Another way to show the information in a divergent bar chart, like the one I posted a few days ago, is in a split bar chart. Datawrapper staff wrote a blog post not recommending divergent bar charts, so they only offer a split bar chart.

I put together this split bar chart to get a better idea of what the Maryland Comprehensive Assessment Program (MCAP) results would look like displayed in this manner. In addition to the five score levels, I added a percent proficient column which is level 4 and level 5 added together. Percent proficient is reported in the official data chart and shown in the official column charts developed by the Maryland State Department of Education for the Maryland Report Card. The percent proficient is shown in green on the right of the centerline in my divergent bar chart.

I’m not sold that a split bar chart is always superior to a divergent bar chart. Especially when the data has categories that are distinctly good or bad and no awkward middle neutral category. However, I like the split bar chart. I especially like that it can be easily sorted by percent proficient, which includes levels 4 and 5. For this data, the percent proficient is more important than the percentage of students at each level.

I find it interesting that the percent proficient ranges from nearly 70% to less than 20%. With such a large gap, it is likely that there are methods that can be learned from the higher-performing counties to increase scores (and hopefully ultimately learning) in the lower-performing counties.

Notes About the Data

  • Values listed as 5.0% are acutally less than or equal to 5.0%. I am not sure if there is a way to show uncertain values in Datawrapper bar charts. In education data with small populations is often repressed. I could probably back into some of these numbers but I have not for this chart as the goal is to look at the data visualization.

I do not really know how to code. I am primarily a policy analyst who just loves data. I have been having trouble getting the demo version of a Divergent Stacked Bar Chart to render using the Amcharts plugin. My googling led me to this simple answer on StackOverflow regarding D3. I tried it with a D3 demo. It worked! First, try. I tried it with Amcharts. Once I got the file saved properly (I am missing something when combining all the pieces that I still have not yet figured out), it worked! So even if I do not know what is wrong with the Amcharts plugin for some of these types of charts, I can make it work. It takes longer and it is harder to debug, but if I want to do it I can.

Here is the code I used. Just change the URL to the URL of the code that you uploaded to your media section of the WordPress site.

`<iframe
  style="border: 0px;"
  src="www.yourwebsite.com/wp-content/uploads/2015/11/d3page.html"
  scrolling="no"
  width="100%"
  height="500px">
</iframe>`

Here is the graph. I used the 2019 Maryland Comprehensive Assessment Program (MCAP) data from 2019.

As I was working on trying to figure out how to make this graph render I read that is controversial and is not considered a good way to visualize data. But by that point, I was too far in. I wanted to see how it looked and judge for myself. I have not made any conclusions about the data or the visualization yet. I am just happy I figured out how to make the code work. This experience also means I will be able to try out D3 visualizations, which is a whole other rabbit hole. Enough explanation. I am going to test this webpage and take it live. The analysis is for another day.

Sources: Maryland State Department of Education, MCAP ELA and Math Performance Data.

The Maryland Comprehensive Assessment Program (MCAP) are the assessments that are used to meet federal Early Student Succeeds Act (ESSA) assessment requirements. MCAP has five score levels, a student who scores at level 4 or 5 is considered proficient on the material. Because of the pandemic, the MCAP was last administered in 2019.

I used data from the Maryland Report Card website to look at the percentage of students who are certified eligible for free and reduced-price meals compared to all students who took the assessment. I choose Math 5 because I wanted a group of students that had been in the school system for some time and to only capture first-time test takers. Also generally in middle school, a subsection of students start taking the Algebra I assessment (because they take Algebra) and thus are not included in the grade-level assessment. The data reports percentages of students, not the number of students. Thus, I can not easily calculate the non-FARMS subgroup, which might be a better comparison group since FARMS students are included in the all students category.

To bring some type of order to the data I reverse sorted it by all students. From a data visualization point of view I think this is an effective way of looking at the data. There is a lot of data, but it is easy to read and all of the counties are on the same chart. I wish I could better highlight the State average data with vertical lines or bolded dots, but I do not think that is an option in this software.

Remiation Rates of Recent Maryland Public High School Graduates Enrolled at Maryland Public Institutions

Today I take a look at the remediation rates of recent Maryland public high school graduates enrolled at Maryland public institutions as published by the Maryland Higher Education Commission in their 2021 Data Book. According to data published by the Maryland State Department of Education, 57,622 students graduated from Maryland public high schools during the 2017-2018 academic year. According to MHEC’s remediation data set, 25,575 students enrolled at a Maryland public institution in fall 2018. Thus, less than half, 44%, of the public school graduates are reflected in the map below. So, the data may tell us less about the quality of public schools than first thought. I will have to see if I can drill down further to this data. My first step will be to look at the number of high school graduates from each county; however, since the percentage of students who are low-income is different in each county, the college-going behavior is likely not consistent.

I was a little surprised to see that 25,575 students enrolled in a public institution immediately after high school because according to the college pipeline data published by the Maryland Longitudinal Data System Center, 17,410 students enrolled in a public institution immediately after high school. I wonder if MLDS is better able to separate students enrolled in multiple institutions. Further, the MLDS dataset only is capturing “degree-seeking” students.

Notes About the Data

  • Students may be enrolled at more than one institution. They are included in enrollment figures for each insituiton at which they are enrolled.
  • Data include all degree- and non-degree seeking students enrolled in credit courses.
  • Maryland residents are identified using their place of residence at the time of application to the insituion.
  • Maryland public instiutitions include community colleges and four-year public colleges and universities.
  • Recent Maryland public high school graduates are defined as those graduating from a Maryland public high school, identitifed using the College Board School Code, who graduated in the 2017-2018 academic year and first enrolled in higher education in Maryland in fall 2018. Analysis relies on high school graduation date and reporting of remedial assesment data; missing data for these variable may result in underreporting.
  • Maryland residents whose county of residence is unknown are included in insitutuional-level remedial data, but excluded from reporting by county of residence.
  • Salisbury University and St. Mary's College of Maryland do not offer remedial coursework.

Ploting the Data

Yesterday I looked at SAT scores in Maryland by county and income and for the incoming freshman classes of the public four-year institutions. Today I take a brief look at SAT scores by county and race/ethnicity from the same dataset published by the Maryland State Department of Education. I made a dot plot using Datawrapper with all the race/ethnicity categories available in the data set as well as “all students”. To be honest, the chart looks very busy and is rather hard to read. I changed the color scheme to reds and oranges to aid with distinguishing the categories, but it only helped a little. I could choose custom colors for each group, and would if I was intending to show this data to a wider audience, but since this is primarily for my own exploration of the data I decided I did not have the energy to make those choices today. I did decide to highlight the “all student” category to help with readability a bit.

An alternative visualization, and the one I have seen used at State Board of Education meetings, is a grouped bar chart. While I think that would work for smaller numbers of counties or race/ethnicities, I think that it is worst than the dot plot for a large amount of data. However, I may explore this visualization in the future.

SAT SCores by county and Race/Ethnicity

It is hard to draw any conclusions from this data. For one thing, I am unsure if this data represents public school students or all students who took the test from that county. For another thing, not all students take the SAT and different local school systems have different policies about pushing students to take the SAT.

Howard County stands out as having very high scores, for students of all races. From the data, I do not know if Howard County encouraged only high-performing students to take the SAT. I would be surprised if that was the case, but it is a possibility. Since there is information about the number of students that took the test, I might be able to infer the policy from that data or I might look at their website to see if they have a SAT policy.

Note: I likely will not be posting for a while as I will be on vacation.