Survey Approvals

U-SAT assumes that you as the surveyor have reviewed and sought any necessary approvals, as needed, to administer your survey project at the University. Additional approvals may be needed by the following prior to submitting your survey request to USAS:

• Collegiate dean or senior-level University administrator (Associate Dean/Dean, Vice Provost, Vice President, Chancellor, President)
  • Required if your request includes:
    •  A total of 1,000 or more students on the Twin Cities campus or a combination of campuses; 
    • 15% or more of the current student enrollment from Crookston, Morris or Rochester;
    • 4% or more of the current student enrollment from Duluth.

You can learn more about enrollment on each campus on the UDIR website, or reach out to usat with any questions;

• Institutional Review Board (IRB)  

Survey Best Practices

A successful survey begins with a strong foundation. In this section, you will find resources and guidelines to help you construct a highly effective data collection instrument. Explore our guides to learn how to:

• Write clear, unbiased questions that yield actionable data.
• Structure your survey logic to reduce respondent friction.
• Design accessible instruments that are easy for all users to navigate.
• Implement strategies to boost your overall completion and response rates.

Use of Incentives

The use of incentives may be desired as a potential way to help increase your response rate (but not guaranteed). When considering incentives, we encourage you to consider lower amount gift cards from local businesses on or near campus that attract student business. No matter your incentive budget, big or small, follow the University's tax management office's guidelines - payment to students.  

Available Resources

• Conducting a Survey
• Survey Best Practices Overview
• Survey Project Design
• Surveys Simplified: A Brief Guide on Conducting Effective Surveys
• Guide to Improve Response Rates

Sample vs. a Census

A sample is subset of units or individuals from a larger, defined population of interest.
A census is a data collection effort conducted on an entire population of interest.

How do I know which to choose?

Due to community survey fatigue and the cost and resources needed, conducting a census is reserved for special cases. The goals of the vast majority of survey research projects are met with a sample. A well collected sample typically provides the same conclusions about the population of interest, as would conducting a census.
Please consult these criteria to determine if your survey necessitates a census. If any of the following are a true, then a census may be justified:

1. The project aims or goals stem from an administrative, programmatic, or other policy mandate to conduct a census.
2. There is longitudinal precedent from prior survey research or projects wherein not conducting a census would reduce the longitudinal validity of the survey.
3. The project scope includes one or more comparatively small population sizes. In these cases, one needs to argue that taking a sample would create methodological, statistical, ethical, or practical problems relative to conducting a census, such as an expected low response rate suppressing total response counts to an unusable level.
4. The project scope includes group comparisons wherein the underlying populations are highly imbalanced. In these cases, there may be justification to take sample(s) from the larger population(s) but employ a census of the smaller population(s).
5. Other conditions from the project scope or context indicating that taking a sample would be methodologically, statistically, or ethically unjustified, relative to conducting a census.
    

Sampling Best Practices

A well-designed survey is only as reliable as the audience taking it. Our sampling resources provide step-by-step guidance on how to reach the right respondents without contributing to campus survey fatigue. Browse these resources to understand:

• How to clearly define your target population.
• Methods for calculating the appropriate sample size for your project goals.
• The differences between probability and non-probability sampling methods.
• Institutional guidelines for pulling student email samples effectively and responsibly.

Use of Samples

Your sample should be as large as needed to achieve the goals of your project, given your project's constraints. For many surveys a sample size around 300 or up to 1,000 is sufficient. Determining an appropriate sample size for your project can depend on factors like, whether the results collected are meant to draw conclusions about a larger population, results intended for research publication, stipulations as outlined as part of a grant/contract or deciding where to order an office lunch from for an upcoming event.

If my project requires that I draw conclusions to a larger population, how large should my sample be?

For many survey projects, proportions or percentages of those who chose a given question option are calculated. For example, what percentage of respondents choose "Very Helpful" on a question? In these cases, please consult the table below as an approximate guideline for sample sizes that meet certain statistical properties, when drawing inferences to a larger population represented by the sample.

