New: Counting Christians in the Australian census

Introduction

In this article I look at how the proportion of active Christians varies across geography and between demographic sectors, mostly using 2016 ABS census data. My intended readers are thinkers and leaders in the Christian community (who are OK with an overload of statistics).

In 2012 I wrote an article on this topic using 2011 census data. I planned to rewrite it after the 2016 data release, but it's taken me over two years to get to it. In contrast, Aaron MacAleese wrote his Census 2016 Religion Summary the same day! He followed with Explaining declining Christianity in Australia a week later. These are worth a read.

This article is viewable both on this site and on Mappage, my webapp for looking at statistics. If you're reading this on a large screen, it's best to read this article in Mappage, where I have set it up with lots of maps and charts. If you read this version, you will see references to the charts you can't see here.

This article is viewable both on this page and in plain form on the GeoMiss site. The one-letter links here will show maps and charts. There are lots of these links, so you need only click the ones that interest you. Most of what I thought important is in the text here. This is not so much an article with a Google Map embedded; it is my statistics & mapping webapp (called Mappage) which I have set up to accommodate articles like this. Clicking the links just changes the settings on the right. You're not restricted to looking at what the links provide; you can look around for yourself at any other stats. I've spent many hours looking at stats in Mappage in compiling this article.

Data sources

There are a few different groups we can measure to gauge the Christian-ness of the population and each has different statistics available.

Average weekly attendance is something easy to measure in individual congregations by a head count. Some denominations ask their churches for attendance numbers, and sometimes denominations may publish totals. Total weekly attendance in Australia was estimated at about 7% in 2011 and has been slowly dropping all this century. (Here's a 2019 article gathering some denominational stats together). The National Church Life Survey is a large (so hopefully representative) subset of weekly attenders, so any demographics measured in the NCLS can be compared against the general population.

Regular church attendance, ie monthly or more often, is measured surveys (eg the International Social Survey Program) and tends to be nearly twice the weekly number. For any other attributes measured by the survey, it may be possible to to compare the profile of attenders with the rest of the population. Publications by the Christian Research Association have done a lot of this. Some churches or denominations may also record the number of regular attenders they have.

Finally, the census measures religious affiliation. Christians by this measure were just over half the population, which is over three times the number of regular attenders. In many cases, the demographics of census Christians are in between that of church attenders and that of the general population, eg census Christians have an old age profile while church attenders have an even older age profile.

What I'm looking at

I am most interested in active Christians - people following Jesus Christ - and I want to know which parts of Australia, by geography or any demographic division, have more or less of us. Whom we count as Christians is obviously a big quiestion, but I'm considering regular or weekly church attendance to be the best measurables we might obtain to indicate where active Christians are. However, the census is the most convenient data source, so much so that I've built Mappage largely to analyse and display census data, and this article mostly looks at census data. Because I'm mostly using a less-relevant source, I think what NCLS has written from 2016 is more significant than anything I'm going to add here.

The ideal for me would be church attendance numbers for geographical areas, but I don't have much of that - the Catholic National Count of Attendance has diocese-level stats, and I have some older data at individual church level for some denominations in SA.

Many of the charts shown here combine religion with other variables such as age, language spoken and geographical region. There are some other variables I couldn't combine directly with religion1Either because I've only using the free version of TableBuilder or because that variable relates not to a person but a household or a region., but I can create a scatter plot that links them indirectly.

The totals

P m8tpd4vi The total Christian figure was down from 58.3% to 52.2%, so this might be our last census with a Christian majority. Hinduism and Islam had increases due to immigration and children born to immigrants. No Religion jumped from 22.5% to 30.2%.

The Not Stated figure of 9.2% includes the 5.2% who did not do the census at all. In 2006 and earlier, Not Stated was higher. I considered it a soft form of No Religion, as the two were highest and lowest in the same areas. If I was being thorough I would remove the 5.2%.

