Are 1 in 5 mothers regularly skipping meals to better feed their children?

In a letter published in the Mirror today, leaders from the Church of England, Methodist Church and Quakers said that:

"One in five mothers report regularly skipping meals to better feed their children",

a claim which was also quoted in the Guardian.

This letter, in support of the End Hunger Fast campaign, comes a day after a rebuttal from the Prime Minister to comments from Cardinal Vincent Nichols on the Coalition's welfare reforms.

The campaign's website features the claim but does not have a source.

It is a strikingly similar claim to one that appeared in the news in 2012, which was attributed to a survey by online parenting organisation Netmums. If this is the source, we can't generalise from it to make a claim about all mothers.

Problems with surveying

The 1 in 5 figure made headlines back in February 2012, when the survey was published.

In February 2012, Netmums surveyed 1,924 parents. Respondents were asked 'What are the main ways you are cutting back household spending?' and 18% answered 'I will miss meals so my kids eat'.

So 1 in 5 survey respondents reported that they were missing meals in order to feed their children. However, to conclude that this is the experience of all mums is not so straightforward.

The survey notes that: "It is recognised that this survey has limitations as those completing it were a self-selected group, but Netmums attracts users from all walks of life and surveys of this size are generally found to be representative."

However, this is not entirely the case. The British Polling Council state that "a biased sample is a biased sample, however large it is". This means polls should take into account potential bias and adjust for it.

Say you ask 10,000 people in Britain to return a survey on whether they prefer tea or coffee, you get 2,000 back and conclude from their response that 80% of all Britons prefer tea.

However, you've asked for no other information, such as their age or gender. How do you know that the majority of people returning your survey weren't in an age group more likely to drink tea than coffee? The answer is you don't. You have no way of knowing if the sample is biased.

To get a representative sample of mothers, respondents need to be asked things like their working status, their household income and the number of children they have.  These answers should then be compared to census results on the same things. If there is, for example, a higher percentage of high income earners in the sample than there is in the general population of mothers, the sample would need to be adjusted for this when making conclusions.

The summary of the survey gives us no data to suggest that they have done this.

Another important unknown is how the survey was described to members. If the title of the survey reflected the content, so something like 'How has the economic downturn affected your family?', it seems likely that those struggling financially would be more likely to complete the survey. Therefore adjusting the responses to take account of income levels and number of children, for example, would be particularly important.

We also do not know if all respondents were indeed mothers - the survey refers to 'parents' and so could include fathers too.

We're waiting to hear back from the End Hunger Fast campaign to confirm if this is where their figure comes from.

Update 24/02/2014

The End Hunger Fast campaign have confirmed to us that they were basing their figure on the Netmums survey.

The context of the figure should have been explained so that it is clear it is not representative of the experience of all mothers nationally. We hope that the figure gets used more cautiously in future.