? So if I were to provide a nice map showing the location of my house and the closest metros, stores, or points of interest, it would provide a better idea of the location then just the address.

 

Ornamental Graphics

Ornamental graphics are images on a page that help to break up the density of the text. These graphics are often used to make the site appear more professional, friendly, or fun. You are probably familiar with these graphics. Let’s say you go to a Web site that offers loans and you see a group of people smiling and shaking hands. It really doesn’t tell you anything about the interest rate or help you calculate your monthly payments. Does it make you feel better about dealing with this company because of the happy people?

 

Picture This – Other Options

Graphics are not inexpensive so they come at a cost whether you buy them from an image company or have a graphic artist create them for your site. Navigation graphics often help users find the items they are looking for better than the just the text alone. Content graphics supplement item descriptions. If you are using ornamental graphics to break up text density, there are cheaper ways to accomplish that task and help the user find the information they are looking for (e.g., headers and sub-headers to help chunk the information into meaningful pieces). Don’t overburden you pages with images that don’t provide any additional information or purpose.

 

Author's Bio

Susanne Furman has been a usability engineer for almost a decade. She earned a Ph.D. in Applied Experimental Psychology, Human Factors from George Mason University. She manages the usability program and Usability.gov at the U.S. Department of Health & Human Services.

Ratings vs Rankings

Up until now, my posts have focused on ratings and rating scales.  I’d like to switch gears for a moment and talk about rankings.  I occassionally hear people referring to ratings as rankings and vice versa, so I’d like to try to clear up some of the confusion.

With ratings, participants in a usability evaluation compare products to a set scale.  The scale generally has labels which anchor the response options along some continuum.  With rankings, respondents compare products to each other based on a specific construct. 

Ratings and rankings both have their strengths and weaknesses.  Ratings are generally easier for participants.  You can even estimate ranks by looking at mean or median ratings.  However, ratings may be susceptible to “nondifferentiation,” where participants tend to give similar ratings on multiple items.

With ranks, you do learn how participants would order the items because they must make a direct comparison between them.  This effort can minimize the risk of nondifferentiation bias.  However, ranks can be difficult for participants, especially if there are many items to consider or multiple items that are roughly equivalent.

There are ways to take advantage of the strengths of each method.  For example, you can ask participants to rank the items first, then rate each item individually.  Another option that may be helpful with a large group of items is to simplify the ranking task by having participants rank their top and/or bottom few choices, then ask for ratings for all items.

 

The views expressed here are those of the author and do not necessarily represent the policies of the U.S. Bureau of Labor Statistics.

Bio:  Jean Fox has been at the Bureau of Labor Statistics since 1998.  She conducts usability design and evaluation work on tools for data collection and dissemination and promotes usability whenever she can.  She wishes to thank Scott Fricker of BLS, who co-authored a paper from which many of these ideas were drawn.

 

Tis the Season

It seems that the new model for a lot of government Web sites features an animated billboard. We should always keep in mind most users are coming to government Web sites to find information and find it quickly. They are not coming to be entertained or see the latest photo or video of some administrator.

Last week I watched some users interacting with a Web site that uses this billboard model where the images rotated approximately every four seconds. An eye tracking device was used to see what the users were looking at. Every time the image changed, no matter what the user was trying to accomplish, you could see them quickly look at the rotating image and then go back to where they were looking.

The usability field has known for a long time that users cannot ignore peripheral movement. You may ask why that happens. Quite simply, there are rods and cones in our eyes. Cones are used for color detection and are tightly packed in the middle of the eye. The rods are used to detect motion and are located on the outside of the eye. No doubt this assisted our ancestors in escaping predators and staying alive.

The other interesting thing I observed is that the use of 1 2 3 4 as identifiers for the slide show really means nothing to users. Did you know that 4 really represented some dinner? Well neither did the users and they didn’t even look there to complete one of the tasks. And the poor Spanish users had a difficult time trying to find where the Spanish translated site link might be located. Of course finding the search box wasn’t much easier for some of the users.

So what does this tell us? Well Web site models that work well in some domains just don’t work well in others. And maybe they didn’t work well in that domain either, maybe someone only thought they did. When designing a Web site, no matter who we are, we need to follow good design principles. Just because we think as designers the Web site looks appealing and cool, doesn’t mean that they are usable.

 

Author's Bio

Susanne Furman has been a usability engineer for almost a decade. She earned a Ph.D. in Applied Experimental Psychology, Human Factors from George Mason University. She manages the usability program and Usability.gov at the U.S. Department of Health & Human Services.

