Surveys: Don’t just set it and forget it!

Surveys are the rotisserie oven of the data collection methods. You simply “set it, and forget it!” That’s why it’s important to be strategic about how you’re reaching your target population. Otherwise, you may be leaving out key subsets of your audience—which are often voices that are already historically underrepresented.  

Is your survey equitable? 

Let’s say you want to send out a survey to library users, so you print off a stack of copies and leave them on the lending desk for patrons to take. While everyone in your target audience may have equal access to the survey (or in other words, are being treated the same), they don’t all have equitable access. Sometimes people may need differing treatment in order to make their opportunities the same as others. In this case, how would someone who has a visual impairment be able to take a printed survey? What about someone who doesn’t speak English? These patrons would likely ignore your survey, and without demographic questions on language and disability, the omission of these identities might never be known. Upon analyzing your data, conclusions might be made to suggest, “X% of patrons felt this way about x,y, and z.” In reality, your results wouldn’t represent all patrons—only sighted, English-speaking patrons. 

Who has access to your survey? 

Start by thinking about who you want to answer your survey—your target population. Where do they live? What do they do? What identities do they hold? Consider the diversity of people that might live within a more general population: racial and ethnic identities, sexual orientation, socio-economic status, age, religion, etc. Next, think through the needs and potential barriers for people in your target population, such as language, access to transportation, access to mail, color blindness, literacy, sightedness, other physical challenges, immigration status, etc. Create a distribution plan that ensures that everyone in your target population—whether they face barriers or not—can access your survey easily. Here are some common distribution methods you could use: 

  • Direct mail – Here’s more information about how to do a mail survey and it’s advantages and disadvantages. 
  • Online – For more information on how to make your online survey accessible, check out this article from Survey Monkey.
  • Telephone – In a telephone survey, someone calls the survey taker and reads them the questions over the phone while recording their answers. 
  • In-person – Surveys can also be administered in-person with a printed stack of surveys or a tablet. However, with this approach you might run into the dangers of convenience sampling

Depending on your target audience, surveys are rarely one-size-fits-all. The best plan is often a mixed-methods approach, where you employ multiple distribution strategies to ensure equitable access for all members of your target population. 

Who is and isn’t taking your survey?

Great! You’ve constructed a distribution plan that you feel can equitably reach your target population, but did it work? The only way to know for sure is by collecting certain demographic information as part of your survey. 

As library professionals, collecting identifying information can feel like a direct contradiction to our value of privacy. Yet, as a profession we are also committed to equity and inclusivity. When administering a survey, sometimes it’s necessary to collect demographic data to better understand who is and isn’t being represented in the results. Questions about someone’s race, ethnicity, income level, location, age, gender, sexual orientation, etc. not only allow us to determine if those characteristics impact someone’s responses, but also help combat the erasure of minority or disadvantaged voices from data. However, it’s important to note that: 

  1. You should always explicitly state on your survey that demographic questions are optional, 
  2. You should ensure responses remain anonymous either by not collecting personal identifying information or making sure access to that information is secure, and 
  3. Only collect demographic information that’s relevant and necessary to answer your particular research question. 

Compare the data from your demographic questions with who you intended to include in your target audience. Are there any gaps? If so, re-evaluate your distribution plan to better reach this sub-group(s), including speaking to representatives of the community or people that identify with the group for additional insight. Make additional efforts to distribute your survey, if necessary.

Conclusion

Inequities are perpetuated by research and evaluation when we fail to ensure our data collection methods are inclusive and representative of everyone in our target group. The absence of an equitable distribution plan and exclusion of relevant demographic questions on your survey runs the risk of generating data that maintains current power structures. The data will produce conclusions that amplify the experiences and perspectives of the dominating voice while simultaneously reproducing the idea that their narrative is representative of the entire population. Individuals who have historically been excluded will continue to be erased from our data and the overarching narrative.

Bad Survey Questions, part 2

Bad Survey Questions – pt. 2

Don’t let those bad survey questions go unpunished. Last time we talked about leading and loaded questions, which can inadvertently manipulate survey respondents. This week we’ll cover three question types that can just be downright confusing to someone taking your survey! Let’s dig in. 

