I pushed my AP Stats learners to use Plot.ly this year for their projects instead of JMP as I have in previous years. I am not sure if I like it better or worse, but I definitely have some frustrations.
Let’s go through the good points first.
1. After the learners had their data in Excel, it was very easy to import the .xlsx file into plot.ly. The learners had to make some changes to the first row, but that is to be expected.
2. As long as the learners hit “save” then they did not have to worry about losing their project. I have had previous years where learners lost flash drives or computer files the week before the project is due and had to start from scratch with their hard copy. I appreciate the fact that plot.ly saves the data IF the learners hit save.
3. Once the plots are present, downloading or screen capturing the plots are easy and quick. My learners liked the ability to quickly make many different plots and then examine them and decide what they plots were really meaning. Changing colors, counts to percents, and other elements of the graphs was easy, fast and very user friendly.
All in all, not too many downsides from the learner’s perspective.
Here are my frustrations with the program as we have been using Plot.ly.
1. Many of my learners did surveys that had categorical Yes, No or Freshmen, Sophomore, Junior Senior responses. After compiling their survey responses, their data looked similar to this:
What kind of graph would you make with this type of data? You are correct a bar chart. Bar charts are for categorical data, histograms are for quantitative. So, I do that.
Not it. I struggle with this for three days, tweet them for help, more than 4 times, and nada. Zilch. Don’t hear anything back, and am ready to give up on Plot.ly. However, I notice they have a “contact us” in the lower right of the screen. I email them, and a very nice person responds the next day and with instructions on how to make a Histogram.
What. The. Hell. A histogram? I follow her instructions.
The y axis is in percent (which takes an extra step to get), clearly what I wanted, but it is not a histogram. I do have a problem with a “Statistics” program that calls a bar graph a histogram. The instructions to make it a stacked bar chart are easy to follow and find, I chose not to do it for this comparison.
2. Ordering the x-axis is a pain. Many of the questions my learners had dealt with the difference between freshmen, sophomores, juniors and seniors. Plot.ly ordered the x-axis based on the order of the data in the spreadsheet. Which means to reorder the x-axis we had to sort the data in excel, (but we don’t want them in alphabetical order?) and reupload.
Really Plot.ly? You don’t have a way to specify the order of the axis? I searched. Trust me I searched long and hard. I ended up just telling my learners to not worry about it.
As a comparison, this is what it looks like in JMP (version 8 is what I have).
After hitting “OK”, look at all the stats automatically generated! Also, in the contingency plot (made by default btw) the width is proportional to how many in that column, so the widths AND heights are informative unlike Plot.ly. JMP also automatically generates axis by percent, not counts.
I guess what I am saying is that clearly Plot.ly is not meant to be used as I am using it in class. There are easier, faster, and more statistically correct software to use. I will have to figure out what to use for next year because I am not completely sold on Plot.ly, but JMP has to be installed on computers. There are always gives and takes to every decision.
And right after I posted this, one of my learners walks in tearing her hair out. She has a mixture of categorical and quantitative data, and Plot.ly will not graph the categorical data at all for her. The menu options work completely different for her than for everyone else. She is installing JMP and getting it done that way. Sigh.