Simple answer: There was already an enormous stock of data on punch cards.<p>So it's rather the other way round: While early computers, like the UNIVAC I, used tapes (in this case magnetic tape) in order to separate themselves from the IBM data empire, IBM was facing a hard competition for its conventional punch card appliances in the Bull Gamma 3, which lead to the massively successful IBM 1401. In this respect, the 1401 was just the next step after electromechanical punch card appliances. Compatibility to existing data stores was a major selling point for IBM computing equipment.
One partial answer might be due to utility billing. Back in the 70's our phone bills always came with a punch card with "Do not fold or staple" written on them. You had to include the punch card with your cheque when paying the bill.<p>The card was envelope size, so it was easy to mail out and easy for the bill payer to carry around. But a paper tape would be pretty awkward to handle in this fashion.
From what I remember, paper tape was a slower input device, while cards could be read much faster. To make a change, you had to repunch the whole tape, while for cards you could just punch one card.<p>Plus punch cards were much nicer to write notes on, and stick in your pocket... :)