Do you know what double-barreled questions are and how to avoid them?

When we design surveys it’s because we’re really curious about something and want a lot of information! Sometimes that eagerness causes us to jam too much into a single question and we end up with a double-barreled question. Let’s look at an example: 

         How satisfied are you with our selection of books and other materials? 

O    Very dissatisfied
O    Dissatisfied
O    Neither satisfied nor dissatisfied 
O    Satisfied
O    Very satisfied

Phrasing the question like this creates two problems. First, if a respondent selected “very dissatisfied,” when you analyzed the data you wouldn’t know if they were saying they were very dissatisfied with only the books, only the materials, or both. Second, if the respondent was dissatisfied with the book selection, but was very satisfied with the DVD selection, they wouldn’t know how to answer this question. They have to just choose an inaccurate response or stop the survey altogether.  

Survey questions should always be written in a way that only measures one thing at a time. So ask yourself, “What am I measuring here?” The double-barreled issue is in the second part of the survey question. What are you measuring the satisfaction of? Books and materials. 

Two ways of spotting a double-barreled question are: 

  1. Check if a single question contains two or more subjects, and is therefore measuring more than one thing.
  2. Check if the question contains the word “and.” Although not a foolproof test, the use of the word “and” is a good indicator that you should double check (pun intended) for a double-barreled question.

You can easily fix a double-barreled question by breaking it into two separate questions.

How satisfied are you with our selection of books?
How satisfied are you with our selection of other materials?

This may feel clunky and cause your survey to be longer, but a longer survey is better than making respondents feel confused or answer incorrectly. 

Do you only use good survey questions every day on all of your surveys, always?

Life isn’t black and white, therefore survey questions shouldn’t be either. Build flexibility into your response options by avoiding absolutes in questions and answer choices. Absolutes force respondents into a corner and the only way out is to give you useless data. 

When writing survey questions, avoid using words like “always,” “all,” “every,” etc. When writing response options, avoid giving only yes/no answer options. Let’s look at the examples below:

                    Have you attended all of our library programs this summer?  O Yes   O No

The way this question and response options are phrased would force almost any respondent to answer “no.” Read literally, you’re asking if someone went to every library program you’ve ever had, whether or not it was offered this summer or for their age group. Some respondents might interpret the question as you intended, but why leave it up to chance? Here’s how you might rewrite the absolute question:

How many of our library programs did you attend this summer?

Instead of only providing yes or no as answer choices, you should also use a variety of answer options, including ranges. For instance, if you also asked the survey respondent how many books they read during the summer, your answer options could be:

O    I have not attended any
O    1-3
O    4-6
O    7-9
O    10+
O    I do not know

Chances are, a respondent would feel like they easily fall into one of these categories and would feel comfortable choosing one that’s accurate.

Have you indexed this LRS text in your brain? 

In libraryland, we LOVE acronyms and jargon, but they don’t belong in a survey. Avoid using terms that your respondents might not be familiar with, even if they’re deeply familiar to you. If you use an acronym spell it out the first time you mention it, like this: Library Research Service (LRS). Be as clear and concise as possible while keeping the language uncomplicated. For instance, if asking how many times someone used a PC in the last week, be sure to explain what you mean by PC, and include examples like below: 

In the last week, how many times have you used a PC (ipad, laptop, android tablet, desktop computer)? 

Do you remember all the tools and tips we covered in our bad survey questions segment?

Hey, that’s ok if not! Here’s a quick review of things to do and don’t do in your surveys:

   Do use neutral language.

     Don’t use leading questions that push a respondent to answer a question in a certain way by using non-neutral language.  

   Do ask yourself who wouldn’t be able to answer each question honestly.

     Don’t use loaded questions that force a respondent to answer in a way that doesn’t accurately reflect their opinion or situation.

   Do break double-barreled questions down into two separate questions.

     Don’t use double-barreled questions that measure more than one thing in a question.

   Do build flexibility into questions by providing a variety of response options.

     Don’t use absolutes (only, all, every, always, etc.) that force respondents into a corner.

