I’m using Network Canvas for a qualitative study and noticed that a long alter name won’t fit in the bubble (sorry, not sure of the correct term) it’s placed in.
For example, if someone writes Dr then that fits, however if they want to write paediatrician I end up with ‘aediatricia’ as the letters are cut off. This is then difficult to read on the sociogram.
Is there a way for me to make this bubble increase so that it fits the full word? I note that the interface scale can be increased to 130%, but I can’t see anything that allows me to increase the alter bubbles.
If there is no option for this I will suggest this change on the community forum. I did not think of mentioning it however recently another qualitative researcher shared the same issue.
This is a common issue that, frankly, we’ve struggled to come up with a good solution to! I would be really interested to hear if you (or others) have any ideas.
Unfortunately it isn’t as simple as increasing the size of the node circles, as this would reduce the informational density across the whole app. Situations where nodes can comfortably be larger (such as the name generator or sociogram) must be balanced against situations where they must be smaller (such as the ordinal bin interface, and name generator side panels). In those situations, the text would become too small to read.
These are some of the thoughts we have had, and the things we have tried:
We used to reduce the font size as the number of characters in the label increased. The thinking was that as text got longer we could squeeze more in by reducing text size. The problem was that unless we use fixed-width fonts (which tend to look computer-y and aren’t very readable) you can’t easily infer much about the amount of space a given string will take up from its length. Consider the different width between a lower case ‘i’ and an uppercase ‘B’.
Then there is the issue of line-breaks. It is simple to insert a line break wherever there is a space - but should you? Not if the label can fit on a single line, which depends on how large the font is (see above). Then, where do you break if there is no whitespace (for example in the string ‘paediatrician’)? Unfortunately, web browsers don’t have an answer to this problem either. In scenarios where you are authoring the content for a website, you can insert hidden characters that give browsers hints about where they should break text, and if they should hyphenate, but this obviously isn’t practical where a participant is entering the strings in an interview.
The choice of a circle for the node shape added additional complications to the above, because it meant that the maximum possible line length changes depending on where your line is vertically within the shape - and this itself varies based on the number of characters you managed to fit on the previous line, as well as how many are left!
We implemented a recursive function that did a reasonable job of “solving” for these problems, but it slowed the app down fairly considerably. The compromise we settled on was the “custom node label worker” functionality (https://documentation.networkcanvas.com/reference/node-labelling/#custom-node-labelling-advanced) which allows you to write a small function that can process the labels however you wish, including inserting spaces and linebreaks. Unfortunately it cannot alter the size of text. Nevertheless, if you can think of an algorithm you think could improve the situation in your protocol, I would be happy to help you to implement it using this functionality.
Another idea we have been thinking about for a while is changing the way node labels are rendered completely, and putting them either outside of the node shape entirely, or else revealing the full node when hovering or long pressing the node. Perhaps even offering alternative node shapes. Unfortunately we are no longer actively developing new features for Interviewer due to the end of our funding, so any implementation of this functionality would have to be done either by the community or through a contract to develop it specifically. Please let me know if this interests you.
Sorry to not be able to offer a simple solution, but I hope that at least helps to explain the problem.
Thank you for explaining this, it’s very clear. Ideally the text would overlap the node circles, with an option for the circles to be hidden or transparent on the sociogram. I realise this may not be feasible however.
The main value of Network Canvas for my PhD study is to present the sociogram and leave it visible to the participant whilst I ask questions. I think I will just ask participants to shorten or abbreviate alters.