In most FPTP constituencies, more people will have voted against, than for, the MP, so not much link for constituents there! Even in a very safe seat, about 40% of the votes will have been against the MP.
With STV in two member constituencies (STV2), which is a little proportional, at least 67% of the final votes will elect an MP, so at least 67% of the voters can identify, and feel a link, with an MP.
With STV3, which is more proportional than STV2, at least 75% of the final votes will elect an MP.
With STV4, which is more proportional than STV3, at least 80% of the final votes will elect an MP.
With STV5, which is more proportional than STV4, at least 83% of the final votes will elect an MP.
With STV6, which is more proportional than STV5, at least 86% of the final votes will elect an MP.
With STV7, which is more proportional than STV6, at least 88% of the final votes will elect an MP.
With STV8, which is more proportional than STV7, at least 89% of the final votes will elect an MP.
The more MPs per STV constituency, the more proportional the result will be, the more people will feel a link with an MP they have helped to elect and the fewer wasted votes there will be, but the figures above show that the law of diminishing returns applies to the proportion of effective votes, and very many MPs per constituency may lead to complaints about the number of candidates and the length of the ballot paper.
Conventional wisdom is that STV5 is about the right balance, and I think STV7 should be regarded as the maximum.