   Do keep language, clear, concise and easy to understand.

     Don’t use jargon or colloquial terms. 

 

Bad Survey Questions, part 1

In our last post, we talked about when you should use a survey and what kind of data you can get from different question types. This week, we’re going to cover two of the big survey question mistakes evaluators make and how to avoid them so you don’t end up with biased and incomplete data. In other words—all your hard work straight into the trash!

Do you think a leading question is manipulative? 

Including leading questions in a survey is a common mistake evaluators make. A leading question pushes a survey respondent to answer in a particular way by framing the question in a non-neutral manner. These responses therefore produce inaccurate information. Spot a leading question by looking for any of these characteristics:

  • They are intentionally framed to elicit responses according to your preconceived notions.
  • They have an element of conjecture or assumption.
  • They contain unnecessary additions to the question.

Leading questions often contain information that a survey writer already believes to be true. The question is then phrased in a way that forces a respondent to confirm that belief. For instance, take a look at the question below. 

Do you like our exciting new programs? 

You might think your programs are exciting, but that’s because you’re biased! This question is also dichotomous, meaning they must answer yes or no. While dichotomous questions can be quick and easy to answer, they don’t allow any degree of ambivalence or emotional preference. Using the word “like” also puts a positive assumption right in the question, pushing the respondent in that direction. A better way to write this question would be: 

How satisfied are you with our new programs?

In order to avoid leading questions, remember to do the following: 

  • Use neutral language. 
  • Keep questions clear and concise by removing any unnecessary words.
  • Do not try to steer respondents toward answering in a specific way. Ask yourself if you think you know how most people will answer. This might highlight assumptions you’re making.

Why are loaded questions so bad?

Similar to leading questions, loaded questions force a respondent to answer in a way that doesn’t accurately reflect their opinion or situation. These types of questions often cause a respondent to abandon a survey entirely, especially if the loaded questions are required. Common characteristics of loaded questions are: 

  • Use words overcharged with positive or negative meaning. 
  • Questions that force respondents into a difficult position, such as forcing them to think in black and white terms. 
  • Presupposes the respondent has done something. 

Let’s look at some examples of loaded questions. Put yourself in the shoes of different respondents. Can you think of someone that would have trouble or feel uncomfortable answering them?

How would someone who has never accrued late fees answer this question? This places someone in a logical fallacy. If they answer “yes,” they are saying that they once had late fees. If they answer “no” because they never started accruing late fees, then they are saying that they are still getting charged.

Why did you dislike our summer reading program?

How would someone who likes the summer reading program answer this question? This places someone in a logical fallacy. Any answer choices they select would be inaccurate. The question is loaded because it presupposes that respondents felt negatively about the program.

When you used our “ask a librarian” service, was the librarian knowledgeable enough to answer your question?

What if the librarian wasn’t knowledgeable, but was helpful? Maybe they didn’t know the answer, but they pointed you in the right direction so that you could find the answer. This phrasing causes the respondent to think in black and white terms, either they gave you the answer or nothing. Not to mention this question assumes you’ve used the service at all! 

Here are some ways to avoid using loaded questions:

  • Test your survey with a small sample of people and see if everyone is able to answer every question honestly. 
  • If you aren’t able to test it, try putting on multiple hats yourself and ask yourself who wouldn’t be able to answer this? 
  • You can also break questions down further and use what’s called “skip logic.” This means you would first ask respondents, “Have you used our ask a librarian service?” If they answer “yes,” then you would have them continue to a question about that service. If they answer “no,” they would skip to the next section. 

How useful was this blog post for learning about surveys and helping you file your taxes?

As the bad question example above might allude to, we aren’t done with this topic! In our next post, we’ll talk about double-barreled questions and absolute questions, so stay tuned! As always, if you have any questions or feedback we’d love to hear from you at LRS@LRS.org.

