Sammy Morse has a useful overview of the constituency (and all of the others) on his web site. For detailed election results, Nicholas Whyte has two separate web pages for Upper Bann, one covering the period 1983-1992, and the other covering 1996-2007. Needless to say, the source data on elections is all available on the web site of the Electoral Office.
There are several caveats that must be made when looking at Westminster elections.
Firstly, elections use the semi-democratic 'first past the post' system in a single member constituency. This has the effect of discouraging both minor candidates, and even minority party voters. If you know that you have no chance of being elected, why stand, and why vote? So the results are not going to give as accurate a picture of real political sympathies as other systems (single transferrable vote, or multi-member constituencies, for example).
Secondly, for the reason above, they are difficult to combine with other types of elections in a series. Although the same constituency is used for elections to the Assembly (using the single transferrable vote system in multi-member constituencies), the outcome of simultaneous elections to the two bodies may be different.
Having said that, though, voters in Northern Ireland are a hardly bunch and will come out to vote even when they know that their chosen candidate has no chance of winning. So although the graphs below mix the results of the two types of elections, it is hard to tell them apart.
Firstly, the percentage of the vote received by unionist, nationalist and other candidates:
Two observations need to be made at this point:
If a trend is sought, it should ignore both 1986 and 1990. Such a trend shows an increase in the nationalist share of the vote, and a decline in the unionist share, though for unionism the last 20 years have been a period of stagnation. Purely statistically, the linear trendlines that can be applied to the graph above show a tipping of the constituency (from unionist majority to nationalist majority) in approximately 12-15 years. We will see below how this is very unlikely though.
Secondly, the percentage of the electorate voting for unionist, nationalist and other candidates:
The same comments apply to this graph as above. However, even taking them into account, there is a visible downward drift in the unionist vote as a share of the electorate.
The statistical trendlines mentioned above are also evident when you examine the religious demography of the constituency. The proportion of Protestants in the overall population at each age is constantly going down. Amongst the elderly it exceeds 80%, but at age 20 it tips, and from then on there are more Catholics than Protestants at each age. Age 20 in 2001 represents age 27 now, of course, and many of the extreme elderly in 2001 are, sadly, no longer with us. The sudden steep rise in the proportion, and numbers, of those declaring 'no religion' is intriguing. Since the increase is amongst babies and toddlers, it is unlikely that they are committed atheists (yet). Nor, clearly, are their parents, since the graph shows no significant numbers of the irreligious around age 30.
So, these babies are being declared as 'no religion' by parents who did declare a religion for themselves. Either the kids are the result of mixed marriages where no decision has yet been taken about their religious affiliation, or there are a lot of parents who feel it is presumptuous to decide on a child's religion, before the child even knows what religion is. Since the 2001 census was the first one for which the particular questions were used, we have no previous example to refer to, and so we cannot yet tell what will happen to these kids in terms of religious or political affiliation. It seems likely, though, looking at the trends of the Protestant and Catholic percentages, that more of these unaffiliated children come from Catholic backgrounds than from Protestant backgrounds. We may have to wait for results of the next census (in 2011) to know more.
The graph below, showing the numbers at each age, shows how the mass of Protestants in the electorate will continue to exceed the mass of Catholics for some time. Amongst the young the numerical advantage to Catholicism is quite small, while amongst the middle-aged, Protestants enjoy a large surplus. The statistical trendline mentioned above appears to be incorrect – on the basis of the demographic picture, it will be several decades before Catholics outnumber Protestants in the electorate.
In 2001 the proportion of the electorate of Upper Bann that was Catholic was 40.5%, while 57.8% was Protestant, and 1.7% was 'other' or none.
In the 2001 Westminster election the combined proportion of the nationalist candidates was 36.0%, and the combined proportion of the unionist candidates was 63.0% ('others' got 1%). So nationalists received 4.5% less of the vote than might be expected, while unionists received 5.2% more.
Both explanations are possible. The graph above that shows the percentage of the electorate voting unionist and nationalist shows that, in 2001, unionism scored a higher proportion than its trend, while nationalism scored lower than its trend. This could be a result of the 'first past the post' discouragement referred to above. On the other hand, in 2001, the first Westminster election following the Good Friday Agreement, the DUP was trying hard to unseat David Trimble, then leader of the UUP and MP for Upper Bann. Some nationalists may have voted tactically for Trimble to avoid a DUP victory. The same thing probably happened in 2005, though this time Trimble lost and the DUP won.
An alternative explanation is suggested by Sammy Morse: "… Possibly because of the large Catholic police vote in Banbridge …" – he assumes Catholic policemen vote unionist, or do not vote.
In other elections in Upper Bann, before and after 2001, the unionist proportion of the vote was closer to the Protestant share of the electorate: 1996 – 57.6%, 1997 – 56.0%, 1998 – 57.2%, 2007 – 56.5%. The nationalist share never exceeded 38.9%, and is usually a point or two lower. It may be the case that most votes for the Workers Party and the Alliance Party in Upper Bann are Catholic.