Yes, polls are reliable enough so long as they meet some criteria. For example, the people who were polled must be members of a fair and unbiased sample of the relevant population. Similarly, the poll should not have misleading questions that prejudice the people who were polled towards a particular response. Polls can be trusted so long as they meet these criteria, which is why pollsters reveal their polling methods.
If they ever had any credibility polls don't have any credibility anymore and are certainly not reliable enough to determine who will win an election. Just look at the failure of the last two main UK elections, the polls were just wrong in the 2015 General Election and the polls failed to predict the leave victory in the EU referendum. Why? Is it due to the increasingly scientific, but ever more depersonalised methodologies used? Or is it due to the fact that as voters are becoming more cynical about politicians and the political system in general and this is reflected in a couldn't care less attitude towards opinion polls?
No, polls are not reliable enough to determine who will win an election. This is because polls are not certain votes and are not factoring in the electoral votes, for example. While polls can give you a good idea of what is to come, it is not reliable enough to determine a win.
Everything depends upon the sample collected and the response received. Everyone may not respond to Cellphone surveys.Internet surveys are becoming more common, with the obvious problem that they can’t survey people who aren’t online.In election polls, pollsters talk to a few hundred or a couple thousand people out of millions of potential voters. This introduces “error” in the poll, which means you would probably get a slightly different answer if you talked to everyone rather than a smaller group.