Are you ready to learn about surveys? Ο Yes Ο No

1. What is a survey?

If you’ve ever responded to the U.S. Census, then you’ve taken a survey, which is simply a questionnaire that asks respondents to answer a set of questions. Surveys are a common way of collecting data because they efficiently reach a large number of people, are anonymous, and tend to be less expensive and time-intensive than other data collection methods. The purpose of surveys is to collect primarily quantitative data. Surveys can be administered online, by phone, by text, or in print. 

2. Should I use a survey to collect data? 

In our last post we talked about how to decide which data collection method fits your evaluation. The first step is figuring out your evaluation question and determining if a survey can answer it. Surveys might be the right option if you want to collect information from a large number of people about their needs, opinions, or behaviors. For instance, they can help you determine what patrons learned from a program, the different ways people use resources at your library, or even what services non-users might be interested in, among other things. 

Surveys might not be the right method if: 

  You’re primarily trying to answer questions of why or how, as these work best as open-ended questions and are better suited for interviews or focus groups. Surveys can contain open-ended questions, but they are typically supplemental to the closed questions that make up the majority of the survey.

  Participant self-reported behavior is likely to be inaccurate. For instance, surveying children on how engaging a program was might not be the best approach.

In addition to these criteria, you should also consider time and costs associated with a survey and whether these line up with the resources you have available. A more thorough breakdown of the costs associated with a survey can be found here

3. How many of these question types have you used? (Mark all that apply)

Although survey questions can be written in a multitude of ways, ultimately every question is either closed, open-ended, or a combination of both. Open-ended questions ask the survey respondent to provide an answer in their own words, like in the example below.

Why did you decide to read this blog post? 

Open-ended questions allow the evaluator to collect robust data by not limiting the respondent to a list of possible answers. For instance, maybe you’re reading this blog right now because your cat walked across your keyboard and accidentally clicked on the link. The survey is unlikely to include that answer option on a closed question, but an open-ended question can capture that sort of qualitative data. 

Although there are many pros to using open-ended questions, there are also some downsides. Most open-ended questions take a long time and a skilled evaluator to analyze the qualitative data they produce. That’s why closed questions are more commonly used on surveys.

Unlike open-ended questions, closed questions provide a set of answer choices and produce quantitative data. Let’s explore some different types of closed questions.

Multiple choice questions allow respondents to select one or more options from a set of answers that you define. A common drawback of multiple choice questions is that they limit answers to a predetermined list like below, which may not reflect everyone’s responses. Often the problem is solved by adding an “other” option where a respondent can write in their answer if it isn’t part of the list. 

How do you feel today?

  Happy

  Sad

  Other, please specify: ___________

Adding an “other” option makes part of this question open-ended. When you analyze the data for this question, pay close attention to the percentage of respondents who chose “other.” If it’s a large portion (usually more than 10 percent), you will need to do some qualitative analysis of these answers.

Likert scale questions give respondents a range of options (usually five or seven choices). They’re often used to gauge someone’s feelings or opinions and can be written as statements instead of questions (see below). Writing a likert scale can be tricky because you need to make sure your response options are balanced. We’ll talk about that more in depth in our next post. Here’s an example of a likert scale question.

I am learning something from this post on surveys.

  Strongly agree

  Agree

  Neither agree nor disagree

  Disagree

  Strongly disagree 

Demographic questions ask respondents about characteristics that are descriptive, such as age, gender, race, income level etc. Demographic questions allow you to gain a deeper insight into your data. For instance, I could use a question that asks a respondent’s age to analyze whether younger respondents were more likely to say they “disagree” or “strongly disagree” on the question above. 

These are the most common question types you’ll find on a survey, but for a deeper dive on different question formats, such as matrix, dropdown, and ranking, check out this article from SurveyMonkey. 

4. Stay tuned for surveys pt. 2?       Yes       Definitely     I wouldn’t miss it for the world

We’ve all probably taken a survey, but there’s a lot that goes into making them balanced, understandable, and unbiased. In our next post we’ll cover why the question above should never be on a survey and other common mistakes people make when writing survey questions. 

Does the (Data Collection Method) Shoe Fit?

You wouldn’t go hiking in a pair of dress shoes, right? Like the variety of shoes in your closet, there are a variety of data collection methods in all different shapes and sizes. The trick is finding which data collection method fits! Today’s post will help you determine which method is best for your evaluation.