P e8qu0514 Breaking down the Christians into denominations, Catholic and Anglican are the largest, but both are much smaller than No Religion. You can click "Residence" on the right, then click on a state to update the pie to that state's religion breakdown. You can also click "Year" and 2011 to see data from that census.

Christian nfd (not further defined) refers to the 2.6% (up from 2.2%) who wrote "Christian".2Many Christians - most evangelicals - think of themselves simply as Christian rather than their denomination and this often feeds through to the census response. The question said "What is the person's religion?" (it doesn't ask for denomination, but implies it by listing denominations as options). This 2.6% would include many from independent churches and denominations that didn't have their own box, particularly pentecostals (there will be a Pentecostal box next census3By virtue of being in the top 10 responses. Had there been a box before, Pentecostal would probably have been in the top 10 in 2006 or 2011.).

McCrindle's 2017 survey on Faith and Belief in Australia also asked about religion identification and returned results of 45% Christianity, 9% other religions, 32% no religion and 14% spiritual but not religious.

Religious groups across geography

Looking at how the census Christian proportion varies across different SA4s4Statistical Area level 4 (SA4) is the next size down from a whole state., C ep6r1vul the highest levels are in rural NSW, some places in Queensland and some outer parts of Sydney including Sutherland Shire. The lowest levels are in the inner parts of Sydney and Melbourne.

The columns are coloured by state. You can mouse over the columns to see what areas they represent. You can click "Residence" then click on a state/territory to break it down and see all the SA3s there. You can click "Vis/algo" then "Map %" to show the same data on a map.

Non-Christian religions C 0bh1e27x account for 8% (up from 7%), spread unevenly, with Parramatta (31%) and three other Sydney areas over 20% while most rural SA4s were below 3%.

The No Religion scores C 8v2g3zbq are highest in inner urban areas, with Melbourne Inner top on 43.4%. Also high are Adelaide South, Hobart and Mornington Peninsula5If we drilled down to smaller areas we'd see lots of "sea/tree change" areas high in No Religion.. Lowest are the multicultural parts of Sydney noted for non-Christian religions, as well as Blacktown. No Religion had its biggest increases in places where it was already higher.

Note that you can click "Religion" and click any of the other religions/denominations in the list to see their distribution.

O v0du1zmh This scatter plot shows both the Christian and other religions percentage of each SA4, meaning the most non-religious are in the bottom-left corner.

Different Christian blocs across geography

To simplify my analysis I sometimes break the Christian denominations into three blocs (I'm leaving out the "marginal Christian" groups). C uh62rnl5 Firstly, the Catholic and Orthodox numbers are highest in Sydney South West and Melbourne North West, which both have many people with southern European or Middle Eastern heritage. They are lowest in Tasmania.

The protestant denominations I divide into two blocs, Old and New. Old Prot (NCLS calls them Mainstream Protestant) refers to Anglican (13.4%), Uniting (3.7%), Presbyterian & Reformed (2.3%) and Lutheran (0.7%). Similar to Catholic & Orthodox, the ratio of weekly church attenders to census figures is below 20% (for Anglicans, around 4%). Together they historically had a large share of the population but that has shrunk a lot over the last 50 years. This decline in affiliation probably lags the decline in regular attendance in the churches by some decades. All of them had a box to tick in the census6Lutherans dropped out of the top 10, so they won't have a box next time..

C kn06hwsa Old Prot numbers are highest in rural NSW and much of Queensland and lowest in most of Melbourne and some parts of Sydney.

New Prot is the rest of the protestants: Christian nfd (2.6%, and these would not all be protestant), Baptist (1.5%), Pentecostal (1.1%) and some smaller categories. None of these except Baptist had their own box in this census.

The New Prot bloc adds up to only 6.2% of the population and accounts for significant slice of active Christians and not many nominals, so where this is high, there are probably a lot of active Christians. Thus the census lets us get a good look at where about a third of Aussie church attenders are, while the others are hidden among the nominals.

