Skip all navigation and go to page content

NEO Shop Talk

The blog of the National Network of Libraries of Medicine Evaluation Office

Mean, Median or Mode–What’s the Difference?

Five stars ratings with shadow on white

Last week I taught the class Finding Information in Numbers and Words: Data Analysis for Program Evaluation at the SCC/MLA Annual Meeting in Galveston, TX.  There is a section of the class where we review some math concepts that are frequently used in evaluation, and the discussion of mean, median and mode was more interesting than I expected.  Mean, median and mode are measures of “central tendency,” which is the most representative score in a distribution of scores.  Central tendency is a descriptive statistic, because it is one way to describe a distribution of scores. Since everyone there had run across those concepts before, I asked the group if anyone knew of any clever mnemonics for remembering the difference.  Several people responded, both in class and afterwards (thanks Margaret Vugrin, Julia Crawford and Michelle Malizia!) Here are a couple of memory tools for you:

Mean: Turns out those mean kids are just average (mean = average)

Median: Just like the median in the road: if you line the values up in order, the median is the number “in the middle”

Mode: Mode has an “O” sound, and O is the first letter in “Often.”  It is the value on the list which occurs most often.


Hey diddle diddle, the Median’s the middle; you add and divide for the Mean. The Mode is the one that appears the most, and the Range is the difference between.

Now that you can remember them, which one should you use?  I think a good way to think about it relates to the ratings you see when you’re trying to pick a hotel or restaurant.  I don’t know about you, but when I’m looking at a restaurant score, when I see 4 stars (out of 5), I’m not comfortable with that number without seeing the breakdown.  There’s a big difference between a 4 where the scores are spread out and one where the scores are heavily skewed towards 5.

Let’s say I’m looking at reviews for hotdog restaurants that have around 4.0 stars.  The first one I look at has a distribution of scores spread out relatively evenly from 5 stars (the most number of ratings) to 1 star (the fewest).  A mean (mean kids are just average) works well here.  To do this you add all the scores up and divide the sum by the number of responses to reach a mean of 3.9.


Here is the chart for a similar hotdog restaurant with a slightly higher 4.2 star rating:


While the mean adds up to 4.2,  you can see in the chart that the scores are heavily skewed towards 4 and 5 stars, and it seems that 4.2 does not accurately describe the ratings of that restaurant (remember that central tendency is a descriptive statistic). However, if you use a (middle of the road) median with this kind of distribution, the result is a whopping 5.  To determine the median, you line all the results up in order and select the one in the very middle.  You might want to use a median when the distribution looks skewed.

This particular restaurant analogy doesn’t really work with mode. Mode is used for categories, which really cannot be averaged mathematically.  For example, if you want to know what is the most representative type of restaurant in your city, you might find out that your city has more hotdog restaurants than any other kind of restaurant (that would be awesome, right?).

I hope this helps. If you know any other mnemonics for mean, median, or mode, please send me an email at and I will add them to the bottom of this post.


How I Learned to Stop Worrying and Love Logic Models (The Chili Lesson)


By Michelle Malizia, Director of Library Services for the Health Sciences, University of Houston

I’ll start with a full disclosure: I am a late convert to logic models. Many years ago, I worked in a department that, for a period of time, became governed by logic models. This experience made me fear… no, hate… logic models.  Several years later, through external workshops and assistance from the NN/LM Evaluation Office, I was introduced to the tremendous value of logic models.

My closest personal analogy relates to my feelings about chili. I grew up eating my mother’s chili, which basically consisted of cans of many different types of beans floating in a type of broth. I hated it. When I was 22 years old, I had no choice but to eat someone else’s chili. This chili had lots of ground beef and spices. It was delicious. Then it occurred to me, my mother’s chili was my only frame of reference for chili. I didn’t dislike chili – I disliked my mother’s chili.

And so it goes with logic models. Once I learned a different way to make and apply them, I became a dedicated user. I now design logic models whenever I plan a new service, activity or initiative.

In 2014, I was hired as the Director of Library Services for the Health Sciences at the University of Houston (UH). In 2017, UH will have its first medical library and my task is to plan the services for the new facility. Of course, I turned to logic models because they provided the framework of not only what I am planning but why I am planning each service and ultimately, how I will evaluate if I achieved the goal.

When I started with my logic models, I was tempted to begin with the activity. I had to remind myself that it is more important to document what I hope to accomplish by that activity (i.e. outcome).  Think about it: Why do librarians teach PubMed classes? Why do librarians want to be embedded in a nursing class? Why do so many libraries provide liaison services? Many of you are probably thinking: “That’s easy, Michelle. We do those things to better serve our customers.” My response is:  How do you know those activities better serve your customers?  How can you prove it to your stakeholders?  That is the reason you should start with your outcomes rather than activities.