What are Data Collection Methods?

Data collection is the process of gathering information from different sources with the goal of answering a specific question (your evaluation question). The method, or procedure, that you use to collect your data is your data collection method. Four common ones are: surveys, interviews, focus groups, and observations.

  • Survey: questionnaires that ask respondents to answer a set of questions. While these questions can be closed or open-ended, the purpose of surveys is to collect primarily quantitative data. Surveys can be administered online, by phone, by text, or in print. 
  • Interview: a conversation between two people—an interviewer and an interviewee—during which the interviewer asks primarily open-ended questions. Interviews may occur face-to-face, on the phone, or online. Interviews provide qualitative data.
  • Focus group: a dialogue between a group of specifically selected participants who discuss a particular topic. A moderator leads the focus group. Focus groups provide qualitative data.
  • Observation: a person (the researcher or evaluator) observes events, behaviors, and other characteristics associated with a particular topic in a natural setting. The observer records what they see or experience. Observations may yield quantitative or qualitative data.  
How to Pick the Right Data Collection Method

By this point in your evaluation you should have: 

Determined the goals and scope of your evaluation

  Written your evaluation question(s)

If not, you can circle back to those posts here and here, respectively. Now you’re almost ready to start collecting data—the fun part! First you need to decide which data collection method to use. Take a look at the pros and cons of each data collection method in the chart below. Use this to help you narrow down which methods might fit your evaluation.

To further narrow down your data collection method search, ask yourself the questions below. Do your answers rule out any of the methods? Reference the pros/cons chart for help. 

  What is most essential to you? Consider whether it is important for you to answer questions of how and why (more likely qualitative data) or what, how often, and to what extent (easier with quantitative data). 

  What will you be asking? Complex topics may lend themselves better to methods that allow for follow-up questions. Taboo topics may require additional anonymity. Think about what methods will make your participants feel most comfortable and safe responding to you.

  What are your constraints? Be realistic about the amount of time and resources you have. Choose a method that meets those constraints.

Conclusion

If none of these methods seem to fit your needs, don’t be afraid to branch out and find a collection method that is best for you or take a mixed-methods approach and use multiple techniques! For some other interesting ideas, here’s some additional articles on a collaborative photography method, oral histories, and other creative evaluation methods.

In our next post we’ll start our deep dive into the most popular data collection method—surveys. Stay tuned!

The Dynamic Data Duo: Quantitative and qualitative data, part 2

In our last post we introduced you to the dynamic data duo—quantitative (number) and qualitative (story) data. Like any good superhero squad, each have their own strengths and weaknesses. Quantitative data can usually be collected and analyzed quickly, but can’t really yield nuanced answers. Qualitative data is great at that! However, it often takes a lot of time and resources to collect qualitative data. So just like Batman and Robin, who balance out each other’s strengths and weaknesses when they’re together, both can also have successful solo careers. This post will walk you through a simple process to determine which data hero is right for the job!

Step 1: What is your evaluation question?

Let’s say we’re doing an evaluation where we want to find out if attending storytime helps caregivers use new literacy skills at home. If we go up to every caregiver and simply ask them, we’ll get a lot of yes/no answers, but not a whole lot of details. For example, imagine if we asked you right now: “Is this blog series helping you use new evaluation skills at work?” You might respond: “Uh…I don’t know. Maybe?” It’s a hard question to answer accurately. Often the evaluation question is too complex to directly ask participants.

Step 2: Break your evaluation question down into simple questions. 

Imagine calling up the Justice League and asking, “Hey, can you save the world?” They might answer yes, but will we know if they have the right skills or perhaps have other plans today? Similarly, our evaluation questions are often broad and abstract. We can’t always ask it outright and get a useful answer. So let’s look at some ways we can break our evaluation question down into simpler questions. 

As a reminder, our evaluation question is “does attending storytime help caregivers use new literacy skills at home?” Go word by word and see if you can come up with additional questions that would break the concepts down further. For instance, “does attending…” What are we assuming/what don’t we know? 