Agree & Disagree Rating Scales

One type of scale that I’ve often used is the “Agree/Disagree” scale.  Basically, you present participants with a series of statements, then have them rate how much they agree or disagree with each statement.  It’s relatively easy to put together a series of items, all using this same scale, to rate the usability of a product.

However, it turns out that these types of questions may not accurately reflect the participants’ opinions.  People like to be agreeable, so they’re more likely to agree with a statement, regardless of what the statement is (“acquiescence” bias).  In addition, presenting a series of agree/disagree statements in a grid may encourage participants to give similar ratings for each statement (“nondifferentiation” bias). 

Further, these items may be difficult to answer.  To rate each statement, participants must first determine how they feel about the topic, then transform their opinion onto the agree-disagree scale.  This extra mental step may not be logical or have a direct one-to-one correspondence (e.g., strongly agreeing with the statement “it was easy to use” is not exactly the same as saying it was very easy to use).  In addition, this step will take extra time to complete (Visser, Krosnick, and Lavrakas, 2000).

One solution is to use response options that directly reflect the construct you want to measure.  For example, to evaluate “ease of use,” you might ask participants to rate their agreement with the statement “The application was easy to use.”  However, to more directly assess ease of use, you could ask:

Was the application easy or difficult to use?

    Very easy

    Easy

    Neither easy nor difficult

    Difficult

    Very difficult

With the direct scale, respondents do not need to take the extra step of mapping their opinions to their level of agreement with the statement.  There is still the risk of obtaining responses biased toward the positive end, but this scale minimizes the potential for acquiescence.  Further, because each scale is different, users may be less likely to provide the same response to each item. 

There has not been much research comparing the agree-disagree scale to these direct scales, particularly in the usability field.  However, this is something we should explore more in the future.

References

Visser, P.S., Krosnick, J.A., and Lavrakas, P.  (2000).  Survey research.  In H.T. Reis and C.M. Judd (Eds.)  Handbook of Research Methods in Social Psychology.  New York:  Cambridge University Press.

The views expressed here are those of the author and do not necessarily represent the policies of the U.S. Bureau of Labor Statistics.

 

Bio:  Jean Fox has been at the Bureau of Labor Statistics since 1998.  She conducts usability design and evaluation work on tools for data collection and dissemination and promotes usability whenever she can.  She wishes to thank Scott Fricker of BLS, who co-authored a paper from which many of these ideas were drawn.

Plain Language and Usability

Plain Language Defined

Plain language is a sister discipline to usability. The most widely accepted definition of plain language goes like this --

Writing in plain language means writing so your intended audience can

find:

• what they need
• understand it, and
• use it to fulfill their needs.

Sound familiar? I’ve had usability folks tell me that’s exactly how they define usability.

So it puzzles me when I see a site that someone has obviously designed with care that’s full of confusing content, obscure headings and links, and content that doesn’t address the reader’s concerns. Why go to the trouble to develop an effective design if you fill it with junk?

 

Let’s look at an example.

I looked at a page about the US Census, entitled “Questions You May Have.” The title is followed by a list of questions with answers. Here’s one pair:

Is there another way to get the form other than the mail?

Be Counted forms are census forms that are available at various community locations for use by people who either did not receive a form in the mail or whose information was not collected on any other form. Be Counted forms are available in English, Spanish, Chinese, Korean, Vietnamese and Russian. These forms can be picked up in various community locations and mailed back in the attached postage-paid envelope.

 

So what’s wrong with this?

First, it includes a lot of information the question (and supposedly the questioner) didn’t ask. And as is common in the government’s writing, it fails to address the reader. Most importantly, it doesn’t answer the question, except to give the vague instruction to pick up a form at a “community location.” What’s a community location?

 

Why didn’t they just say:

Yes. You can pick up a form and a postage-paid return envelope at your local post office. That would have saved a lot of space, and served the reader’s needs more effectively.

Examples like this are multiplied hundreds of thousands of times on federal web sites. If you care enough to test the usability of your design, why not test the readability and usefulness of your content as well?

 

Annetta Cheek Bio:

Dr. Cheek received a PhD in Anthropology from the University of Arizona. She spent most of her Federal Career writing and implementing regulations until the early 90s when she became interested in the Plain Language movement. She spent four years as the chief plain language expert on Vice President Gore’s National Partnership for Reinventing Government. She is currently the Chair of the board of the private sector Center for Plain Language. She works as a consultant providing plain language training and writing. She was instrumental in getting the U.S. Congress to introduce legislation mandating plain language in federal documents.