For example, my new library will be providing assistance with NIH Public Access Policy compliance. When I developed my logic model, I called upon my inner 3-year old to ask the question best asked by toddlers:  Why.  Because I have a creative side, I use the software Visio (a Microsoft Office software) to create my logic models. It allows me to visually see connections between activities. The chart below shows  a portion of my logic model.

NIH Public Access Policy Assistance Services logic Model One. Activities are conduct workshops and assistance with compliance. Outputs are number of workshops and number of consultations. Short-term outcomes are increase awareness of NIH Public Access Policy and increased knowledge of compliance specifics. Intermediate outcome is compliance and long-term outcome is UH retains and receives NIH grants.


As you can see, my long-term outcome for this activity was to ensure that UH retains and receives NIH Grants. If UH researchers don’t comply with the NIH Public Access Policy mandate, their current and future funding is in jeopardy. The intermediate outcome leading to the long-term outcome is increased compliance with the policy. In order to increase compliance, I need to make researchers aware of the policy and how to comply.  That’s when I was able to determine the best methods for me to accomplish those outcomes. For my university environment, the best way to achieve these outcomes is through workshops and consultations.

Now that I knew the “what and the why,” I needed to determine the how.  How would I know if I accomplished my goals? Again, I turned to Visio to visualize how I could assess if I achieved my outcomes.

My final step was to determine my measurable indicators. For example, in the case of workshops, my indicator was “% of workshop attendees who reported being more knowledgeable about how to comply with the policy.”  My target was 85% of attendees. To evaluate this outcome, I would use a pre- and post-test.


Evaluation plan for NIH Public Access Policy Assistance Services. Outputs are workshops and consultations, leading to short-term outcomes of increased awareness of NIH Public Access policy and increased knowledge of compliance specifics. The outcomes will be assessed with a pre-post test and follow-up questionnaire. The intermediate outcome is increased compliance, which will be assessed with a survey and/or other follow-up

My overall work with logic models led to a pleasant surprise. Mid-way through my process, UH Libraries adopted a new strategic plan. Strategic plans are usually written in terms of goals. Some of my colleagues tried to feverishly determine where their activities fit into the library’s overall goals. Because I had already determined my outcomes, it was easy to slot my activities into the library’s overall plan.

If you have had a previous bad experience creating logic models, try it again. Ask the NEO for assistance and look at their extremely helpful guides. Like me, you may finally realize that logic models are worth the time and energy. Remember, there are many different types of chili.  Find the one that you like best.

NEO note: The evaluation field has come a long way in discovering new, less painful approaches to creating and using logic models.  If, like Michelle, you had bad experiences years ago with logic models, you might want to give them another chance.  You can learn one approach through the NEO booklet Michelle mentioned, which is  Planning Outcomes-Based Projects (Booklet 3 in our Planning and Evaluating Health Information Outreach Programs series). For alternative approaches, check out our NEO Shop Talk blog entries  Logic Model for a Birthday Party and An Easier Way to Plan: Tearless Logic Models.


How Many Interviews Does It Take to Assess A Project?

A green piggy bank standing with a group of pink piggy banks to represent the cost effectiveness of individual interviews

FAQ from NEO users: How many interviews or focus groups do we need for our qualitative assessment project?

Our typical response: Um, how much money and time do you have?

At which point, our users probably want to throw a stapler at us. (Karen and I work remotely out of an abundance of caution.)

Although all NEO users are, in fact, quite well-mannered, I was happy to discover a study that allows us to provide a better response to that question.  A study published by Namey et al, which appears in the American Journal of Evaluation’s September issue, provides empirically based estimates of the number of one-to-one or group interviews needed for qualitative interviewing projects.  More specifically, their study compared the cost effectiveness of individual and focus group interviews. The researchers conducted an impressive 40 interviews and 40 focus groups (with an average of eight people per group). They then used a boot-strap sampling methodology, which essentially allowed them to do 10,000 mini-studies on their research questions.

They first looked at how many individual and focus group interviews it took to reach what qualitative researchers call thematic saturation. In lay terms, saturation means “Not really hearing anything new here.”  Operationally, it occurs when 80-90% of the information provided in an interview has already been covered in the previous interviews.