  • Did the caregiver attend a storytime session? 
  • Why or why not?
  • How many times did a caregiver attend a storytime session?
  • Which storytime sessions did the caregiver attend? 

Continue on with the rest of the evaluation question, keeping in mind you might not come up with simpler questions for every word or phrase. 

“Caregivers”

  • Who are the caregivers? 
  • Were they already using the literacy skills taught during storytime at home prior to attending a storytime? 

“New literacy skills” 

  • Are caregivers learning new literacy skills during storytime? (If caregivers aren’t learning new literacy skills at storytime, they can’t then use those skills at home!)
  • Why or why not? 
  • What new skills are they learning? 
  • How many new skills are they learning?

“At home”

  • Do caregivers use new literacy skills from storytime at home? 
  • Why or why not?
  • How often do they use new literacy skills from storytime at home? 

Step 3: Determine if each sub-question can be answered with numbers or a story

Go back through your list of sub-questions and try to answer each one with a number. Can you do it? If so, the question would give you quantitative data. If not, it might be a qualitative question. 

Let’s look at the question, “What new literacy skills are caregivers learning during storytime?” We need words to answer this question, not numbers—right? Not necessarily. We could create a list of 10 literacy skills that we taught during storytime and ask caregivers to check which ones they learned. By creating these parameters, we’re limiting the response options to a finite quantity (10 possible choices) and can count how many people choose each skill. This process transforms what would be an open-ended question yielding qualitative data into a question yielding quantitative data. 

You can generally apply this process to questions that either have a finite number of options or where a likert scale is appropriate. However, there are numerous (no pun intended) cases where you’ll want more nuanced, qualitative answers. For instance, try answering the question, “Why did you attend storytime today?” with a number! We could still create a list of possible answers, but it’s likely that someone would look at those choices and feel like none of them really fit. If we want to better understand our caregivers’ reasoning, then we don’t want to limit their responses. We want a story—we want qualitative data.

Step 4: Batman or Robin? Or both?

Now that you’ve classified your questions as quantitative or qualitative, do you have the means (capacity, resources, etc.) to collect data on all of them? Remember the pros and cons of each data type and review which questions are most important to you. Are a majority of them qualitative or quantitative? Knowing which type of data you need to collect will help you decide which data collection method to use. Our next several blog posts will address the different data collection methods you can use and their pros and cons, so keep reading!

Ready to meet your (data) match? Introducing number data and story data

Shows two different shapes to illustrate the two different categories of data

Hey, there! Welcome to 2021! We’re glad to see you here. It’s a new year and we’re ready to dive into research methods. Not what you expected to rejuvenate you in 2021? Well, hold on—research methods are actually pretty rad. First, though, what are they? 

Research Methods

Research methods are the different ways we can do the research or evaluation. If you’ve already tried out our tips on doing desk research, you may have found that the data you need is just not out there. You’re going to have to collect some data yourself! 

What kind of data should you collect? Two very broad categories of data are quantitative and qualitative data. Quantitative data are numbers data and qualitative data are story data. Wait—isn’t all data numbers? Nope! Story data are real! 

Quantitative Data: how much or what extent

What kind of information can quantitative data provide? Think about questions that you could answer with a number. Here are some examples from libraries:

  • How many books were checked out this month?
  • How often did families attend more than one storytime in a month?
  • What times for storytime have the highest attendance?
  • What percentage of our patrons rely on mobile services for library access?

You can see from the examples that quantitative data can answer questions about how much, how often, what, and to what extent. Quantitative data can often be collected by consulting data you already track within your library or by distributing a survey. This data can generally be collected and analyzed relatively quickly. The downside to quantitative data is that it can’t tell you how or why something is a particular way. If you collect data on how often families attended more than one storytime in a month, you still don’t know why some families came more often. That’s where qualitative data comes in. 

Qualitative Data: why or how

What kind of information can qualitative data provide? Think about questions that are difficult to answer with a number. The questions below cover the same topics as the quantitative questions above, but approached in a qualitative way:

  • Why are some patrons super-users? 
  • Why do some families attend storytime once and never return?
  • What reasons other than convenience determine whether families attend storytime? 
  • How do patrons who use the mobile services feel about the library in general?