Note:The table assumes that the true proportion of the larger population choosing any given response option is 0.50 or 50%. This is the safest assumption. This table also assumes the sample is randomly drawn from the population, or is at least representative of the population on factors relevant to the project context.

Desired sample sizes for closed-ended survey questions, given different population sizes, margins-of-error and confidence levels.

Table Key

• Population size:The number of individual people or units in the larger population from which a sample is drawn and represents.
• MOE: Margin of Error. This is the precision around the sample proportion (plus and minus) that we would expect the true proportion to be.
• CL: Confidence or Certainty Level. Roughly, this is the amount of certainty that a survey result from a sample reflects what is true in the larger population.
• Putting MOE and CL together: For example, suppose the sample proportion who chose "Very Helpful" is 0.75, with MOE = 0.05 and CL = 95%. That indicates we expect the true proportion in the population to be anywhere from 0.70 to 0.80, with 95% certainty.

If I'm drawing conclusions to a larger population, how do I calculate the exact desired sample size I need?

All sample size calculations require statistical assumptions and depend on the type of information collected and the needs of the project. Some calculations are simple and some are complex. See resources below.

For example, suppose you ask survey respondents their age as a numeric question, with the goal to estimate the mean age in the larger population. Sample size calculations for a mean involve making assumptions that are more ambiguous than the assumptions for a proportion. In general, the more complicated the metric, the more complicated the sample size calculation.

There is not a single, ideal sample size for a given project. Every sample size calculation is based on statistical assumptions that should reflect the context and needs of a given situation.

Available Resources

Sample size determination should be informed by sound statistical guidance, whether based on your expertise or someone else's.

If you are unsure how to calculate the needed sample size for your project, please consult one of the many powerful sample size calculators already freely available. If you are unsure how to navigate a sample size calculator, please consult one of the many expert offices or centers here at the University of Minnesota.

Recommended Online Resources

• Statulator: Free online sample size calculator for the most common analysis situations.
• G*Power: Free downloadable sample size and power calculator for numerous analysis situations, including those of high complexity.
• General Online Research Sample Decision Tree
• Considerations on Maximizing Sample Size
• Nonprobability Samples and Validity: A Brief Socratic Dialogue

Consulting Offices and Centers at the University of Minnesota

• Biostatistical Design and Analysis Center (BDAC), Clinical and Translational Science Institute, Office of Academic Clinical Affairs.
• Center for Applied Research and Educational Improvement (CAREI), College of Education and Human Development.
• Liberal Arts Technologies and Innovation Services (LATIS), College of Liberal Arts.
• Lynn Y.S. Lin Statistical Consulting Center, Institute for Research in Statistics and its Applications (IRSA).
• Research Methodology Consulting Center (RMCC), College of Education and Human Development.
• University Survey and Assessment Services (USAS), Institutional Analysis.

Survey Software & Platform Support

The university provides access to a robust, enterprise-level survey platform through Qualtrics to meet a wide variety of research and administrative needs. Support and governance for Qualtrics, and other tools, are managed collaboratively to ensure data security and operational excellence.

Qualtrics

• Qualtrics Technology  
• Business Owner: University Survey & Assessment Services (USAS)
• Technical Platform Owner: Office of Information Technology (OIT)

Additionally, the University provides access to REDCap; it is a specialized, highly secure application built specifically for complex database management and clinical research. 

Important Note on Unauthorized Survey Tools

• We strongly discourage the use of non-approved, free, or low-cost external survey platforms (such as SurveyMonkey or consumer-grade polling apps). While these tools may seem convenient, they pose significant risks to both your project and the university:

• Data Security & Privacy: Unapproved platforms have not undergone the university's rigorous security review process. Using them to collect student, staff, or research data puts sensitive information at risk and may violate institutional privacy policies (such as FERPA).
• Data Ownership & Control: With many free or third-party tools, you do not have sole control over your data. The vendor's Terms of Service may allow them to access, mine, monetize, or retain your data indefinitely.

To ensure your data remains completely secure, legally compliant, and fully under your ownership, always use the preferred university-approved and supported platform detailed below.