Line Graph vs Bar Graph

[Niksko]

Ok, so here's the story. A week ago we did our first SAC (School assesed coursework) in my Year 12 Biology class. A sac is roughly the equivalent of a test at our school, but they're usually more extended response type questions. The sac was on enzymes in general. The experiment consisted of placing some liquefied liver in test tubes and then adding pH buffers to each one (pH's 1,4,7,9 and 14 used - this is somewhat important). We then added some hydrogen peroxide to the test tube and observed the reaction. For those who don't know, the liver contains an enzyme, catalase, that aids in the breakdown of hydrogen peroxide into oxygen and water. One of the questions on the sac involved graphing the height of the bubbles (measured using a ruler ) against the pH of the solution. Here's where the controversy arose. I asked all my other teachers whether the graph should be a line graph or a bar graph. I assumed it would be a line graph, since pH is a form of continuous data (because there are infinite pH values - 3.5, 3.42343, 3.23423472384623 and so on) and our teacher said that we should use a line graph for continuous data. All my other teachers agreed including my maths teacher and the head of the maths department. But I got the SAC back and I got the question wrong - apparently it should have been a bar graph. I brought this up with the teacher, and he said that it should have been a bar graph because we only measured those set pH values and not all the values in between. The strange thing is, our textbook shows exactly the same graph as a line graph. I also brought this up with him, and he made some comment about the publishers having done tests for every single (infinite) pH value, which sounds slightly implausible to me. So my question is : Do Biologists obey different rules to the rest of the scientific world? Does how many measurements you take actually affect whether the data is continuous or not and therefore whether you use a line graph or a bar graph?

 

I doubt that he'll listen to me, but it will be a personal victory. I only lost 1 mark for doing the wrong type of graph, but it's a matter of principle. We also have another SAC tomorrow where we'll have to graph the percentage concentration of salt water against the change in weight of a piece of potato that is immersed in it. My knowledge of continuous and discontinuous data tells me that this should be a line graph as well, since percentage concentration is continuous, but I have a feeling if I got with line graph I'll get the same answer as I did last time. "We didn't test a 3.23723782342345% solution of salt water so therefore you should have used a bar graph"

 

Thanks in advance

 

-Nik

 

PS not sure if this is the right forum, but I didn't see any general sciences forum and I wanted to get opinions from as many people in as many different fields as possible. Feel free to move it mods if you think there is a better place for this.

[the tree]

From my (albeit limited) experience, scientists use graphics that convey information in the most representative and helpful way that they can think of, rather than following exact rules. Unfortunately, school doesn't seem to be the appropriate place for good communication and sometimes you just have to jump through hoops.

 

In this case, I'd have drawn a line chart with the experiments actually done strongly marked so as to not imply that you did more experiments than you did but also that you understand pH is more-or-less continuous.

[Niksko]

Ok, thanks for that. I should have mentioned that I indeed did put at the bottom "Actual data points marked with crosses. Additional data was generated by interpolation between the actual data points."

 

-Nik

[hermanntrude]

you did the right thing, scientifically. Bar graphs are far rarer in science than line graphs, and you can see why. Often, as in this case, the gradient of the line which interpolates between actual data points is significant in some way. With a bar graph, the line isn't there and so the gradient cannot be seen.

 

Of course... you CAN draw a bar graph and plot a line on top, i've seen that done...

[Glider]

Line charts are best used to illustrate changes in the same measure over time (e.g. pulse, temperature etc.). Bar charts best illustrate differences in central tendency (e.g. mean, median, mode) between different samples.

[iNow]

Glider's post reminded me of an example we use in the office.

 

When we want to show the entire company's stock and how it moved throughout the quarter, we use a line graph.

 

When we want to analyzy how much profit came in across each of the divisions and departments across the company, we use a bar graph (this makes comparing them to one another much simpler and you can quickly see which of your divisions is bringing in the income).

[Niksko]

Granted, I could have used a bar graph. But am I right in saying that it's not wrong to do a line graph?

 

-Nik

[NeonBlack]

I would say that showing the measured data points with a fit line connecting them would be most appropriate.

The only times I really see bar graphs is in experiments where you're counting something and where you have what we call "binning" of data, where data is grouped into relatively large discrete steps.

 

Unfortunately some teachers are never wrong. I really wonder how they can teach when they aren't even willing to learn.

[Glider]

Granted, I could have used a bar graph. But am I right in saying that it's not wrong to do a line graph?

 

-Nik

Not so much 'wrong' as 'less appropriate'. As you were comparing the pH values from different samples a bar chart is more appropriate as each of the bars represents a discrete sample.

 

Inherent within a line chart is the suggestion that each point represents a repeated measure of the same sample over time.

 

Hermanntrude is not quite correct in the statement that bar charts are rare in science. They are used specifically to compare the summary statistics (measures of central dendency and dispersion, e.g. means and SDs) from different groups/samples and are quite common, particularly where there are many values to present. However, Where there are fewer values to present, tables are more common.

 

The tree is quite right: "scientists use graphics that convey information in the most representative and helpful way that they can think of, rather than following exact rules.". However, they do conform to general convention (for the sake of consistency) and conventionally, line charts represent changes in a repeated measure over time (i.e. longitudinal) and bar charts represent measures of several different samples all at once (i.e. cross sectional).

[Niksko]

Ok, thanks so much Glider, that's exactly what I was looking for. You've explained to me a good reason why I should have used a bar graph. If my teacher had only explained it like this then I would have had no quarrels with it. Thanks again, and I'll be sure to keep that in mind in future.

 

Regards

 

-Nik

[Glider]

No worries