You can collect some qualitative data on surveys by asking open-ended questions. You also can collect qualitative data from observations, interviews, and focus groups. While it yields detailed information, qualitative data collection and analysis can be complex and time-consuming. These data don’t always yield information that is actionable right away. Going back to our storytime example, if we ask why some families attend storytime once and never return, we may get a lot of different answers and need to spend time looking for common themes. 

How to choose?

Now that you know what both types of data could look like, how do you decide what data is the best to collect for a project? Did you notice how those quantitative and qualitative questions matched up on similar topics? That was on purpose! Different types of data can give you insight into different aspects of your evaluation question. 

To get the most meaningful results, it’s a great idea to collect both quantitative and qualitative data for your project. They can work together to provide a more complete picture of the topic. An easy way to incorporate both is to create a survey that includes mostly quantitative questions, but also a few key qualitative questions.

Now, is it always realistic that your organization has time and capacity to collect both types of data? Not really, right? That’s ok. The most important thing is to match the kind of data you collect with your evaluation question. 

Conclusion

Now you have a basic idea of how quantitative and qualitative data are different and how they can be used to find out different kinds of information. In our next post, we’ll show you a simple process for breaking down your evaluation question into smaller questions and determining if you need to use quantitative or qualitative methods. 

 

Happy Holidays!

Snow globe with mountain scene inside

We have loved having you all with us on our data journey! We are putting our blog series “Between a Graph and a Hard Place” on hold in December.

We’ll be back in January with more exciting information about doing your own evaluation, including specific ways of collecting data like surveys, focus groups, and observations.

In the meantime, we wish you all happy and safe holidays! Special thanks to Mary Bills for the beautiful artwork.

How to conduct a secondary research evaluation in four steps

 

In our last post, we assured you that it was possible to complete an evaluation without ever leaving your desk! So as promised, here’s how to conduct a secondary research evaluation in four simple steps.

Remember, in the scenario in our last post, you are a youth services librarian at a rural public library that serves a population of 4,000. You want to know if your summer learning program is effective at engaging youth with developmentally enriching content (our evaluation question). You don’t have the time or resources to go out and collect your own data, so you decide to conduct secondary research instead to help you make a decision about how to improve your summer learning program. In our last post, we talked about the different ways you can conduct secondary research. Now we’re going to apply the multi-data set approach. Here’s how you can do that in four simple steps.

  1. Identify your evaluation question

We’ve already determined that our evaluation question is: do summer learning programs engage youth with content that is developmentally enriching? If you need help determining your own evaluation question, you can revisit our post on the topic.  

  1. Identify a secondary data set (or sets)

Review the existing literature on your topic of interest. In our last post, we identified different external and internal data sources that you can investigate. You may find other libraries, organizations, or agencies that have explored your topic and collected data. Reach out and ask for permission to use their data if necessary. For this example, let’s say we found this publication of key research findings on public libraries’ role in youth development. To get a well-rounded understanding of your topic and enough data to analyze, you’ll probably need to find multiple data sets. For the purpose of this post, we’ll just look at one.

  1. Evaluate secondary data set

Congrats, you’ve chosen a data set! Sometimes that can be the hardest part. Now we need to evaluate whether we chose the right one. To do so, we’ll try to answer the questions below. If you need additional help understanding how to answer these questions, read this first.

  • What was the aim of the original study?
  • Who collected the data?
  • Which measures were employed?
  • When was the data collected?
  • What methodology was used to collect the data?

Based on what we found, the data set we selected comes from a reliable source and is relatively recent. Some of the libraries in the study also serve a population that is close in size to our own. However, the aim of the original study is a little different than ours (the role of libraries as a whole on youth development). Therefore, we might want to find an additional data set specifically on summer learning to help us answer our evaluation question. If one of the public libraries who participated in the study has a similar population or demographics as our library, we could also reach out to them directly and ask to see their data.

