Ever since I started observing variable stars at the age of 11, I have used rough, quickly-drawn graphs as a way of getting a feel for the behaviour and consistency of numerical data. At the end of a long observing session I would often sit down beside the fire with a cup of cocoa and do quick free-hand sketches of the light-curves of the eclipsing binary stars that I had been following through the night. This enabled me to see if anything unexpected had occurred and, if necessary, to plan follow-up observations on later nights.
I now no longer observe variable stars (not since becoming a parent) but I occasionally come across data which leads me to plot a quick graph. This happened earlier this summer when I received my daughter Zoe's school report. I had got a vague feeling over the year that Zoe was occasionally feigning sickness, especially on Mondays. But, as I normally have to leave for work before Zoe goes for her school bus, I have always given her the benefit of the doubt. However, when I saw an attendance table in the back of her school report I immediately totted up the absences for each day of the week and plotted the results in the above graph. I was astounded: Zoe's sickness rate varied almost linearly from 15% on Mondays down to 0% on Fridays. I showed Zoe the graph and she too was quite surprised by it.
The most likely cause of this Monday morning effect is that Zoe was staying up too late at night on the weekends. The tiredness produced by this could persist through Tuesday and Wednesday to produce the observed linear decrease in 'sickness' throughout the week. Another possibility, that Zoe had more lessons she didn't like early in the week, can be discounted because it would be very unlikely that her lessons were so arranged as to produce the linear decrease. So, this year Zoe will be in bed by 10:30pm at weekends and getting up by 08:30am. I have stuck the above graph on the fridge door to remind me (and Zoe) and am looking forward to her next school report so I can plot this year's attendance data and see if the new regime has had the desired effect.