The researchers found that 80% saturation occurred after 2-4 focus groups or eight individual interviews. It took 16 interviews and five focus groups to reach 90% saturation. Note their estimates apply to studies that focus on one specific population.  If you want to explore the experiences of two groups, such as doctors and nurse practitioners, you would hold eight interviews per group to reach 80% thematic saturation.

For comparative cost assessment, the researchers used a formula that combined hourly rate of the data collector’s time, incentive costs per participant, and cost of transcription for recordings. They chose not to include cost for factors that vary widely, such as space rental or refreshments. Using more predictable costs made for cleaner and more generalizable comparisons.

Bottom line, they found individual interview methods cost 12-20% less than focus group methods.

Of course, many of us operate on shoestring budgets, so we are our own moderators and transcribers.  Even though most of us DIYers collect hourly wages, the cost for outsourcing these tasks is probably higher than for conducting them internally. Knowing this, the researchers looked at variations on moderator, transcriptionist, and incentive costs.  They also compared cost effectiveness of the two methods when lowering the standards for thematic saturation (i.e., aiming for 70% saturation instead of 80%). Across the board, individual interviews were more cost-effective than focus groups.

Cost is not always the only consideration when choosing between focus groups and individual interviews. Some assessment questions beg for group brainstorming, while others demand the privacy of one-to-one discussions. However, for many assessment studies, either method is equally viable.  In that case, cost and convenience will drive your decision. Personally, I often find individual interviews to be more convenient than focus groups, both for the participants and for me. It’s nice to know that the cost justifies using the more convenient approach.

The full article provides details on their methods, so it is a nice primer on qualitative analysis of interview transcripts. Here’s the full citation:

Namey E, Guest G, McKenna K, Chen M. Evaluating bang for the buck: a cost effectiveness comparison between individual interviews and focus groups based on thematic saturation levels. 2016 September; 37(3): 425-440.


Meet the NEO’s New Program Assistant Kalyna Durbak

Kalyna Durbak

I am pleased to introduce the NEO’s new program assistant, Kalyna Durbak, MLIS, who joined our staff on October 3.  Kalyna will be our go-to person for managing the NEO website, providing technical support with webinars, and helping with the “roll-up-your-sleeves” work involved in carrying out evaluation projects.

Kalyna began working for the UW Health Sciences Library in May 2016. Prior to joining the NEO, Kalyna was the Web Content Assistant on the team that created and promotes the Response & Recovery App in Washington (RRAIN), designed to provide emergency responders with quick access to disaster-management resources. It also provides local information such as weather alerts and traffic reports. Kalyna also provided web content and social media assistance for the Health Evidence Resource for Washington State (HEALWA), a portal that provides affordable online access to clinical information and health education resources. The portal is available to health professionals who are licensed through 23 state organizations. A 2015 evaluation study conducted by HEALWA showed that many health professionals eligible to use the portal are not aware of it.  Kalyna helped promote HEALWA through social media and exhibits.

Kalyna earned her MLIS degree from University of Illinois at Urbana–Champaign and a BA in History from University of Illinois at Chicago. She was an intern at the Smithsonian Ralph Rinzler Folklife Archives and Collections and the Rochester Institute of Technology Archives. A Midwestern native, she recently moved to Seattle with her husband because they were attracted to Seattle’s mix of urban and outdoor opportunities. To introduce herself to our readers, Kalyna agreed to answer a few questions about herself.

What made you want to pursue an MLS?

I always considered myself a “jack of all trades.” At school I did not excel in one subject, but rather did fairly well in most areas of study. I also fell in love with researching, and doing “deep dives” into different subjects. I figured that with an MLS, I could end up working in vastly different environments, help others with research, and pursue my dream of being a lifelong learner.

What made you want to join the NN/LM Evaluation Office?

I recently realized that I needed to strengthen my evaluation skills. Whether I am working or volunteering, I am constantly trying to solve issues concerning outreach and training. For most of my career, I just created solutions without ever thinking “How can I measure my success in solving this issue?” and “Are these solutions working the way I intended?” These questions are key in determining whether the solution is actually solving any problems, or just wasting time and energy.

What experience have you had with evaluation?

My experiences with evaluation come from managing social media accounts. Once I realized I had a whole dashboard of statistics to my disposal, I used them to set optimization goals in terms of posting times and types of content that resonate with my audiences.

What evaluation skills do you particularly hope to develop?

I am very interested in developing my outcome assessment skills. I am usually the big idea person of a group, and enjoy setting lofty goals. In the past, I have measured the success of an initiative based on the number of tasks my group completed for the project. What I want to do going forward is measure success by the initiative’s impact on the intended audience and community.

What other interests do you have?

