Graphing: Bar and Column Graphs

Column Graphs

Let’s start with bar and column graphs. Column graphs are the default graph that Excel makes for you. It might not be appropriate, but Excel can’t make those decisions. Google Sheets is slightly smarter and might return to you a line graph or scatter plot. No matter what, whatever it is these program spit back at you upon first try, it’s only a suggestion. Before we go any further, we should note that a bar graph has bars extending horizontally across. These graphs are more commonly used in displaying money-associated growthAn increase in size and number of cells.. Column graphs have up and down columns and are commonly used in science.

A column graph is appropriate when you are simply comparing values of an independent variable. For example, I could run a taste test on different type of a food product (jam, beef, bread, etc…). Each column on my graph would represent a rating of a food product. I can also choose the sequence of the independent variable along the x-axis. I mean, why not? Right? Who’s to say that beef comes before pork or apricot jam should be last on the graph? These are the types of choices that you have to make when graphing, but what graph to choose is only one of those devilish details.

For the graph above, the student entered the data exactly as it was given to them. No decisions were made.

This graph was created from the same data. In this graph, the student made many decisions about entering in the data and even more decisions about the elements of the graph. Note the following:

• The the randomly chosen letter labels are replaced with the type of meat eliminates the need to ask what happened to B.
• The raw data number is displayed on each column, eliminating the need for a data table.
• The numbers all have the same number of significant figures.
• The columns are ordered in an increasing trend.
• The graph name has a Figure number so that it can be referred to in text.

Dealer’s Choice

With column graphs, we can rearrange the data because there is no predetermined sequence on the x-axis. Imagine if the x-axis were time, age, temperature, etc… Examine the graphs below. The one on the right is preposterous!

Time has a predetermined sequence. If we were to measure the height of a child for a year, we’d have to report that data in a predetermined sequence: the sequence of months in a year. It would make no sense if we listed the months in alphabetical order. We’d have the child growing and shrinking throughout the year. I don’t have kids, but I don’t think that’s how it works.

Stacked Column Graphs

Let’s say that we measure the fat rendered from each of these lovely meats at two points in time: after 1 minute and then again after 5 minutes. This is the resulting data table:

Meat Type	Fat Rendered at 1 minute	Fat Rendered at 5 minutes
Venison	1.8	3.2
Pork	3.1	4.8
Beef	4.0	7.0
Lamb	5.0	11.1

Notice some changes in the table. The data is organized in an increasing trend and all the letter labels have been swapped out for the name of the meats. With these changes in the data table, Excel or Sheets will have better suggestions for you. Here is what a student created with a little more knowledge about the choices they have to make when graphing. This graph is much better at showing data sets such as this one where there are two points in time when the same measurement is made. Excel very nicely put those labels for me down at the bottom, so I don’t have to tell you what the numbers mean. But, I can’t see the data in the blue portions of the columns! Excel made the choice for this blue. In order to make those data labels more visible, I need to either change the color of the text or the bars.

What will you do if your reader wants to the total grams of fat rendered at the end of the 5 minutes? You can make a proactive choice to sum your data and provide that number for them. Sometimes you have to put yourself in your readers’ shoes. A great way to do this is to think of yourself 5 years from now. What would you need on this graph that would make you remember all the details of the experiment?

Everything AND the Kitchen Sink

On the other side of the spectrum, students throw everything they possibly can into the graph. A graph is meant to summarize and synthesize highlights of the data set rather than show the reader all the raw data. Sometimes, the raw data is not the most appropriate display of the data.

In the example graph below, a student wanted to see if blueberry was preferred to apricot and peach jelly types. The student noticed that the strawberry and blueberry jellies were always low at the breakfast bar. They couldn’t locate enough strawberry for the experiment, so they just went with blueberry, apricot, and peach. They tested 5 people testing 3 jelly types and rating them on a scale of 1 to 3.

The graph above shows all the data from this experiment. It is organized by tester, but that’s not what we are interested in. We are interested in knowing how the jellies ranking in comparison to each other. Switching the columns and rows on the graph gives us the organizationThe structured arrangement of biological systems. below. The names of the jellies made it onto this graph, so that’s a plus!

