Better example, if you photograph the cat sitting on a pile of coal in a snowy landscape:
With the 200mm, you might only frame the cat itself (presumably close to neutral grey in tone) and so get a "correct" exposure.
With the 50mm (if you are now taking in the whole scene), you might be getting more or less of the snow or coal in the image, and the camera's meter might be "fooled" by the lighter or darker objects one way or another.
But with both lenses, if you are careful to only meter the cat, no exposure compensation is needed (the meter should give the same results).
Also, if the zoom's aperture varies with its focal length, you need to take that into account if using a hand-held meter.
I think Rol_Lei Nut said it best. I would just add that if one is to trust the meter alone, then the "ideal cat" would be average/18%/Zone-V in reflectance. A black cat or white cat would have some of the same problems as would the coal or snow. IMO, any metering system requires some interpretation to arrive at the desired exposure for each imagemaking case. That comes with experience.
I have come to trust incident metering more than the other methods. Many, including Ansel Adams, prefer/preferred spot metering. What I trust the least are the "smart" metering algorithms which have minds of their own. I can't tell what they are thinking, and so I cannot as easily predict or compensate for when they get things wrong.
A good reflective/incident ambient/flash hand meter is a good addition to one's kit, in my opinion. I have been using the Sekonic L308B for a few years, and I like it very much. It runs on a single AA cell. If one is on a tight budget, a Kodak gray card can give "incident" type readings while using a reflective meter, hand-held or in-camera.
Also, there is a direct relationship between the reflectance of the palm or back of your hand and what a gray card will yield. One should know that difference. Then one always has a way to get a reference meter reading that can be trusted as a starting point. The back of my hand is one stop brighter than a gray card. Knowing this, I can read from it, open a stop, and know that I have an approximation of what an incident or gray card reading would have provided.