Should we strip out the undecideds when reporting the findings of opinion polls? And how far should we bore into the figures beyond a simple yes/no answer?
Does how the questions are asked skew the results?
Those were hot topics on Belfast Telegraph's DebateNI web area and we have also had phone calls and emails about them.
Often a high proportion of people don't register an opinion. Forty-four per cent said they wouldn't vote, pretty close to the actual figure in elections and on today's policing question about half of people either didn't have an opinion or were neutral.
Some polling companies encourage a response to see if there is an underlying preference.
For instance, in the Life and Times survey, conducted by the two universities, only 8% of people weren't counted as supporters of a political party, only about a fifth of the number who don't vote. The pollsters there wanted to get at underlying feeling.
The result was to record higher support for the SDLP (13%) than Sinn Fein (12%) and to find that the DUP was supported by 17% of people, nothing like the 30% who voted for it in 2011.
That wouldn't be much use for predicting election results; it may flow from reducing the number of undecideds.
We agreed with LucidTalk not to encourage 'leaners' to commit to an opinion if they don't initially volunteer one.
That got us close to the actual non-voting figures and, we believe, gave a realistic indication of the extent of both committed opinion and apathy or indecision.
In something like a referendum on Irish unity or an election to the Assembly the 'don't knows' generally wouldn't vote.
So, it makes sense to remove them from the equation and work our percentages for those who have made their minds up. Yet, when we give the full poll findings to political parties and other bodies who have asked to see them they will also examine the undecideds so we give both sets of figures.
These are the people it would pay to listen to and win over if they want to build popular support.
That holds true for political parties, police services and ministers.
So, we also asked LucidTalk to analyse the opinions of non-voters to see if they differed much from voters and we published a summary of the findings on Tuesday.
It wasn't quite a chalk and cheese division but non-voters were more likely to be younger or women.
Just over half of the under 25- year-olds and of the female population said they wouldn't vote.
Unlike most societies, more than half of the affluent upper middle classes, the ABs, also said they didn't intend to vote.
Very broadly speaking, non-voters tended to be more liberal on issues like same-sex marriage and slightly less likely to identity as Protestants and Catholics, though they are a million miles away from the voting population.
This shows us where the party political system may be losing touch with sections of the electorate in Northern Ireland. It points to policies that need tweaked and it can also point to geographical areas of weakness.
In countries like America, where political parties have more resources, there is regular private polling to try to increase political reach.
It happens here, too, but it is smaller scale.
Then there is the approval rating.
That strips out people who express a neutral view on a scale of one to five as well as the 'don't knows'.
The rating then measures the feeling of those who gave a definite view.
Approval ratings are fairly crude but in bigger places like England and the US they are taken every few days to spot trends in the comparative performance of political leaders, and governments over time and with different target audiences.
Some companies also do it for marketing purposes.
All these measures are valid in different ways, as is the Life and Times methodology.
This week we have tried to use as many as we could to get the fullest possible picture of opinion here.
In a few days LucidTalk and the Belfast Telegraph will make the full data tables available to let you, the public, see if there is anything we have missed.