C ozji2qhc New Prot is highest in Toowoomba (10.9%), Logan-Beaudesert and other SA4s around the outside of Brisbane and some outback SA4s. Melbourne Outer East towers above the rest of Melbourne. Outer surburban SA4s are high, inner urban ones are the lowest.

Another thing I've done is weight each denomination by their attendance ratio, with some adjustments for Christian nfd. This can give us a vague estimate of weekly church attendance for an area. For now I'll call this Sum-product, and it scores 7.5% (around 8% in some contexts7When we combine with some variables such as birthplace or language, it's normal to exclude Not Stated from totals, which excludes census non-respondents and so boosts the percentage of everything else., and should really be adjusted down to average 6-7%). I've also added a protestant-only version of Sum-product.

C qrmc0621 At SA4 level, Sum-product correlates strongly with New Prot, with Toowoomba winning by even more. At SA3 in each state level this high correlation is maintained in every state except SA, where the Catholic & Orthodox population is unevenly distributed. The changes in Sum-product since 2011 mostly reflect changes in New Prot. The biggest increase was Sydney South West and the biggest falls were Ballarat, Bendigo and parts of Brisbane.

C f1r8kyg9 This chart shows NSW at SA3 level, ordered by distance from central Sydney. Sum-product is lowest in inner Sydney, and highest in the outer suburbs, with the rest NSW above average. You can change to other states or other religion categories and see how the numbers vary by distance from the CBD of the state capital. (Changing Vis/Algo to Map % also shows this).

Age, language, gender

We can also look at religion stats broken down by other demographic categories. Firstly, we can see C 7my60g8v the Christian proportion of each age group. The over 70s are over 70% while ages 20-34 are under 40%. C pc7a6ibd The Sum-product is not quite so weighted to older ages. You can select any religion/denomination/bloc to see what proportion of the country it has in each age group. The age profile of church attenders measured by NCLS is even more weighted towards the older age groups.

Looking at the age profile of all religions, S lrkda4p9 we see that while the oldest age groups are the most Christian, young adults and young children have the most in other religions and no religion.

If we break down the population by language8Language spoken at home, which gives priority to any language other than English. (which is my preferred variable for breaking down into ethnic groups), S qmnypjo9 we see nearly all the English-only majority identifying as Christian or no religion, with only 2.3% for all other religions. S 6tnw8hso Looking at the other languages, there are some that are mostly of one religion and some that are mixed. The Sum-product C q7x68sik shows other language speakers at the same levels as English-only. Excluding English-only C ksi4x5nf (there's a lot in this chart, so note that you can put the mouse in the margin to zoom vertically) high scores include African, Pacific Islander, Aboriginal and Burmese languages as well as Tagalog, Romanian, Armenian and Aramaic9Could be wrong - I really don't know what ratio to use for the Assyrian church and Malayalam. Many European ethnicities are mostly Christian in the census, but they would be largely nominal. Some Asian groups have a large proportion of active Christians. NCLS reports that 27% of church attenders were born in a non-English-speaking country, compared to 22% of Australians. Chinese, Korean and Vietnamese all saw No Religion increase by over 10%10Much of this increase reflects recent immigrants, but there was also there is also a shift in the existing population from Buddhist and Christian to No Religion..

54.7% of females were census Christians compared with 49.5% of males, while the males were more often non-religious. Put the other way, 53.1% of census Christians were female (54% of those aged 15+), while only 47.3% of No Religion were female. Sum-product returns 53.2% female (54.1% for 15+) . NCLS reports that Australian adult church attenders were 60% female11Part of this imbalance is due to the old age profile and women living longer, but women outnumber men in church in every age group. I would also guess that another part of the imbalance is that women are more likely to attend more often, so that the gender ratio of active Christians is a bit more balanced than the 60% suggests..

