In this article
Overview
This document provides sampling recommendations for clients who wish to survey a smaller subset of a large population – pulse surveys administered only to small groups or targeted subpopulations should census those respondent groups
Pulse surveys average a response rate of 50% - clients should expect no more than half of individuals who are administered a survey to return one
Press Ganey recommends a respondent count of 50 as a best practice
At minimum, respondent counts should be at least 30 where possible
Respondent count recommendations apply to the smallest grouping desired (e.g., demographic, work unit, etc.)
Survey administration periods may be extended provided the project has not already closed
Additional respondents cannot be added to the sampling pool once a survey project has begun
When to Sample
Sampling strategies should be employed when an organization wishes to collect feedback representative of a large population, such as an entire system or potentially an entire facility (depending on the size of the facility). For example, if an organization wants to simply conduct a pulse survey to get a snapshot of Engagement scores between annual census administrations, this can likely be accomplished without conducting a census pulse. This document outlines some considerations that are important when taking this approach to surveying.
If a pulse survey is already designed to target a specific targeted population or smaller group of respondents, then census pulse surveying should still be employed. Common examples of these would be clients looking to conduct follow-up measurement of Tier 3 work units, gathering information from specific respondent groups like RNs, etc. In these cases, since the starting population is already a smaller subset of the total organization, all members of these smaller groups should be given the opportunity to participate.
Clients who are unsure whether to sample can initiate a conversation with their account team so that all of the details can be gathered, and proper assessments and recommendations can be given. In most cases, however, more data is always better.
Response Rate Considerations
Press Ganey has historically achieved an average response rate of 50% with pulse surveys. There is a number of considerations that can impact this, however, such as the nature of who is targeted to receive the survey, the communication about the surveys from organizations to their employees, the perceptions that potential respondents have about how likely their feedback is to initiate changes, etc. All other considerations being equal, it should be expected that no more than half of a total sample will respond to the administered survey. Therefore, when selecting the initial sample population, simply double the established minimum sample size to determine how many individuals the survey should be administered to.
Minimum Sample Sizes
The general rule of a minimum respondent count being at least 30 respondents is in place because that is often viewed as the smallest acceptable sample for conducting inferential statistics or establishing comparative data performance (e.g., percentile ranks or comparisons to database averages). That said, more data is always better, so Press Ganey recommends a target of 50 respondents as a minimum sample to further minimize error.
These recommendations hold at the smallest level of aggregation desired. For example, if a client wants to drill down to measure specific job categories, then ideally there would be 30-50 respondents present within each job category. Whenever there is a desire to disaggregate larger samples to examine patterns and differences present among different subsamples, those subsamples must be adequately represented.
In some cases, it may be impossible to survey 30-50 respondents who fit a certain set of filters (e.g., physicians who specialize in pediatric nephrology split out between each facility within an organization). In these situations, the goal would be to simply collect as much data as possible from the subpopulation. All in all, this means that the level of specificity in respondent characteristics sought when administering a survey should be considered before making a final decision regarding the total number of people the survey is going to be administered to.
Survey Project Administration Notes
Provided that a given project has not already closed and is no longer accepting incoming surveys, clients may opt to extend administration periods to potentially increase response rates. Once a project is in motion, however, clients cannot add to the sample population. Given these considerations, it is better to administer pulse surveys to a larger group and underestimate response rates rather than close a survey without enough data.