- Measuring survey data requires scales of measurement, research questions, quantitative and qualitative data, historical analysis and more.
- To organize and present your survey data, first consider your audience, then create visuals and reports.
- To analyze your survey data, keep sample size, data correlation, statistical analysis and software in mind.
- This article is for business owners who want to know how to analyze survey data and apply it to their business to improve processes, brand reputation, customer satisfaction and more.
More companies struggle to properly use data than you might think. A Deloitte survey found that 37% of executive respondents said their companies are not data-driven, and 67% of executives expressed discomfort about accessing and using data. These trends are understandable given that data analysis often seems complex at first glance. However, in reality, data analysis is usually straightforward if you follow certain steps.
Survey data analysis is a great example. You can obtain so much information from a survey that, sometimes, you might not even know where to start your analysis. That said, once you find a starting point, analyzing your data should be easy as pie (which is also one of the types of charts that might come in handy when analyzing survey data). Below, we'll walk you through the fundamentals of how to analyze survey data and show you how simple analysis can be.
What is survey data analysis?
Survey data analysis comprises all the steps involved in obtaining, measuring, interpreting and visualizing the information you collect from your company's surveys. It is the practice of turning quantitative and qualitative data into meaningful insights. Armed with these insights, you can adjust your company's products, marketing campaigns and more to bring in additional revenue.
Types of survey data
Depending on the questions your survey aims to answer, the data you gather may fit into one of several types. These types include:
- Quantitative data. All numerical data is quantitative data. However, numbers mean different things in different contexts. That's why, in learning how to analyze survey data, you'll usually encounter four levels of measurement, three of which pertain to quantitative data. We'll explain these levels in more detail later.
- Closed-ended qualitative data. This type of data is nonnumerical and comprises simple answers. These answers can be as straightforward as yes or no, or they can be slightly longer answers that you provide as part of a multiple-choice survey. Generally, qualitative data is closed-ended, meaning respondents have limited choices of answers to choose from.
- Open-ended qualitative data. This type of non-numerical data offers respondents the freedom to answer questions however they'd like. If you ask your customers to respond to a question by typing original thoughts into a text box, their answers are open-ended qualitative data. Often, you can ask questions that begin with "why" or "how." Since open-ended data gives the customer more room to elaborate, it may contain less of your bias (which is inevitable when you craft a survey) than closed-ended data. Conversely, it can be more difficult to categorize and analyze, depending on how many subjects your survey includes.
Key takeaway: The data your surveys gather will be quantitative, closed-ended qualitative or open-ended qualitative. Numerical data is quantitative data. With closed-ended qualitative data, respondents have a limited set of responses to choose from. With open-ended qualitative data, respondents can elaborate with their answers.
How to measure survey data
As you structure your survey, you'll need to determine how you'll measure the data you'll collect. To appropriately structure your data for analysis, take the following steps:
1. Learn about the four levels of measurement.
There are four levels of measurement that each merit different statistical analyses. Some data can only yield a mode value, whereas others can yield a combination of mode, median, and mean. These levels are:
Any non-numeric type of data is nominal. From this data, you can determine your data's mode.
For example, if you're preparing a survey asking customers which of your products they like best, your data will be nominal. That's because your products aren't themselves numbers, and they have no inherent relationship to one another. Measurement along the nominal scale is primarily useful for measuring how many times each data entity (each of your products, in this case) is chosen by a respondent.
Ordinal data is ordered. You can obtain mode and median through ordinal data.
For example, you would obtain ordinal data from a survey asking customers to rank their preferences. If your survey includes five items to rank, then one will emerge as the most frequently chosen, and another will emerge as the middle value among the five.
An interval scale is a line along which your data points exist. You can obtain mode, median and mean through interval data.
For example, water is liquid at temperatures between 32 and 212 degrees Fahrenheit. Additionally, the 30-degree difference between 32 and 62 degrees is the same as between 92 and 122 degrees. However, 32 degrees will feel painfully cold, whereas 62 degrees will feel lukewarm, so different numbers have different meanings.
Ratio scales are interval scales that have true zeros. As with interval scales, you can get mean, median and mode from both.
For example, college exams are usually graded on a scale of zero to 100, though in most cases, passing grades are 65 and above. Additionally, in ratio data, a zero still has meaning – for example, a zero on a test might mean the student never took the exam. As such, zeros provide insight rather than lacking information.
2. Keep your research questions in mind.
Often, the questions you ask in surveys all tie back to several broader research questions that only you and your team know. Let's look back at the example we gave when explaining the nominal scale: There, your company asked customers to choose their favorite product. You might be asking this question to determine which of your products you should produce more or less often.
Put another way, the external question of "What is your favorite product?" answers the internal question "Which products should we produce more, and which should we produce less?" That's why you should link all external questions to an internal question as you measure your survey data. This approach will make it easier to connect your survey results with operational decisions.