I am very active in a Ukrainian Scouting Organization called Plast. Through scouting I found my love for the outdoors, and made countless friends all over the United States and around the world. When I’m not working on scouting activities, I find myself crafting. My favorite crafts include quilling, card making, and traditional Ukrainian embroidery.

When I am crafting or commuting to work, I listen to various nerdy podcasts. Some of my favorites include 99% Invisible, LibUX, and Reply All

What is the bravest thing you’ve ever done?

I took a hike with my husband down the Grand Canyon. I didn’t make it that far, because I’m afraid of heights. There was one foot between my body and a drop into the canyon, and that was not where I wanted to be. I had to turn back partway down. My husband said, “I love you. Do you mind if I keep going?”  So I had to walk back up the trail alone. Looking back, I’m glad I went through it.  Once I climbed up, I felt so proud of myself.

What’s in a Name? Convey Your Chart’s Meaning with a Great Title

Some of you may be working on conference posters and paper presentations for Fall conferences.  And some of those will probably include charts to demonstrate data representing a lot of hard work on your part.  In most cases you have minutes to use that chart to get your audience to understand the data.

Stephanie Evergreen has great advice for displaying chart data.  She literally wrote the books on it: Presenting Data Effectively and Effective Data Visualization.  Her recent blog post is about one of the simplest and most powerful changes you can make to effectively present your chart data: “Strong Titles Are The Biggest Bang for Your Buck.

What many of us do is present the data with a generic title, like “Attendance rates.” Then the viewer has to spend time working through the data and you hope that they see what you wanted them to.  What Stephanie Evergreen proposes (backed by persuasive research) is to give your charts a clear title that explains what the data shows. Your poster or paper is almost certainly making a point.  Determine how your chart supports the point of your presentation and state that in the title.  Here are some reasons why:

  • It respects your viewers’ time
  • It forces you to be clear about the point you want your data to make
  • It makes the data more memorable

Stephanie Evergreen’s post has some great examples of how a good title can really improve the impact of the chart.  In addition, here is an example from the NEO webinar Make Your Point: Good Design for Data Visualization.

Looking at this original chart, you might notice that in each activity the follow-up showed an increase over the baseline.  If you, the viewer, didn’t have a lot of time, that might be all you noticed.

Chart with title: Comparison of emergency preparedness activities from baseline to follow-up

With a simple change of title , you can see that the author of this presentation is highlighting the increased number of continuity of services plans.  This is designed to enhance the point of the presentation, and not waste the viewers’ time. Also, note that the title is left justified instead of centered.  Because the title is a full sentence, a left-justified format is easier to read.

Chart with title: The biggest improvement in emergency preparedness from baseline to follow-up was the number of network member organizations reporting that they had or were working on a service continuity plan.

So, while Shakespeare might have been correct when he wrote “What’s in a name? that which we call a rose / By any other name would smell as sweet,” what if the presenter was trying to show the fortitude of Texas antique roses to survive in harsh weather conditions, and the viewer only noticed how sweet the rose smelled?  Maybe the heading “A Rose” sometimes isn’t enough information.








Six Degrees of Key Informants: Finding Interviewees for Community Assessment

Linking grid of the social networks of a young adults of various nationalitiesSix degrees of separation is a concept that describes social interconnectedness. The idea, popularized in a 1990s movie, is that each human on the planet could reach any other human, by way of friend-to-friend introduction, in six steps or less. For example, you are six (or fewer) people away from meeting your favorite celebrity you happen to adore. Research has explored this social interconnectedness and, while the actual maximum number of intermediaries may be in dispute, we do live in a small world.

If you are faced with doing a needs or community assessment, social interconnectedness should be comforting. Most good community assessments conducted for project planning involve interviewing key informants selected for their special knowledge about their communities. Key informant interviews can provide a quick source of detailed information important for planning. This tip sheet from USAID talks about the basics of conducting key informant interviews.

However, when Karen and I provide training on community assessment, we find our workshop participants feel daunted by the prospect of actually identifying key informants. So take heart, readers. Our social interconnectedness means good key informants are just a friend or two away.

The interview sampling approach I use is described in this article by Tiffany.  It is a participatory evaluation method that engages stakeholders in framing evaluation questions and recruiting interviewees. These stakeholders are your first key informants. You most likely will find them among members of the project team initiating the community assessment. It is likely that their interest in a community was sparked because someone on the team knew someone in the user community.