Education, SEIFA

Education level is an interesting one. I can't compare religion and education directly12The free version of TableBuilder restricts what combinations of variables I can use., but O faq3i87w comparing the proportion of census Christians per SA4 with the proportion with university degrees shows a strong negative correlation. This page (ABS) using 2011 data confirms that degree holders are less likely to be Christian than the less educated. This would be partly mitigated if we broke it down by age groups. This chart O qzomsfe1 shows the proprtion of degree holders and Christians in each age group.

O jg23iqr4 The Sum-product measure of Christian-ness across SA4s also correlates negatively with degree holders even though Sum-product is less skewed by age.

However, NCLS reports that 37% of church attenders had a uni degree, compared to 27% of the general population (despite the more Christian age groups being less educated - if we broke that down by age group, it would be even more stark). Putting this together, we find that the less educated are more likely to be nominal Christians while the more educated are more likely to be active Christians or No Religion. This confounds any attempt to estimate relative Christian-ness using census data.

O fihbt09y Comparing the Christian percentage with the SEIFA13SEIFA = Socio-economic indices for areas. Four different indices, each a combination of several variables. Pertains to an area rather than indiviuals. Index of Advantage and Disadvantage for each SA4, the correlation is -46% - census Christians clearly tend not to live in rich areas. If we change to the Index of Disadvantage (which isn't that different), this moderates to -33%, reflecting higher Chrstian numbers in low-wealth rural areas than rough metropolitan ones.

The Sum-product measure correlates even more strongly with both indices. But again, if we looked at active Christians, we might see a different picture, particularly with so many Anglican church attenders (who hardly show up in Sum-product) living in rich areas.

More thoughts on Sum-product

An area with a lot of active Christians in the Old Prot or Catholic/Orthodox blocs will be largely hidden, while a smaller number in the New Prot bloc will show through. Using Sum-product guesses that the active Christians and nominals will mix uniformly. If instead you guess that non-church-attending Anglo-Aussies will be uniformly divided between No Religion and nominal Christians, the outer suburbs will prove to be more Christian and the inner cities more secular. Then again, New Prot accounts for a bigger share of the church attendance pie in outer metro than they are in inner...

The Sum-product across geography could be refined if we had our denomination attendance stats broken down further. I have the attendance for each Catholic diocese and I could use the diocese-level attendance/census ratio rather than using 11.8% across the board. I did this with the Anglican figures last time.

The ratios for the Catholic dioceses range from 15.7% for Sydney to 6.7% for Hobart, with metro areas higher and Queensland lower. The ratios correlated strongly with the proportion of non-English speakers among the Catholics of each diocese. Adjusting the Sum-product according to this would increase it by up to nearly 1% in parts of Sydney and decrease it by nearly 1% in much of Queensland.

The ratios for Anglican dioceses in 2001 ranged from 6% for Sydney to 3% for Tasmania and Brisbane. Does this suggest that the same would hold for other denominations - could we assume that the Uniting and Presbyterian (or even Baptist and Christian nfd) figures each contain a higher proprtion of nominals in Queensland and Tasmania than in Sydney?

State-level attendance numbers for other denominations would help paint the picture, but we have already seen where the breakdown of active Christian differs from the nominals - by language, age, education level and other factors. I could tweak my multipliers in more detail14Philip Hughes pointed this out in an email after I wrote the 2012 version of this., but I would mostly be illustrating what is already known.

As one example, the Sum-product for Port Augusta was 5.8%, after some adjustments it would be below 5%, while actual church attendance is known to be below 2%.

Some bottom lines

We know some sectors of the population have a larger proportion of active Christians than the 10-15% average and some a smaller one.
Seniors > young adults and children.
Uni-educated > less educated, especially when we compare within age groups.
Women > men.
These imbalances, which NCLS has long been telling us, are well-known. The Church needs to consider why some people are absent and how to reach them.

My look at the geography doesn't show that much:
Queensland and Sydney > Melbourne and Hobart.
Outer suburbs > inner suburbs (but I need to look further at this).

What next?

My next step will be looking specifically at Adelaide, where I may be able to look at numbers in each church in one or more denominations and thereby have a close estimate of the proportion of attenders in each SA3.