How we know.

This page sets out how the European Christian Media Survey 2026 was conducted, how we report from it, and where its limits are. It is written to be read in full before any figure on this site is quoted.

The survey

The European Christian Media Survey 2026 was conducted by New Faith Network using the research platform Iterate. Fieldwork ran in July 2026.

An invitation was sent to an email list of approximately 150,000 addresses in the Netherlands and Belgium, comprising current subscribers, former subscribers and people who had never held a subscription. Participation was voluntary. Respondents who completed the questionnaire were entered into a draw for one of ten annual subscriptions.

In total 4,071 people took part, a response rate of a little over 1.5 per cent.

Who responded

Of the respondents, 48 per cent were current subscribers to New Faith Network at the time of fieldwork, 21 per cent were former subscribers, 16 per cent had held a trial only and 15 per cent had never been a customer. Just over half of the sample therefore sits outside the current customer base.

Age was given by 3,251 respondents. Of those, 290 were under 35, 1,175 were between 35 and 54, and 1,786 were 55 or over. The sample is considerably older than the general population of either country.

Gender was given by 2,789 respondents, of whom 68 per cent are women and 31 per cent men. That imbalance is consistent across every age band, at 70 per cent women among under-35s and 67 per cent among those 55 and over. It may reflect the composition of the list, a known tendency for women to respond to surveys more often, or both. We report it because it is a second skew alongside age, and readers should weigh both when interpreting any total.

Church tradition was given by 2,778 respondents, with Evangelical (779) and Protestant (665) the largest groups, followed by no specific tradition (323), Reformed (241), Pentecostal (209), Catholic (203) and Baptist (176).

Of the respondents, 2,501 live in the Netherlands and 267 in Belgium.

What this survey is not

It is not representative. Respondents selected themselves by choosing to complete the questionnaire, and they were reached through one platform’s mailing list. The results describe an engaged Christian audience in two countries. They do not describe Christians in the Netherlands and Belgium as a whole, and they do not describe Europe.

It is not weighted. No correction has been applied for age, region, tradition or any other characteristic. The figures are the raw distribution of those who answered.

There is no control group. No non-Christian sample was surveyed. Any statement comparing Christians to non-Christians is outside what this research can support, and none appears on this site.

We state these three points wherever a reader might otherwise assume more than the data allows. They are limitations, not disclaimers, and they are the reason the comparison rules below exist.

How we report

Every figure carries its base. Bases differ between questions because not every respondent completed the questionnaire. A figure quoted against 2,957 respondents and one quoted against 4,071 are not interchangeable, and we give the base each time rather than implying a single sample size throughout.

Thresholds. A subgroup of 100 or more is reported normally. Between 30 and 99 it is reported as indicative, with the base shown alongside. Below 30 it is not published at all.

Rounding. Percentages are given as whole numbers. Decimals would imply a precision this sample does not carry.

Comparisons within the survey are made between groups of similar construction, with both bases stated.

How we compare with other research

Because the sample is older than the general population, comparing our totals with a national total would largely measure that age difference. We therefore compare within age bands only: respondents under 35 against the general population under 35, and so on.

Where an external figure is cited, we name the study, its publisher, its edition or year and its sample size. Where the question wording differs enough that a numerical comparison would mislead, we say so in the same place rather than leaving the reader to assume the figures are equivalent.

Sources used in edition 1 include NMO Mediatrends and the NMO podcast study (Nationaal Media Onderzoek, conducted by Ipsos I&O), the Digital News Report Netherlands 2026 (Commissariaat voor de Media with the Reuters Institute), the Reuters Institute Digital News Report 2026, the Markteffect Podcast Monitor, Geven in Nederland 2026 (VU Centre for Philanthropic Studies), God in Nederland (Radboud University with HDC), CBS StatLine and imec.digimeter 2025 (imec-mict-UGent).

Known limitations of edition 1

Trust was measured by source, not by medium. Respondents were asked which sources they trust most for faith, inspiration and Christian information. Churches, speakers and streaming services were not offered as answer options. This survey therefore says nothing about trust in those, and no such claim appears on this site. The question will be extended in edition 2.

Question wording is our own. It was not taken from the standardised instruments used by NMO or the Digital News Report. Comparisons with those studies are therefore indicative rather than exact. Harmonising the wording is planned for edition 2.

Belgium can be reported only in total. With 267 Belgian respondents, statements about the group as a whole are supportable. Breakdowns within it are not: only 23 Belgian respondents are under 35.

Behaviour and stated preference are different things. Several questions ask what respondents would like to see more of. That is what people say, not what they watch, and the two are known to diverge.

Personal data

Responses were collected with an email address for the prize draw. No personal data is published on this site or in the report. Figures are published in aggregate only, and no cell derived from fewer than thirty respondents is shown. Raw response data is not shared outside New Faith Network.

Open-text answers are held internally. Where any is quoted in the report, it is anonymised and carries no detail by which a respondent could be identified.

Cadence and corrections

The survey is intended to run annually in the same field window, so that year-on-year comparison is possible from edition 2 onward.

If a figure published here is wrong, we correct it, and the correction is noted on the page rather than made silently. Questions about method, or requests for a cut that is not published, are welcome; see about.