Those stakeholder key informants can, in turn, direct you to more key informants who can talk about their own and their peers’ needs, desires, opinions and lifestyles.  After each interview, you ask a key informant “Who else do you recommend that I interview?” and “What important information can that person add to my understanding about the community?” As a result, your key informants share in refining your community assessment inquiries. Their recommendations allow them input into the direction of your project plans. Because your key informants are likely to be opinion leaders in their communities, you can generate enthusiasm and possibly forge important partnerships, assuming they respond positively to your project.  Key informant interviews are your first step in building trust in the user community you’re assessing.

To find key informants  who can truly help you gather good project-planning information, be clear about the information you’re seeking.  That way, your stakeholders can refer you to the best  interviewees for your needs.  For guidance on the type of information you should gather in a community assessment, check out these NEO blog posts on Diffusion of Innovation Part One and Part Two.

Ideally key informants get something in return for participating in interviews. At the very least, key informants who are opinion leaders have valuable information about your project or organization to share with their peers. More significantly, your interviewees will assist you in bringing valuable services or resources to their communities.

I want to share two examples of projects where I used this approach to key  informant sampling.  A few years ago, I led a community assessment project for Cumberland County (North Carolina) Public Library and Information Center, which wanted to improve its service to the military community affiliated with Fort Bragg. (Public Libraries published an article about this project here.)Public librarian Jennifer Taft received funding from the State Library of North Carolina for this project and also participated in the community assessment process. We started by interviewing a cadre of her colleagues from the Fayetteville Community Blueprint Network, composed of representatives of local organizations that served military families. Jennifer and her colleagues had worked together to put on a community forum on post-traumatic stress. After each interview with her forum colleagues, I collected recommendations for other key informants. I did the same with my second wave of interviews. Our sample grew until we had a good sample of interviewees and focus group participants with experience-based perspectives on the military community. All worked in organizations that provided services to military families. Most also were members of military families (that is, service members, veterans or spouses).  The key informant interviews had an advantage beyond providing useful information. Relationships established in the interviewing phase provided the library with the contacts it needed to participate, for the first time, in on-post activities.

In a different project, I worked with the National Network of Libraries of Medicine South Central Regional Medical Library to explore how it could support public libraries in hurricane-prone counties. Our sampling process began with contacting librarians at the state libraries of Louisiana and Texas, both of which actively supported public libraries during Hurricanes Katrina and Ike. They introduced us to key informants from “further-in” libraries that valiantly helped waves of evacuees from communities that suffered direct hits.  Our contacts pointed us to libraries that were struck by the storms and restored services quickly in order to help their community members. After we completed our interviews, these librarians became valuable partners in helping us develop NN/LM resources. (You can read about the Gulf Coast library community assessment here in Public Libraries.

Still worried about locating good key informants?  I assure you, you can have faith in the interconnectedness phenomenon.  It has always worked for me, starting with my very first qualitative interviewing project. That project occurred in the 1970s when I was an undergraduate at Penn State. I was enrolled in an undergraduate class taught by an American folklorist.  Our extra credit assignment was to find three legends from our own families or social circles. In those pre-Internet days, most modern-day legends were ghost stories. I was immediately overwhelmed.  What friend of mine could possibly have a ghost story to share? Turns out, the first person I saw after class had a haunting tale. And so did the next person. Within days, I was 10 points closer to getting an A in the course. Everyone, it seemed, knew someone who had seen a ghost.

So remember, you are probably less than six people removed from a great key informant. Just get a handle on what you want to know in your community assessment, talk to anyone affiliated with the community, and you’re on your way.

And, if you know somebody who knows somebody who knows Kevin Bacon, kindly send their contact information to Cindy or Karen?


Reference: One of the most well-known studies of interconnectedness was

published by Travers, Jeffrey, and Milgram, “An Experimental Study of the Small World Problem”, Sociometry 32(4, Dec. 1969):425–443


Update Your Evaluation Toolbox: Two Great Conferences

It’s the fall, also known as the beginning of conference season. It’s a very exciting time if you like evaluation/assessment.  If you want to improve your evaluation skills, two great conferences are coming up, back to back.  Take a look at some of these highlights and pick one to go to!

Oct. 24-29, 2016 Evaluation 2016, Atlanta GA

Eavlaution 2016 October 24-29, Atlanta, GA

This is the annual conference of the American Evaluation Association, an international organization with over 7000 members, and interest groups that cover topics like Assessment in Higher Education; Collaborative, Participatory & Empowerment Evaluation; and Data Visualization and Reporting.  The theme of this year’s conference is Evaluation + Design.