How Many is Enough

How many options should I include in a rating scale?

There’s been a longstanding debate over how many response options rating scales should have. Some people swear by 5, others prefer 3 or 10. The debate will rage on, but here are some thoughts on determining the best size for your rating scales.

Fewer vs More Options

With more options,

• The scale conveys more information.

• Respondents can better communicate the exact direction and intensity of their attitude.

• The scale is more likely to include an option that corresponds exactly to the respondent’s attitude.

However, with too many options,

• Respondents may have a difficult time differentiating the meaning of each individual option. As a result, they may pick randomly from several reasonable responses, rather than carefully considering the best response.

• Respondents with less familiarity with the topic may struggle with their response.

Therefore, the best guideline is to include as many options as respondents can distinguish. In general, that optimal number seems to be between 5 and 9 points, depending on the topic covered and the experience of the respondents (e.g., Alwin, 1997; Krosnick and Fabrigar, 1997).

Odd vs Even Scales

Odd-numbered scales have a neutral middle, while even-numbered scales do not. Without a neutral option, respondents will have to decide which way they lean. Usability-related scales typically offer a neutral midpoint, but some practitioners prefer no midpoint.

This is particularly relevant to bipolar scales (with opposite anchors on either end), where there is likely to be a natural neutral option in the middle. For unipolar scales (ranging from a zero value to some positive value, such as "never, rarely, occasionally, frequently, always"), there might not be a natural middle. However, respondents consider the position of the response options in determining their response, and may consider a middle option to be "about average."

The research has mixed results about the impact of having a midpoint. All agree that if there is a midpoint, people will choose it. However, the experts don’t agree on whether having a midpoint affects the results (see, for example, Krosnick, 1991 and Schuman and Presser, 1981).

To help decide, it’s important to consider the context of the survey. In some cases, it’s important to know whether respondents lean one way or another (like political polls). However, for usability questionnaires, it is usually reasonable for respondents to have a neutral opinion; if so, it is appropriate to offer it as an option.

References

Alwin, D.F. (1997). Feeling Thermometers Versus 7-Point Scales: Which Are Better? Sociological Methods and Research, 25(3), pp 318 – 340.

Fox, J.E. and Fricker, S.S. (2009). Designing ratings scales for questionnaires. Presented at the Usability Professionals’ Association Annual Meeting, Portland, OR, June 11, 2009.

Krosnick, J.A. (1991). Response strategies for coping with the cognitive demands of attitude strength in surveys. In J.M. Tanur (ed.) Questions About Questions: Inquiries into the Cognitive Bases of Surveys. New York: Russell Sage Foundation, pp. 177 – 203.

Krosnick, J.A. and Fabrigar, L.R. (1997). Designing rating scales for effective measurement in surveys. In Lyberg, Biemer, Collins, de Leeuw, Dippo, Schwarz, Trewin (Eds.) Survey Measurement and Process Quality. John Wiley and Sons, Inc.

Schuman, H. and Presser, S. (1981). Questions and Answers in Attitude Surveys. New York, NY: Academic Press.

The views expressed here are those of the author and do not necessarily represent the policies of the U.S. Bureau of Labor Statistics.

Bio: Jean Fox has been at the Bureau of Labor Statistics since 1998. She conducts usability design and evaluation work on tools for data collection and dissemination and promotes usability whenever she can. She wishes to thank Scott Fricker of BLS, who co-authored a paper from which many of these ideas were drawn.

Why Would They Do That

But Why Did They Do It That Way?

So far we have had an assortment of blogs including Section 508, diverse routes in becoming a usability specialist, outsourcing usability testing, and Likert Scales. Quite an assortment of topics and I hope they have been minimally entertaining if not educational. In an effort not to become our own statistic, I am going to fill in as a guest blogger.

With so many topics that I could possibly choose from, because we all have interacted with poor designs, I find myself at a loss for words. Believe me, I am always saying to myself - 'why would someone do that, what in the world were they thinking?' And I don't just save that question for Web sites but lots of things I interact with.

Pretty soon I will have to change my car clock because the time will change. Of course, every 6 months I make a mental note of how I changed the clock time but it never seems to stay noted. As a result, I am always looking in the owner's manual and thinking to myself why can't there be a button with a clock on it?

