API is my preferred test kit, initially it was based on cost but I have adapted the instructions a bit and now get very consistent and accurate readings and completely trust them.
Firstly I don't use the test tube 5ml measure because I tried a few and they are not consistent. I scrounged a 5ml syringe from my doctor, they are manufactured to a accuracy standard and are nice and easy to read.
Secondly I store bottle #2 on the air pump - you know that nasty vibratey thing. Like Leeatl I hold it on with an elastic band, but I also shake the bottle as well just to be sure.
The final change is the really significant one, I ignore the stupid printed colour charts.
I got fed up of finding different colours on two different charts or the same colour for three ppm levels.
Initially I found a chart from their web site and stored it as a picture on a tablet, if you then hold the tube with a patch of clear screen lighting the tube from behind is takes the effect of ambient lighting out of the reading.
This was considerably better but I then went a further step and bought a Colorimeter - see I can spell colour when I want to (stupid Brit) from IORodeo Open Source Colorimeter Project ? IO Rodeo
This REALLY solves the problem for good, and works with most of the API test kits. All the thing does is to measure the amount of light that shines through the sample so it replaces the eye for the reading, nothing else. It has the colour absorption curve v ppm for the API tests built in and will repeatedly read retested tank samples to a couple of ppm and I really mean that. A new sample of water retested and giving a reading within 2 or 3 ppm - never mind the "Does this look like 40, 80 or 160ppm Hun?" Given there are still potential errors in the testing process this is a massive improvement and having seen so many consistent results I now completely trust the result.
I expect someone will want to tell me that we don't need this level of accuracy and for Nitrates its true, but we DO need a better accuracy than 'between 40 and 160ppm' and as it will also measure Ammonia and Nitrite with the same level of accuracy you will see problem coming from a country mile off.
Since starting to use this method I now test, then do a quick calculation, and know exactly how big the water change has to be to get to my target of 15ppm. I can also see if there was an increase in the amount of NO3 produced (or not consumed by the plants) over the week and this for me is another very good indicator of how the filtration and the tank is performing.
The kit cost about $85 but its the best fishy investment I have ever made - and I can still use the cheaper test kits. In the words of your famous spokes person, "Woohoo!"