The conference has 40 workshops and 850 sessions.  Here are some example programs:

  • From crap to oh snap: Using DIY templates to (easily) improve information design across an organization
  • Developing Evaluation Tools to Measure MOOC Learner Outcomes in Higher Education
  • Evaluation Design for Innovation/Pilot Projects

There’s still time for Early Bird Registration (ends October 3)!

Oct. 31-November 2, 2016 Library Assessment Conference, Arlington VA

Library Assessment Conference 2016

This conference only happens every other year and is co-sponsored by the Association of Research Libraries (ARL) and the University of Washington (UW) Libraries (disclosure – the NEO is part of the UW Libraries–something we’re quite proud of).   The theme for this conference is Building Effective, Sustainable, Practical Assessment.

This conference is bookended by workshops like Getting the Message Out: Creating a Multi-Directional Approach to Communicating Assessment and Learning Analytics, Academic Libraries, and Institutional Context: Getting Started, Gaining Traction, Going Forward.

Scholarly papers and posters with titles like “How Well Do We Collaborate? Using Social Network Analysis (SNA) to Evaluate Engagement in Assessment Program” and “Consulting Detectives: How One Library Deduced the Effectiveness of Its Consultation Area & Services” are organized around a variety of topics, such as Organizational Issues; Ithaka S+R; and Analytics/Value.


This is an exciting time to be in the assessment and evaluation business.  Take this amazing opportunity to go to one of these conferences.















A Diagram is Worth a Thousand Words: Visual Evaluation Plans

What would you rather look at?  Some paragraphs of text and bullet points that explain in a step-by-step fashion your process and outcomes evaluation plans, or a diagram of those plans?  For me the answer is easy: a diagram.  Diagrams have the advantage of being quickly understandable, interesting to look at, invite participation of the viewer, and possibly most important for me, they’re colorful.  A textual explanation can walk me through the same process, but I would play a much more passive role, and I might not understand the big picture without having, well, a big picture.


Obviously you would also need the text.  Somewhere you need to explain the details of what you’re going to do in your evaluation. But a diagram can make the plan immediately comprehensible, and the reader can then read the textual explanation while understanding the overall context.

Bethany Laursen, an evaluation consultant, posted some examples of what she calls visual evaluation plans in her blog, Laursen Evaluation and Design.  These are created by students in a class at the University of Wisconsin-Madison.  I like them because by looking at them I have a basic understanding of their projects and how they will be evaluated.

Her blog post presents visual evaluation plans as a way of getting non-evaluators to understand your evaluation plans.  But I think they can also be a way that people (whether evaluators or non-evaluators who find themselves writing evaluation plans) could begin to think about how to plan their evaluation strategy to fit their project.

Microsoft products like Word and Power Point have drawing tools that can work to make diagrams.  But I think best with pen and paper, so if I were designing an evaluation plan for my daughter’s birthday party (see February 4, 2016 post), I would do something like the drawing here. Then I could create a plan for evaluating each of the process evaluation questions (in blue arrows) and each of the outcome evaluation questions (in red arrows).

This video, Faster Program Evaluation Planning: a New Visual Approach, shows how you could use a product like DoView to create a snazzy looking evaluation plan that also can link images to the textual description of the evaluation, and even further, link to your actual evaluation.

That famous phrase in the title, “a picture is worth a thousand words,” works really well to show how you can use your diagram to communicate your evaluation plan to others.  But if you’re using a diagram to design your plan in the first place, the quote that might work better is Gloria Steinem’s: “Without leaps of imagination or dreaming, we lose the excitement of possibilities. Dreaming, after all is a form of planning.”

And as Winnie-the-Pooh says “Nobody can be uncheered with a balloon.”









My Report Writing Toolkit


I once heard evaluator extraordinaire Michael Patton say that an evaluator could staple an executive summary to a bunch of pages ripped from a phone book and no one would notice. Possibly our readers have developed a fear of drowning in numbers and technical information?

(For our younger blog followers, a phone book is that thick paperback that materializes on your doorstep about once a year and you trip over it a few times before throwing it in your recycling bin.)

Many of us are trying to write better reports, thanks to proactive efforts in our professional associations.  Many such organizations provide excellent training on report writing, often to sold-out audiences. The first step toward better reporting is better synthesis of our evaluation findings.  You yourself must understand your data well before you can effectively share findings with others.  However, there are many other design elements in a report that you can use to help your readers understand key points and retain important information. Nonverbal elements such as color, font choice, page layout, and graphic design, all contribute to effective evaluation reporting.

I have picked up a few tricks of the trade over the past few years.  So in today’s blog, I’m giving you my personal report-writing toolkit.