Of course being a Federal employee, I have to interact with various Web applications to enter my leave time, check my pay and do all those things we do to keep informed about our professional careers. Of course, they all require different passwords and of course I forget. And of course there isn't a way to let them know I forgot the password like Yahoo, Google, etc do. Then I think, who am I supposed to contact to get a new one because there isn't a link or even a name or number to call. And of course I say to myself - who did this?

I watched a usability test last week and the site had an optional secure sign-in. So I was watching and listening to the participant's comments. Now he knew that he had never signed up for this optional service. But he made comments about what ID/password combination he might enter that he already has used other places. The most important comment he said was "Why should I do this?"  Of course it was not clear why anyone would use this feature. And even though the help and registration information are below the secure sign-in, they don't look like they are part of the process. And of course I asked myself "why don't they tell the users what they receive if they sign up for this optional service?" and I wonder how many registered users they have.

So with this said, I realize every day how resilient humans are when they interact with various systems. But I also wonder why we don't make it easier for them!

Author's Bio

Susanne Furman has been a usability engineer for almost a decade. She earned a Ph.D. in Applied Experimental Psychology, Human Factors from George Mason University. She manages the usability program and Usability.gov at the U.S. Department of Health & Human Services.

 

Scales & Data Analyses

Rating Scales

Usability professionals often rely on questionnaires to collect data as part of a usability evaluation.  We often use ratings scales as a way to facilitate data collection and analysis.  Rating scales provide response options along some continuum.  Rating scale responses are designed to have little overlap between neighboring options.¹ The resulting data are generally ordinal in nature, which means that there is a logical order to the answers but the distance between values is not known.

There are many types of rating scales, and I’ll spend a little time on the most popular in the usability field, starting with Likert scales today.

 

Likert Scales

Probably the most well-known rating scales are "Likert Scales." Rensis Likert (actually pronounced lick-urt) originally developed his scale as a way to combine items to compute an overall score related to attitudes. His scales have three characteristics²:

(1) The response options are evenly-spaced,

(2) The options are "bipolar" and symmetric (meaning that the endpoints of the scale are opposite, as with a typical "agree" to "disagree" scale), and

(3) Multiple items (questions) are combined to get an overall scale "score."

Usability questionnaires are more likely to use what are technically called Likert items, with characteristics (1) and (2) above, but not (3). For example, the questionnaire might include just one question on satisfaction and one other on ease of use.

Further, some usability questionnaires use Likert-type items, which do not meet any of the three characteristics listed above, but still present responses along some continuum, such as a scale of "helpfulness," including "Extremely helpful," "Helpful," and "Not at all helpful."

Analyzing Likert Scale Data

I’ll provide more information about analyzing rating data in the future, but for now, it’s important to remember that Likert items and Likert-type items generate ordinal data, but not the interval data required to use parametric statistics. Therefore, it is better to rely on non-parametric statistics to analyze these data.

True Likert scales combine ratings from multiple items, which can have some advantages. Combined ratings have better reliability and validity than single items³. In addition, you can use parametric statistics on these combined scores, making analysis a little bit easier (Carifio and Perla, 2007)⁴.

The views expressed here are those of the author and do not necessarily represent the policies of the U.S. Bureau of Labor Statistics.

_____________________________________________________________________________

¹ Krosnick, J. A., Judd, C. M., and Wittenbrink, B. (2005). The measurement of attitudes. In D. Abarracin, B. Johnson, and M. Zanna (Eds.) The Handbook of Attitudes and Attitude Change: Basic Principles. Hillsdale, NH: Lawrence Erlbaum.

² Likert, R. (1932). A technique for the measurement of attitudes. Archives of Psychology, 140, 1-55.

³ Trochim, W. (2000). The Research Methods Knowledge Base, 2nd Edition. Atomic Dog Publishing, Cincinnati, OH.

⁴ Carifio, J. and Perla, R. (2007). Ten common misunderstandings, misconceptions, persistent myths and urban legends about Likert scale and Likert Response Formats and their Antidotes. Journal of Social Sciences, 3(3), pp. 106 – 116.

__________________________________________________________________

Author's Bio

Jean Fox has been a Research Psychologist at the Bureau of Labor Statistics since 1998.  At BLS, she conducts a variety of usability activities including usability testing, expert reviews, and focus groups.  She is also very involved in the User Experience Task Force of the US Federal Government's Web Managers Council.  Previously, she worked as a Usability Consultant for American Institutes for Research.  She earned her Ph.D. in applied experimental psychology from George Mason University. She earned her Master's degree in Human Factors Engineering from Virginia Tech, and a Bachelor's degree, also in Human Factors Engineering, from Tufts University.