Seems we have changed topic, so can I enter my bid:
I learned SPSS using punched cards:
http://en.wikipedia.org/wiki/Punched_cardDeliver a set of cards for batch processing, wait for 2 hours to get output, with only one single message: 'Fatal error in card 23, job aborted'
On topic:
In terms of reliability of the measurement it's always best to have many measurements, especially when there is a lot of variation in the object of measurement (drinking a glass of water increases 'weight' with 200 grams). My weight will vary 1 full kg. between different measurements. So for a reliable measurement measure often (daily), but take only the average of a number of measurements (say 7 or 10) to calculate your 'true weight'. Make sure that in your head (in terms of assessing your progress) you only calculate with the average.
So, when measuring daily, put on your NERDY glasses (or hat)
and say in a loud voice: 'This is not me, it's a scientist collecting one of a set of measurements to calculate my weight'. Enter the data in some kind of spreadsheet and remove them from your mind (should be easy, as this was only some nerdy scientist doing her job). Then look at the average number, compared to one or two weeks ago. Say in a loud voice: 'This is me, looking at my weight progress.
Make sure you use only one digit (it makes no sense to calculate more if one drink will have an effect of 0.2 kg.)
Happy measuring!