Many columns in the lesson

[brownsarahm]

There are a lot of columns in the data set, for this lesson, only a subset are used and the scrolling left and right to get to the right one often distracts learners; it's even slower for an instructor who has their screen zoomed in to be readable.

[brownsarahm]

We had the idea that maybe the beginning of the lesson works with a truncated version that is only the columns that we know have issues and in the script section, load the whole dataset and apply the script

[leegalletly]

I found the number of columns distracting as well. Although researchers will often be working with data that looks like this, for the purpose of the lesson, I think it would be easier to have a table that just exceeds the window, this way learners can see what it looks like to scroll but don’t have to navigate too far to find the column they’re working on.
I don’t feel that using the larger dataset during the script lesson would be quite as effective because you would still have to scroll through all the irrelevant columns that were unchanged. Using a smaller set for both sections would be easier for learners to immediately see changes.

[brownsarahm]

we thought of using the wider data set for the script to make it more clear that a script can apply to a different dataset. Definitely optional, but I think some form of a different dataset would reinforce the purpose of the scripts.

[murraycadzow]

+1 for this suggestion

[jevbelikov]

We could consider using a re-ordered data set with relevant columns at the front.
While I found scrolling distracting, using a larger dataset gives emphasises that we are working with real data.

[leannethelibrarian]

Just wanted to express my support for reducing the number of columns, I too found it distracting to scroll.

[bencomp]

It has been quite a while since you all discussed the large number of columns in the dataset for this OpenRefine lesson. I see several good suggestions in your comments:

• perform the cleaning steps on a smaller dataset and (optionally) show the power of OpenRefine's scripts by applying the steps on the large dataset
• don't remove columns, but put the relevant columns on the left to prevent having to scroll all the time

As I also find that the number of columns and the scrolling make it harder to give clear instructions, I would like to solve this issue too. Last time I taught this lesson, I added a short introduction of the column reordering/removing menu, as suggested in #83. We moved the relevant columns to the left and removed a few columns for the sake of trying, but I mentioned that I knew beforehand which columns were going to be relevant. Because of the number of columns, I would strongly recommend that learners get a list of relevant columns to move in the menu through a shared document. (#83 is still open, as there is no vetted exercise for the menu yet. Suggestions are very welcome!)

As a maintainer, I see that using the menu for column reordering/removal is the quickest way to resolving this issue (even though we'd need appropriate instructions). Please let us know if you feel otherwise! Without objections, we can close this issue by closing #83.

To change the dataset, i.e. have a smaller dataset or maybe one with reordered columns, we need an opinion from the curriculum advisory committee (CAC). This makes sense to me, as it was mentioned that this dataset represents real-world data, as it should. (There will be a new CAC for the social sciences curriculum in a few months. 🎉 )

A previous CAC already suggested to add more rows to the dataset, in #29. The CAC will hopefully also discuss adding more 'dirt' to the dataset, as I suggested in #108.

[brownsarahm]

I think this approach makes sense.