The method was a good way of carrying out the experiment as it was quite easy to conduct once the circuit was connected, the results seem to be fairly accurate and the obtaining of results was systematic. However it would have been better to use two people conducting the experiment as one would to connect point A to point B while the other partner would start the stopwatch. Also the experiment took extremely long due to the high resistances in some cases like the 680kΩ experiment. Instead we should have used lower resistances, this would also have made the readings more accurate. Also to ease the process of connecting point A to point B we could have used a switch making a short circuit and looping the battery or something along those lines (see below).
In my opinion the results are accurate enough to establish a trend as the graph shows a definite straight line which proves the prediction that the resistance is directly proportionate to the time the capacitor requires to discharge. The results were obtained using an accurate stopwatch, a digital voltmeter and the resistors were accurate to 2%. Ofcourse the results aren't perfect, as the results on the graph would fall exactly on the graph-line if they were, but it is nearly impossible to achieve that.
There was only one result, which did no fit in with the main pattern, it was the value at 680kΩ. This may be due to many reasons. Firstly because the resistor value is so high even at a 2% error margin the resistor could be up to 13.6kΩ different from the supposed value. Also the time lapse between connecting A to B and starting the stopwatch may have accounted for the discrepancy.