--- title: "Sampling" canonical: "https://jbi-global-wiki.refined.site/space/PT/4390920/Sampling" format: markdown --- Before you can begin collecting data, you will need to set a Sample Size against your Criteria. This may be a pre-determined number (Ad hoc, for example: a sample size of 20) or you may choose to use the Sample Size Calculator. There are 3 types of projects in PACES, as shown in Figure 1 below. All 3 include the option for ‘target sample size’ which is based on the same formula to deliver a reliable and valid sampling framework. ![Project-Type-Paces.jpg](media://6afc534b-516b-43bf-923f-81a6943e44c7) **PLEASE READ PRIOR TO USING THE CALCULATOR** - The PACES sample size calculator estimates the number of observations required to measure a population proportion with a specified level of precision. The calculation assumes independent observations and is most accurate when sampling approximates random selection. In practice, consecutive sampling of all eligible cases over a defined period is commonly used in clinical audits and is generally acceptable, provided there is no systematic time-related or selection bias. When consecutive sampling is used, users should consider potential temporal or clustering effects and, where appropriate, extend the sampling period or increase the sample size to improve representativeness. - When conducting two data collection cycles (e.g., baseline and follow-up), the calculated sample size is not intended to detect statistically significant differences between proportions (for example, a change from 10% to 80%). Detecting such differences would typically require a larger, hypothesis-driven sample size. However, identifying statistically significant change is not the primary aim of evidence-based quality improvement initiatives; rather, any positive change—regardless of magnitude—is considered a meaningful improvement in practice. - *In quality improvement, the aim is not to detect statistically significant differences but to estimate practice performance with reasonable accuracy. The sample size calculator helps ensure that results are meaningful and not distorted by random variation due to small samples. It also supports a balance between precision and feasibility, avoiding unnecessary data collection while still enabling confident conclusions. By applying a consistent and transparent approach, the calculator adds credibility and ensures that estimates reflect actual performance levels — making the audit process more rigorous and useful for guiding improvement.* - Each audit criterion in a project represents a distinct outcome, even if the data type (e.g., yes/no) is the same across all audit criteria. Therefore, sample size should be calculated **separately for each audit criterion**, based on its own expected compliance rate, desired precision, and margin of error. If users prefer to use the same sample size across all criteria, they should still perform separate calculations for each and adopt the largest calculated sample size as the final target. This ensures there are enough samples to reliably assess compliance across all key audit criteria. - The **sample size calculator is designed specifically for binary outcomes**—that is, audit criteria recorded as yes/no (Boolean) data, such as compliance with a particular practice. If the outcome of interest is not binary (e.g., continuous data such as pain scores or other numerical measures), this calculator is **not** appropriate. In such cases, users should seek additional statistical support to determine the appropriate sample size based on the nature of the data and the intended analysis. - For multiple sites: - The sample size calculator assumes equal allocation, a type of *disproportionate* stratified sampling. That is, the calculated sample size represents the total number of cases needed across all sites to assess compliance with a particular criterion (e.g. 50 records per site). In this case, the user must apply survey weights during analysis to avoid biased estimates. These weights adjust for the overrepresentation or underrepresentation of certain sites in the total sample. Statistical support is recommended when using survey weights. - An alternative approach is to perform separate sample size calculations for each **site–criterion** combination, using site-specific assumptions (e.g., expected compliance rates). This may be appropriate when: each site will be **analysed independently** (e.g., for local reporting or internal benchmarking); sites have **substantially different characteristics** or **expected compliance levels**, making a combined analysis inappropriate; or if the aim is to ensure **adequate sample at the site level**, not just overall. This approach increases data collection burden and complexity and is not required unless site-level comparisons or decisions are essential. # Set your sample size using the Sample Size Calculator - From the **Project Dashboard**, click on the **Data Collection Period** (usually you would begin with the **Baseline** cycle) you would like to apply your Sample Size to - Click on the **Sampling** tab - You will see a list of your chosen Audit Criteria - Click on the **Sample Size Calculator icon **at the top of that list ![Baseline-Paces.jpg](media://68ae54a2-0930-4bdb-8673-c446feaea7c8) - The window that pops up will be automatically populated with the recommended sample size per site, calculated using the default values. You can adjust these values if needed before proceeding. - Margin of error: Specify the acceptable margin of error. For example, if the estimate should fall within ± 5% of the true proportion, margin of error would be 0.05. - Desired confidence level: Typically, a 95% confidence level is used (corresponding to a z-score of 1.96). This indicates 95% confidence that the results are not due to random variation. - Expected proportion: Provide an estimate of the expected proportion for the outcome of interest. For instance, if it is anticipated that 50% of patients will be screened (i.e. compliance to screening), p would be 0.50 or 50%. - Click **Apply** ![Sample-Size-Calculator-Paces.jpg](media://18fdd381-e1ed-4de7-8314-6cca2d1597f6) - Scroll down to calculate the sample size of other sites (for multiple site audits). - After calculating the sampling for all sites, click **Save**. --- **Any questions? Please visit the JBI Help Desk for **[PACES support](https://jbi-global.atlassian.net/servicedesk/customer/portal/3)