I've seen 4-5 different respectable sites test the display. But dismissing there battery test based on a display calibration that's not identical is just silly. Oh and I'm in no way associated with Notebookcheck. Most likely what's at play here, is different usage patterns, awareness of problems, and computer knowledge, more than hardware that varies. If you are referring to the hardware performing different on identical setups, then even Poul Thurrot questions that. Lastly I just want to touch on the statement that people are experiencing wildly different results. Twins are great of testing, as you exclude as many variables as possible. If you want to use a people analogy, it kind of represent identical twins. So the average usage pattern for the i7 could be wildly different than the i5's.Īnd thats why a perfectly controlled sample of 1 of each device gives a lot better results when comparing the battery life of the difference CPUs. Further more you would be a large set of samples with every type of CPU as you were interested in the difference between the i5 and i7.Īnother thing is even the average battery life of 1000 i5 and i7 devices would be hard to compare, as people who opt for the i7 typical have a different usage pattern than people who opt for the i5. But 30 samples is so far from doing that, especially on a site like Reddit where people tends to tinker with there devices and setup, have tons of little programs installed that could run in the background etc. In other words its not only the CPU type that affects the battery life.Īnd you are right, once the selection gets large enough you start to gain a confident result with respects to the average battery life. I bet you there are 30 different use cases and setups in play. The problem with comparing results from 30 people is an uncontrolled environment. In this case you are comparing three different devices, and by using the exact same testing criteria and controlling the variable that could affect battery life other than the CPU (i.e. When doing any kind of comparison the most important thing is a controlled environment and knowledge of your uncontrolled variables. That's not to brag, just to point out that I have extensive knowledge about testing. I've done a lot of lab work, and have a background in science and engineering. It wasn't designed to get me insulted by random people on the internet. I only posted this reddit to generate Information to help me and others with their purchase. I will not be responding to you any longer, because on reflection your last statement is the perfect troll statement that appears to have done its job.
Another Tech site that I trust greatly, Anandtech, came up with wildly different results for their battery life that I could ever achieve with a fresh installation of windows and doing exactly what they were doing.īut should I believe an internet site above and beyond my own eyes, experiences and observations? erm, sorry do you work for notebookcheck? I do not know anything about their qualifications. Yes, everyone's use will be wildly different, but once the selection gets large enough you start seeing trends and start getting statical Information on the Deviation from the trend, and can start weighing the confidence level of the statistics generated.Īnd as for me 'doubting their qualifications'. So again, one sample is not enough.ģ0 samples or more from a general selection like this hold much more value.
1 sample is not big enough in any test on any product in any place on this planet (outside of North Korea and maybe Tobacco companies).Īs for the surface pro, people are having wildly different results on the exact same model with the same software installed. We could pick a baby and say that Norway's Population is all under 1 year old.
We could also pick driver driving a truck, and say that 100% of all drivers in the USA drive a truck. If it had significance, we could pick a 90 year old smoker and say that smoking does not cause cancer. Firstly there is no need to be rude.Ī sample pool of 1 has 0% worth in the field of statistics and science.