PowerPoint:  You may think of PowerPoint as a presentation tool, but I have discovered it is also a great tool for producing written reports.  Slide layouts provide flexibility in organizing graphics and text on a page.  The text boxes also force me to be succinct with written content. My favorite resource for PowerPoint reports is Nancy Duarte’s Slidedocs.  You can download free PowerPoint templates at her website, but truthfully, I seldom use them.  They never seem quite right for what I want to present and I don’t think all of them are accessible (508 compliant).  However, I use them to guide my design.  The templates provide examples of good layout and color palettes.  Also, Duarte’s templates exemplify effective practices for readability, such as ideal column width and line spacing.

“Presenting Data Effectively” by Stephanie Evergreen. I routinely consult this primer on presenting data when I write evaluation reports. Her book gets into the nitty-gritty of reporting evaluation results.  How do you choose font type? Where do you place data labels in a chart? How do you layout a page to incorporate text and charts. She leaves no stone unturned in this book.

Free photos: Photos have their place in both written reports and presentation slides, particularly when they serve as visual metaphors for key findings. Google’s advance search has a “usage rights” option that allows you to quickly find images online that are free to use or share. However, the quality of images from Google searches is variable.  I prefer to start with Pixabay, which provides consistently high quality pictures that are free to use.

Color Picker Tools:  Accent colors add visual interest to reports and direct readers’ attention to key findings. There are two color picker tools that I use routinely to find accent colors for my headers and graphics. PowerPoint now has an “eye dropper” feature that allows you to add custom colors that match images in your reports..  This is the fastest way to add a custom color to your theme palette.

However, when I have time to perfect my color choices, I rely on Adobe Color. You upload your image and Adobe Color shows you a palette of complementary colors to choose from. (I like to use my report cover photo or a screen snip of a logo or web page as my image.)  Adobe Color will allow you to adjust your palette to find, for example, brighter or more muted versions of your colors. Once you have the colors you want, you can get the RGV codes to create a custom color scheme for your report.

Two views of a line graph with four lines (four groups). The left is the graph as seen by the graph creator. It has a three black lines and a red line to emphasize results from a subgroup. The right version of the graph shows how a color-blind person sees it. The red line is black and the other lines are lighter gray

Color blindness checker: This exciting new multi-colored reporting world has its downside.  A small percentage of people are colorblind, so improper color choices may make your reports less understandable to them.  (The American Academy of Ophthalmologists estimates that colorblindness affects 8% of men and .5% of women.) So it’s a good idea to check your images through an app like Vischeck.

The two charts on the left show the results of a Vischeck on a line graph  I designed, where I made one line red to draw attention to results for one group. If you are not color blind, you will see that the left-hand chart has a red line to highlight a specific finding. The right chart shows what colorblind readers see: the line is darker, but it is not red. The darkness of the line does provide some contrast, so it is probably acceptable. But a different color or possibly a wider line would make that finding noticeable for all readers.

Printer: If your report is going to be printed and reproduced, chances are the copies will not be in color. I have learned the hard way to print my reports in black-in-white before distributing them to be sure the contrasts are still visible without color. If not, you can try varying intensity (gray versus black) or patterns (solid versus dotted lines).

There you have it: my go-to tools for creating evaluation reports. If you have others, I hope you’ll visit the NEO’s Facebook page and share them!

Here’s the full test for Stephanie Evergreen’s book: Evergreen, SDH. Presenting data effectively. Los Angeles, CA: Sage, 2014.


From Logic Model to Proposal Evaluation – Part 2: The Evaluation Plan

Photo of black and white cat with fangsLast week we wrote some basic goals and objectives for a proposal about teaching health literacy skills to vampires in Sunnydale.  Here’s what the goals and objectives look like, taken from the Executive Summary statement in last week’s post:

Goal: The goal of our From Dusk to Dawn project is to improve the health and well-being of vampires in the Sunnydale community.

Objective 1: We will teach 4 hands-on evening classes on the use of MedlinePlus and PubMed to improve Sunnydale vampires’ ability to find consumer health information and up to date research about health conditions.

Objective 2: We will open a 12-hour “Dusk-to-Dawn” health reference hotline to help the vampires with their reference questions.

There are also three outcomes that we have identified:

  1. Short-term: Increased ability of the Internet-using Sunnydale vampires to research needed health information.
  2. Intermediate: These vampires will use their increased skills to research health information for their brood.
  3. Long-term: Overall, the Sunnydale vampires will have improved health and as a result form better relationships with the human community of Sunnydale.

To get to an evaluation plan from here you have to know that there are basically two kinds of things you’ll want to measure: process and outcomes.