3. Go quantitative, then qualitative.
All smart business decisions are numbers-driven; it's often best to look at your survey's quantitative results first before its qualitative results. In the aforementioned product survey example, if the smallest percentage of customers say their favorite product of yours is Product X, you'll know from this quantitative data to look at qualitative data about Product X.
Similarly, low customer satisfaction with a product can indicate it's time to look at your online reviews and find complaints. These complaints are also data; they give you the exact information you need to change your product in ways that should satisfy your customers.
Quantitative data reveals trends, while qualitative data can shed light on what's driving those trends.
4. Don't conflate causation and correlation.
One of the most fundamental principles of statistical analysis is that correlation does not imply causation. This statement means that if your data shows a relationship between two variables, then one variable is not necessarily the cause of the other. Two-variable statistical analysis is inherently limited – in the real world, far more than one factor influences most changes.
Think about it like this: If you see that millennials prefer one of your digital products, you might assume that millennials' love of digital technology explains why. But that might not be the reason. Maybe the product in question is low-cost, a common decision-making factor for many millennials. Or maybe that product's higher ethical standards are the reason millennials are more likely to choose it over others. Without additional data, you can't know for sure.
5. Compare current and prior data.
If this isn't your first survey, then you likely have past data to which you can – and should – compare your new survey data. If you discover that the product that customers now least often choose as their favorite was last year's favorite, ask yourself: What changed?
Maybe the answer is something obvious, like a product update that was great in theory but poor in execution. It could also be something subtler, like a shift in consumer spending habits. Whatever the case, take time to learn why your data changed and what you can do to return to prior levels. This task will likely take teamwork, which means you'll need to properly organize and present your survey data to others.
Key takeaway: When designing a survey for your business, select the proper scale of measurement, prioritize quantitative data over qualitative data, distinguish causation from correlation, and make sure you're making equitable comparisons with the data sets.
How to organize and present survey data
Measuring your data is just step one of how to analyze survey data. You'll likely also need to present your data to other members of your team or people outside your company. To effectively do so, take the following steps:
1. Consider your audience.
If you're presenting your survey data to your team, you probably don't need to spend too much time contextualizing the information – your team likely understands the context. If you're sharing survey data with people less familiar with your day-to-day operations, whether that includes other departments, board members, shareholders, or potential investors, providing context will be more helpful. That's why some presentation formats may work better for some audiences than others.
2. Build a report.
Here's a great example of how your presentation format will differ by audience: A report is necessary only if you need to contextualize your survey data. A thorough report expands on your survey findings over several chapters and works best for shareholders and investors. Of course, report creation is time-consuming – quick graphics will be easier, not to mention better, for most situations.
3. Use visuals.
Graphs and tables are the bread and butter of survey data presentation. Bar graphs, tables and pie charts are all great options for creating easy-to-read visualizations of your survey data.
Graphs and tables make for meaningful, memorable presentations, especially in slideshow form. You can also group the information from these visuals into larger infographics, which can be especially useful for incorporating open-ended qualitative data. Infographics are also more appropriate for sharing with investors and shareholders, since they include space for explaining data further.
Key takeaway: To organize and present survey data, consider your audience, use visuals and/or write reports.
Tips on how to analyze survey data
Before presenting your data, analyze it internally. When you do this, you'll better understand what your survey data says about your company, and you'll know which of your data points are the most important to present to your audience. Keep the following tips in mind as you analyze your survey's data:
1. Consider your sample size.
A fundamental assumption of any statistical analysis is that the sample size is appropriately large to yield meaningful results. To determine how meaningful your data is as you analyze it, compare your sample size to the ideal value for your population size. A sample size calculator can explain the difference between these numbers and illuminate what your sample size says about your data.
2. Complement data points with additional data.
Often, a data point taken from a survey is not as meaningful without additional data for context. For example, if you collect survey data about which of your products men and women prefer the most, don't just look at data regarding the least popular products in each category. Compare the most and least popular answers to come up with some potential conclusions about your customer base.
3. Cross-tabulate your data.
Let's say you ask rural, suburban and urban customers a closed-ended question. To cross-tabulate the data you receive, first create a table. Then, add rows for rural, suburban, urban, total percentages and total numbers. Next, add columns for all possible answers and total percentages and numbers. Finally, input your data.
With cross-tabulated data, you can more easily visualize differences in responses among groups. As such, this method facilitates the data complementation described above.
4. Use statistical methods to analyze your results.
With your survey data collected, you can make logical inferences and use statistical formulas to confirm many of these inferences. Three statistical methods are especially helpful for this purpose: ANOVA, t-tests and correlation analysis.
5. Use software.
If you still find survey data analysis overwhelming or confusing, data software can help. In many cases, you won't need to buy new software or platforms – your basic business programs should do the trick.
Key takeaway: To analyze your survey data, consider the sample size and how data sets relate, cross-tabulate your data, conduct statistical analyses, and use software.
What are the tools for data analysis?
You're probably already familiar with some data analysis tools. Excel is great for visualizing your data, and the popular survey administration platform SurveyMonkey includes analysis tools. Other popular data analysis tools include SAS, Tableau and Google Data Studio.