Do Dxo marks really have an impact on how good an image is?

I'm stuck between the Canon 100D and the Nikon D3300, and here is what Dxo says:

Nikon: 82
Canon: 63

Is this b/c of the 18mp vs the 24mp, because might print A4 or A3 size, max so will there be a difference? Also what one has the better kit lens? So what one can I take out of the box and get the better image ( I know it depends on the user, but still)?

If you intend using your camera to photograph 2 dimensional black and white matrices and patterns then by all means Dxo marks show the way.
In the real world of photography it is a load of pretentious bow locks.
If I were to show you 2 identical images taken with those 2 cameras with equal quality lenses you and 99.9% of the population will not distinguish between them.

DXO is mainly focusing on the performance of the sensor. So only worry if shooting a black cat in a coal cellar is something you need to photograph often. We could print A4 and even A3 with from properly-lit images from DSLR cameras that had only 6mp sensors. If you want to print A3 often then get a camera with 35mm sensor because if you crop from smaller sensor then there are problems at that size.

Kit lenses are similar as long as both have VR image stabilisation (you don't have to worry with Pentax or Sony as the VR is in the camera sensor mechanism itself)

'Out of the box' image characteristics depend on the colour saturatation, white balance and sharpening settings which you have set up in your camera

Does composing on the LCD panel interest you? If so then live view on the Canon can simulate the effects of settings changes. Nikon doesn't do this but in real life is so accurate at exposure that it is not likely to get it wrong very often. In pre-live view days we would take sthe shot, check it and adjust the blinking areas with settings changes so that the histogram came back from the edges. Then the next shot would be perfect if metred in the right place.

Handle both cameras and pick the one you like best. You don't want to end up in 5 years time having to sell all your accessories because you have decided to jump ship to a different and incompatible brand.

Dxomarks measures sensor performance in a controlled way. All are treated the same. So, yes, the sensor of the Nikon D3300 does measure better. Read their testing to see why it is better.

The MP has a little to do with it, mainly in resolution.

Both can print on almost any size of paper.

The kit lenses on both cameras are quite similar. You can read what photozone.de has to say about them.

Both will give you great service, right out of the box, after you charge the battery.

To decide between the two, try to get your hands on both. Lacking that, see what the photographer you admire uses and go with that.

Photographers are always looking for an edge.

This is why many magazine photographers shoot using 35 mm Kodachrome film before digital cameras were available because it had the highest resolution and best colour of the other transparency films of the time

This is why many pros use Nikon D3s and D810 cameras if they shoot any sports that is indoors or at night. Both cameras sensors perform well in low light and as a plus, they also have good colour depth and dynamic range.

I recently purchased a Nikon D800E and it has one of the highest performing sensors and it shows, especially when shooting in situations where the dynamic range of the shot is rather large.

I have linked you to an image that shows the kind of scene that is best shot using such a camera.

Generally, no. I own an Olympus OMD-EM5 which has a higher rating then my Canon Rebel T3i. The biggest difference that is noticeable is the dynamic range, the EM5 is better, however you can adjust the Canon when shooting raw in Lightroom to make-up for this. I think it all comes down to experience and what you like to shoot. Canon cameras have lower ratings but they are still great cameras for any professional. As others have stated DXOMark uses very restrictive and specific testing which does not always correspond well to the real world. They also do not take account for the type of lens being used which can make a big difference, as the ratings are sensor based for cameras.