Process assessment measures that you did what you said you would do and the way you said you would do it. For example, you can count the number of classes you taught, how many people attended, and whether their survey responses showed that they thought you did a good job teaching.

Also you might want to show that you were willing to make changes in the plan if review of your process assessment showed that you weren’t getting the results you wanted.  For example, if you planned all your classes in early evening, but few vampires attended, you might interview some vampires and find out that early evening is mealtime for most vampires, and move your classes to a different time to increase attendance.  Your evaluation plan could show that you are collecting that information and that you will be responsive to what you see happening.

Outcome assessment measures the extent to which your project had the impact that you hoped it would on the recipients of the project, or even greater on their overall organizations or communities. We showed the first step of outcome assessment in last week’s assignment, but I’m going to break it down a little more here.  Put in basic terms, to do an outcome assessment, you state your outcome, you add in an indicator, a target, and a time frame to come up with a measurable objective, and then you write out the source of your data, your data collection method, and your data collection timing to complete the picture.  Let’s talk about each item here:

Indicator: This is the evidence you can gather that shows whether or not you met your outcomes.  If one of your outcomes is that the vampires have increased ability to research health information, how would you know if that had happened? The indicator could be their increased confidence level in finding health information, or it could be improvement in skills test scores given before and after a training session.

Target: The target is the goal that makes this project look like a success to you.  For example, if the vampires improve their test scores by 50% over a baseline test, is that enough to say you have successfully reached that outcome?  And how many of the vampires need to reach that 50% goal?  All of them? One of them?  Targets can be hard to identify, because you don’t want them to be too hard to reach but if they’re too easy your funder may not be impressed with your ambition.  Sometimes you can work with the funder or other stakeholders on setting targets that are credible.

Time frame: This is the point in time that when the threshold for success will be achieved.  So if you want to make sure the vampires increased their ability by the end of your training, then the time frame would be by the end of your training.

Data Source: This is the location where your information is found. Often, data sources are people (such as participants or observers) but they also may be records, pictures, or meeting notes. Here are some examples of data sources.

Data Collection Methods: Evaluation methods are the tools you use to collect data, such as a survey, observation, or quiz.  Here is more examples of data collection methods.

Data Collection Timing: The data collection timing is describing exactly when you will be collecting the data.

What does your final evaluation plan look like? 

Here is a sample piece of an evaluation plan for the Dusk to Dawn proposal.

Objective 1: teach 4 hands-on evening classes on the use of MedlinePlus and PubMed to improve Sunnydale vampires’ ability to research consumer health information and up to date research about health conditions.

Process Assessment: The PI will collect the following information to ensure that classes are being taught; expected attendance figures are being reached; teachers are doing a good job teaching classes (including surviving the classes).  Data will be reviewed after each class and changes will be made to the program as needed to reach target goals:

◊ Participant roster to measure attendance figures
◊ Class evaluations to measure teacher performance
◊ Count of number of teachers at the beginning and ending of each class to measure survival of instructors
◊ Project team will meet after the second class to review success and lessons learned and to consider course corrections to ensure objectives are met

Outcome Assessment:
Measureable Objective: In a post-test given immediately after each class, a minimum of 75% of Sunnydale vampire attendees demonstrate that they learned how to find needed resources in PubMed and MedlinePlus by showing at least a 50% improvement over the pre-test.

Based on Level 2 (Learning) in the Kirkpatrick Model, a test will be created with some basic health questions to be researched. Class participants will be given these questions as a pre-test before the class, and then will be given the same questions after the class as a post-test.  This learning outcome will be considered successful if a minimum of 75% of Sunnydale vampire participants demonstrate that their scores improved by at least 50%.

Last wishes, I mean, thoughts

This is not a complete evaluation plan, but the purpose of these two posts has been to show how you can go from a logic model to the evaluation plan of a proposal.  Don’t worry if all your outcomes cannot be measured in the scope of your project.  For example, in this Dusk to Dawn project, it might have been dangerous to find out if the vampires had passed on needed health information to their brood, even harder to find out whether the vampires had become more healthy as a result of the information.  This doesn’t mean to leave these outcomes out, but you may want to acknowledge that measuring some outcomes is out of the scope of the project’s resources.

As Grandpa Munster once said “Don’t let time or space detain ya, here you go, to Transylvania.”

Photo credit: Photo of 365::79 – Vampire Cat by Sarah Reid on Flickr under Creative Commons license CC BY 2.0.  No changes were made.












Last updated on Monday, June 27, 2016

Funded by the National Library of Medicine under Contract No. UG4LM012343 with the University of Washington.