One of the most exciting parts of a survey is analyzing responses because you’re likely to find patterns in the data that help you better understand your assumptions and learn about your topic of interest. However, like many elements in the process, your analysis and conclusions can be inaccurate without research, forethought, and expert input.
This blog is the culmination of my three-part blog series, which started with Thoughtful Survey Design, on defining your population and survey purpose, followed by Intentional Query Design, on question types and tips for writing queries. In this blog, I’ll cover how to share your data.
I hope this blog can be a springboard for further study about survey analysis. Many people focus their entire careers on statistics, data science, and data visualization, so if you don’t fully understand something, I recommend seeking help from one of these experts. Ensuring accuracy is crucial because numbers have a unique power for many people as they seem to be objective pieces of evidence. However, data can easily be misinterpreted or misconstrued.
Once you’ve closed your survey, pulled the results, cleaned up the data, and comb through your data to find trends and identify outliers. Qualitative data includes labels, names, and descriptive information that can either be unordered (nominal) or ordered (ordinal). In comparison, quantitative data reflects the measurement of a quantity where the difference between integers is the same throughout the scale.
When finding trends within qualitative data, look for frequency calculations like proportion, percent, and ratio. Even if numbers represent qualitative data, you can’t calculate means, standard deviations, or other statistics because the intervals between responses may differ.
When analyzing quantitative data, there are generally three qualities that describe data.
One fantastic way to discover trends and convey data is to use data visualizations such as histograms, scatter plots, maps, timelines, and word clouds, among other options. Data visualizations allow you and your audience to visually see trends.
Once you have your calculated statistics, the next step is to draw conclusions through inference, taking many data points and uses them to draw a general conclusion. One key element to remember when forming inferences is that your sample’s demographics should reflect the characteristics of the population. Ideally, your sample should be random and representative of the entire population.
To effectively extrapolate to your population, it’s essential to collect enough samples. Check out this sample size calculator to ensure you have adequate responses from your population, but a sample size of 30 is generally the minimum.
When sharing your findings, you may have multiple categories of audiences, including the respondents, current and potential customers, stakeholders, or media, and each may need a different set of messaging or visuals.
In addition to understanding the key findings, stakeholders may want to see your survey methodology, population descriptions, statistical significance claims, and notes about pervasive trends and notable outliers. Others may make future decisions or pose future surveys based on your work, so helping them be as educated on the data as possible may give them confidence.
The media is likely to appreciate newsworthy information. Often this means timely survey findings, which can include information that shows conflict, is relevant to many people, has proximity to their beat or geographic location, relates to well-known cultural figures, or appeals to emotion.
Here are some final tips to keep in mind when developing the messaging for each audience:
Surveys are a powerful mode of social research. With forethought and intentionality, a survey’s conclusions can be compelling and allow stakeholders to leverage the data for future business decisions.Tags: analysis, analytics tools, analyze, bias, biases, Customer Survey, Data, data science, demographics, descriptive statistics, extrapolate, extrapolation, inference, inferential statistics, Informed decision making, Know your audience, Likert scale, market research, Newsworthy, Online survey, population, qualitative data, quantitative data, research, sample size, sampling bias, Sensitive Data, skew, social research, statistics, Survey, survey design, writing tips