Unfortunately, the graph above is still not the best visual display for the experiment. A reader may not be able to tell from this graph which of these jelly types was ranked highest among the three. WAS it blueberry? Or, is the sequence along the x-axis just random? Also, honestly, do I care about Marge, Matilda, Madeline, Margaret, Missy, Morgan, Maria, and Maggie? I mean, I care about them as people, but they are a controlled variable of this experiment. Maybe they don’t need to be on the graph?

They don’t. The graph below is all that is need to provide support for the original hypothesis about jelly types. The decision to employ the calculation of an average was made long before this graph was conceived. Using this calculation allows us to aggregate the data and show only what is important to the hypothesis: how blueberry fared against other jellies. The average has also been rounded to attend to significant figures.

Explore the use of other simple statistics.

This final graph includes a very subtle detail concerning the y-axis. The jellies were all rated on a scale of 1, 2, and 3. From the experiment description, it doesn’t seem as though the testers could have chosen values such as 1.5 or 2.75, etc… The y-axis has been scaled to have increments that are realistic to the experiment. This detail is devilishly subtle.

Can I Have Negative Bars?

Yes! It is possible to have the y-axis on any graph contain negative values. However this depends on the student entering in the negative sign when they enter in their data. It is surprisingly frequent for students to forget to enter the negative sign. Take the example below. This is a weight loss study. The patients were all instructed to eat the same diet and have the same levels of activity over a 3 month period.

The difference between the initial and final weights is presented as a percentage. I can’t say that a 10 pound weight loss for patient 1 is similar to the 10 pound weight loss for patient 3. In factA statement based on direct observation that is repeatedly confirmed., percentage-wise, Patient 1 lost twice as much as patient 3. Before I even go to graph, I can see that Patients 1 and 3 lost weight, Patient 2 gained weight, and Patient 4 has no change in weight, expressed as 0. I’ve been presented this table, complete with the negative signsObjective clinical findings observable by a provider (e.g., edema, fever).. When students are required to generate percent change, they frequently forget the negative sign.

Patient	Initial Weight	Final Weight	Percent Change
1	100	90	-10%
2	150	165	+15%
3	200	190	-5%
4	250	250	0%

After plugging in the numbers, a students created this graph in Excel.

Right away, we can recognize the mistake of graphing the independent variable as though it were dependent variable data. Excel thought that the patient number (1, 2, 3, and 4) was recorded data for the dependent variable and gave you the left-most series of columns on this graph. Those need to be removed.

Data Sets with Different Ranges

Even with removal of that group of columns, we would be left with two data sets extending over a range of 90 to 250 and another data set extending over a range of -10 to +15. Because these ranges are so different, the percent change bars are so small, and comparison is almost impossible without the data labels. Further complicating the data labels for the percent change group of columns is the placement of the x-axis labels.

This is resulted when the student removed the initial weight and final weight data and graphed just the percent change data.

The graph is getting closer to perfection. But close only counts in horseshoes and hand-grenades. When you change small elements of a graph, it usually results in the need to make more changes. Before moving on to the next paragraph, see if you can find 3 aspects of this graph that could use improvement. Put yourself in the reader’s shoes and don’t look back at the data table.

First, the label on our y-axis needs to be changed to “Percent Change.” Second, we have only one series of data. We no longer need what is called the series legend. This is the blue square and “percent change” on the bottom of the graph. It’s like having a table of contents for a book with only one chapter. You don’t need the chapter. The third and final correction comes with the placement of data labels and the x-axis labels. Your reader might not realize which numbers in your graph are the patient numbers and which are the data labels. We can either move the labels, or the axisSecond cervical vertebra; has the odontoid process (dens) for pivoting head (“no” motion).. There are quite a few ways to do this. Pictured above is only one way in which that could be done. Compare the two graphs to see how the second graph is more direct in communicating what the reader needs to see.

After reading all this, you might be intimidated about making graphs. A little fear can be motivating, but try not to let that get to a point where it stops you from trying. Because it is so important to put yourself in the position of your reader, it is essential to make a graph and then come back to it in 24 hours. Doing this can sometimes reveal to you some of the lacking or confusing detail in your graph. Proofreading graphs is a skill. It can be developed.