The picture of C.D.N.-N.D.G. Mayor Sue Montgomery, the borough mayor of Montreal knitting her way through a council meeting has now gone viral. She knits in red when men speak and in green when women speak. Her knitting is tapered and starts narrow but gets quite wide. I’ve read since her initial tweet that the ratio is around 80:20 men speaking, to women speaking.
As a knitter and a social researcher, this got me thinking. The first thing that occurred to me was that I had to try it myself. I am already a local politics enthusiast and so knitting my way through a Full Council meeting sounded like a brilliant use of my time.
I decided to even cast on by gender colour. Bristol City Council had/has a woman Lord Mayor and the casting on began with pink. I chose stereotypical colours to make the choices immediately visible.
Within a few rows, I encountered a methodological problem. Switching between colours after knitting for a while, meant that yarn was crossing the material all over the place; also, garter stitch meant that the previous row’s colour [once a change occurred] showed up on the next row.
This impacted the design of the project in three ways:
- The colours were not isolated so it was difficult to determine who spoke when, in small clusters of stitches. People would have to speak for a long time for it to be visible.
- The material is very ‘messy’ and making it into any kind of blanket or shawl would be almost impossible.
- Yarn would constantly have to be cut or moved around, and it was slowing down the knitting and wasting yarn.
I realised that knitting the colours together would not be a useful method. I suspect that the mayor was turning a row every time a new speaker began, or a different gendered/sex speaker began talking. However, this limits the use of using knitting to measure differences.
Method 1: Fail
I had also decided to make the knitting into squares so the narrow and wide parts of the shawl did not give misleading impressions on length of time spoken. This is one of the criticisms of pie charts — because of their narrow points and wide edges, it’s difficult to measure differences between categories. Bar graphs from the same axis are the best way to compare, in some situations.
Squares would also mean I could make a blanket at the end of say, a year’s worth of meetings.
An alternative I would like to also do is use different colours for different political parties, but one weakness to that method would be my lack of knowledge as to who is in which party. I recognise some politicians but not all. Would the Lord Mayor/chair have a different colour? Non-politician / city council people speaking, such as at public forum?
For now, gender [cis/trans] seems the easiest way to practice this.
I decided to try a new method. I would have two sets of needles (same size – 4.5mm) , two balls of yarn, and I would knit two separate squares. I would pick up either the male or female knitting set each time someone spoke.
One limitation to this might be that it could be cumbersome to have both sets at full council, in the chamber; or at home, next to wherever I was watching the council meeting.
One slip knot was tied on each needle to begin.
Each square would be 20 sts long and therefore the number of sts could be counted and measured at the end of the year.
This approach worked much better.
The red yarn was used whenever a woman was speaking; the yellow yarn was used whenever a man was speaking.
The evident, as in ‘visible’ conclusion is that women spoke more than men. Which seems perfectly acceptable.
However, another methodological problem appeared. The two yarns were slightly different in weight and texture. The red yarn has tighter rows with few gaps in between. If these two square-ish materials were to be measured in inches or centimetres, the result could be misleading.
Also, when the speakers finish in the middle of a row, I have so far knit to the end before turning and casting off. This is for practical purposes in order to make sure all the squares/rectangles can be joined at the end. I feel it
Method 2: almost successful
Same size needles; same size and weight and type of yarn; CO as the speaker speaks, two separate knitting projects; Bind off; join.
Once all the squares for the year are knit up — and this can be done in retrospect too since the full council webcasts are available — I intend to conduct chi-square tests for how many stitches are expected according to the gender balance in the chamber, and whether reality signficantly differs from expectation.
Will we be able to reject the null hypothesis of: the length of time people speak in the chamber is unrelated to gender?
Method 3 seems to account for most issues with the research method. Note the difference in clarity between knitting both colours at the same time and separately [see image].
Both are approximately the same length of time.
Some further thoughts:
Should every speaker be counted to ascertain the numbers for the chi square test, or is it sufficient to note the ratio of attendees at the meeting? For example, there were a number of councillors at Full Council on 21 May but not all of them spoke.
Full Council: 19 March 2019 [3:54:57]
Full Council: 21 May 2019 [1:48:37 https://www.youtube.com/watch?v=Rr18SL88dxI&feature=youtu.be]
List of attendees.
Conclusion from meeting 21/05/19
Of those expected and present, there were 50 men and 45 women, which means there were 52.6% men in the chamber: Of speaking time, men spoke for 40.9% of the time.
Data: 50 men (52.6%); 45 women.
men 18×20 =360 sts (40.9%) ;women= 26×20 =520 sts;
The chi-square statistic is 46.25, which gives a statistically significant result. More women spoke than expected.
— TBC —Tweet