  1. Analyze secondary data set

Pick the variables from your data set that are most relevant to your evaluation question. You may also need to recode variables. For instance, maybe the data set includes a variable for school district, but that’s not important to you. You’re more interested in seeing if there’s a correlation between poverty and youth development. Therefore, you can recode the school district variable by percentage of people who live below the poverty line in each district (using another data set in tandem!). Here’s a short video on how to recode variables in Excel. Once you’ve got all your ducks in a row, you’re ready to employ all your statistics mastery (mean, median, mode, correlation, etc) to draw conclusions from your data. 

Conclusion

There you have it! An evaluation without ever leaving your desk. As always, if you have any questions or comments, please feel free to reach out to us at LRS@LRS.org. In our next post, we’ll cover another evaluation methodology, so stay tuned.

Conduct an Evaluation Without Ever Leaving Your Desk

Are you ready to get your hands dirty and start evaluating? After covering outcomes, the logic model, evaluation questions, and research ethics, our next step is to start collecting data. I know many of you might be thinking, “But we’re still in a pandemic. How could we possibly do an evaluation now?” Well that’s one of the many advantages of secondary research.

What is secondary research and why should I do it? 

Secondary research involves data that has been previously collected by someone else. As opposed to primary research, where you collect the data yourself, secondary research uses “available data” and various online and offline resources. Also called desk research because you can do it without ever leaving your desk, it’s a particularly useful evaluation method when you have a limited ability to collect your own data. In many ways, it is similar to a literature review—it gives you an idea of what information is already out there. However, secondary research focuses more specifically on analyzing existing data within the confines of your evaluation question. 

What are different ways I can use secondary research? 

Secondary research can be useful whether you have limited resources and time or have no limits whatsoever. Your evaluation might only consist of secondary research or it could simply be the first step. No matter what your goal, secondary research can be helpful. 

Let’s say you are a youth services librarian at a rural public library that serves a population  of 4,000. You want to know if your summer learning program is effective at engaging youth with developmentally enriching content (our evaluation question). You don’t have the time or resources to go out and collect your own data, so you decide to conduct secondary research instead to help you make a decision about how to alter your summer learning. 

One approach you could take is to conduct a classic literature review and in the process, look for studies on topics that align with your evaluation question. If possible, also look for data that is similar in some aspect (demographics, size, location, etc.) to data you would collect yourself. For instance, you might find a study on how public libraries facilitate youth development. Within the study, you see data was collected from another rural library. Perfect! 

Depending on your evaluation question, you may even find multiple data sets that are useful and relevant. For example, let’s say we find data on summer learning from three different libraries. Each recorded what their main activity was and participation numbers. Great! We can compare these data sets and extrapolate some conclusions. Just remember, when using multiple data sets, it’s helpful to have a variable they all share. In our example, even if one library recorded participation rates in weekly numbers and another in monthly, we can recode the data so that the variables match.

Even if you also plan to collect primary data, secondary research is a good place to start. It can provide critical context for your evaluation, support your findings, or help identify something you should do differently. In the end, it could save you time and resources by spending a little extra time at your desk!  

What are the different kinds of secondary data I can collect?

Internal sources

You don’t have to go far to find data. Your library has probably been collecting some sort of data ever since it opened! This is called internal sources—data from inside your organization. Here are a few common examples:

  • Usage data (visits, circulation, reference transactions, wifi, etc.)
  • User data (ex: number of registered borrowers)
  • Financial data 
  • Staff data
  • Program data (attendance, number of programs, etc.)

External sources

Maybe your library doesn’t have the data you’re looking for, like the demographics of children in your service area. Perhaps you are more curious about what other libraries have found successful or challenging in their summer learning programs. Or maybe you want to look at peer-reviewed research  about summer learning loss (summer slide). These are all examples of external sources—sources from outside your organization. Here are a couple of common examples:

  • Government sources
  • State and national institutions
  • Trade, business, and professional associations
  • Scientific or academic journals
  • Commercial research organizations 

Conclusion

Now you have the what of secondary research. Next time we’ll cover how to do secondary research in four simple steps, so stay tuned. As always, if you have any questions or comments, please feel free to reach out to us at LRS@LRS.org