Hello, I'm Karen Monsen, Professor at the University of Minnesota School of Nursing. This is the fourth of five data applications modules for our social determinants of health data to action specialization. In this module we will explain and conduct ANOVA analyses and use the line graph to visualize our findings. These learning activities will equip you to conduct ANOVA analysis and create line graphs. In this course we focus on relationships and perspectives as we examine between group differences. In each data to action module, we consider ways to ensure we are ethical analysts and allies to the most vulnerable. Recall that participation is essential to unbiased findings. Given that we are comparing groups in this analysis, this is even more critical. As the very notion that we will be analyzing data by comparing multiple groups, challenges our thinking about who counts and who can be counted, and in what ways. We need to ensure that those who are represented in our data agree about their group membership that is assigned to them, so that our findings are not biased. Even if the groups come from the data, such as levels of education, or numbers of problems as in our analyses, the people who are represented in the data do not have a choice about how they are labeled in our analysis. They may or may not agree with their classification into a particular category. Keep in mind that every analysis needs to be planned with the community, and to be shared back with those who provided the data. So that group labels are fair and valid, and findings are interpreted with community perspectives at the center. This is becoming more important as group membership is the basis of many algorithms increasingly employed in health data to guide and monitor health. As we advance in our use of artificial intelligence technology and data in healthcare, it is critical that we are alert to the possible biases in data and the algorithms that cause harm. You can review the implications of algorithms and the importance of vigilance related to artificial intelligence and health in this paper. The analysis we use in this module builds on our analyses in course two, in which we compared means for two groups. But what if we have more than two groups and we want to understand differences between them? Analysis of variance or ANOVA allows us to do so. ANOVA tells us if there are any statistical differences between the means of three or more independent groups. That is, if differences between groups are due to chance. In each of our two textbooks, we point you to optional readings pertaining to ANOVA analysis. Likewise, we point you to optional resources pertaining to line graph visualizations. Why line graphs? When we are comparing groups you may ask. It's true bar graphs are usually used to show data for different groups. And line graphs are usually used to show change over time. However, bar graphs and line graphs can be created using the same data, as you can see here. I think it is much easier to recognize patterns in line graphs than bar graphs. Especially when the groups have a natural increase or decrease in their order. For example, we are looking at groups here that have increasing numbers of problems, from one problem to five or more problems. Another reason we introduced line graphs here, is that we already introduced bar graphs in course one. So here's our chance to focus on line graphs. Line graphs connect points of a variable of interest in the series by a straight line, usually showing progress over time. Simple line graphs show one line and multi series line graphs show more than one line. A variation of line graphs, parallel coordinates graphs, shows many variables connected by straight lines to show patterns in diverse data. Again, I encourage you to try out all kinds of blinds to gain experience and improve your visualization literacy skills. Just a quick reminder here, that to perform ANOVA analysis, you have to have three or more groups and a continuous variable of interest. If the group means differ overall, the ANOVA tells you how significant the differences between group means are. That is how likely these differences in means could have happened by chance. To tell which groups differ from each other, you'll need to do a follow up or post talk analysis, such as the Turkey or Bonferroni test. This video shares more information about such analyses. Tying all of this into our course material, recall the notion of collective impact that is critical to the success of a data to action initiative. Our collective impact example for this course is planned families. As you read it, think about the data sources that the project used and how the team used data to tell the story as inspiration, and especially to decision makers to create policy change. Recall also, that we have positive system archetypes to guide us in our data to action efforts. For each collective impact example, we draw on a positive archetype that underlies that example successes. In this case we see that the planned families project success leveraged the bite the bullet archetype. As you review the positive archetypes, do you agree that the bite the bullet positive archetype was instrumental in the success of the planned families project? Which other positive archetypes may have contributed to the success? Given everything you've learned so far, as well as your previous data experiences particularly with T-tests, you're ready to move on to course four, module five, Part 2, ANOVA and Visualization. In which we will answer the questions, which groups defined by their social determinants of health differ in their health outcomes? And is this likely due to chance?