Best Practices

Data is essential for effective public health work. It helps us to identify trends, track progress, and target resources where they are needed most. But data is only useful if it is accurate, timely, and relevant.

However, data can also be misused or misinterpreted, leading to inaccurate conclusions and ineffective decision-making. To ensure that data is used effectively in public health work, there are a few best practices:

Interpret data cautiously:

There are limits to what kinds of claims you can make from a given data set based on how it was collected, what kind of questions were asked, and how big the data set is. Interpret data cautiously and consider other factors that may be influencing the results.

In general, you can make more (and broader) claims from bigger data sets. Larger samples increase the "power" of the study, meaning that they are more likely to detect small but meaningful differences. Finally, large sample sizes can also be helpful in reducing bias. This is because they provide a greater representation of the population, which makes it less likely that the results will be skewed by any individual characteristics of the participants.

Remember, though, that no matter the size of your sample, the way you ask survey participants questions shape what you can claim based on those data. For instance, if you ask a survey question that states "How great was the communication prior to the meeting?", you are implying that the communication that was great rather than asking them to rate the communication prior to the meeting on a scale from 1-10. Drawing conclusions from the first version of the questions may not be accurate, because maybe not everyone thought that the comunication prior to the meeting was great, so they may be unable to rate "how great" it was. 

We have more info on survey design here, and the Pew Research Center has a helpful and thorough guide on survey design: Writing Survey Questions | Pew Research Center 

Communicate results transparently and accessibly: 

When sharing results from data analysis, it is important to do so in a transparent way. You can never be completely unbiased, so be explicit and reflective about the limits and perspectives built into the data and its interpretation. 

Visualizing data in graphs, charts, maps, and interactive dashboards can help to make complex information more accessible and easier to understand. According to the CDC, adults in the U.S. have lower data literacy than adults in other comparable counties. Try not to rely on the numbers to speak for themselves.

However, be careful to be precise about the language you use to talk about data. For instance, describing an increase as "massive" is imprecise. Similarly, "significant" means something very specific in the context of statistics, so unless you're sure the thing you're talking about is statistically significant, don' say that.

The CDC Blog has good information on ways to think about audience and how to present data in accessible ways: 10 Health Literacy Tips for Reporting Data | Blogs | CDC 

Use multiple sources of data when possible: 

Relying on a single source of data can lead to biased or inaccurate results. Instead, use multiple sources of data to get a more comprehensive picture.

Use disaggregated data when possible: 

Data that are disaggregated by factors such as age, gender, and geography can help to identify patterns and target interventions more effectively. However, it's important to balance disaggregated with protecting privacy.

Small Community Complications

Of course, we realize that it can be harder to follow these guidelines in a small community where less data collection happens, and the small numbers make anonymity hard. 

One of the challenges of working with small local public health data sets is that it can be difficult to de-identify the data. This means that there is a risk that personal information could be released, which would violate the privacy of the individuals involved. There are a number of steps that can be taken to de-identify data, but this can be time-consuming and expensive. In addition, even de-identified data can sometimes be re-identified by matching it with other data sets. For these reasons, privacy concerns are often a barrier to sharing small local public health data sets. 

However, there are actions that can be taken to work around privacy concerns to still be able to ultize data that is crucial for planning, monitoring,  decision making, and evaluating. For example, data can be grouped together or "aggregated" when it's shared publicly, so no indivual-level data can be indenitifed. We can also include what data can be shared with who in a data-sharing agreement between parties, so each scenario is laid out before data-sharing begins.