*Understanding statistical significance in engagement surveys and pulse*

Statistical significance is an important concept for analyzing Pulse & Engagement Survey scores over time. This is also particularly important to consider when comparing results from different populations & attributes within your data.

The **certainty (%)** is the confidence in the accuracy of your Survey results. In this case, the confidence level of 95% certainty allows us to say that we are 95% sure that this sample of employees is representative of the population. To break it down further, we can say that if we run this survey 100 times, we’d expect results to match what we’re seeing now ± margin of error at least 95 times.

The **margin of error (MOE)** is a statistic that predicts the amount of the random sampling error in survey results. To calculate the MOE, you need to multiply the % by the scale. For example, if the average for a given question is 4, and your MOE is ± 10%, then there’s a 95% chance that the full population’s average is between 3.5 and 4.5.

As an Admin, you may be wondering how many responses you need to receive for there to be value in distinguishing the scores. After you launch your survey, you can take a look at the graph below to see what your margin of error is. The graph shows the relationship between the **size of the population** and **what % of that population needs to respond **in order to have a given margin of error.

## A few more things to note here:

- This does not mean that if you have fewer responses, you won't have usable results – the margin of error will just go up! The margin or error is particularly important when looking at results over time.
- The amount of times a question is asked is less important than the number of people who respond to it, and what that latter number represents as a percent of the total employee population. (per the above)
- It's also important to note that if you are planning to do any slicing on the data, e.g. gender, department, etc, that will have an impact on the statistical significance for that subset of employees. For example, if your company has 200 employees is split evenly between men and women, then you'll be looking at a population of 100 of each (for this attribute). This will then change the margin of error due to the decreased population size.

## Looking for ways to increase Survey Participation?

- As an Admin, you can write a reminder to employees to submit their surveys. This article here walks through how to remind those employees in Lattice! The reminder will only go to employees who have not yet submitted their surveys.