The impact of parental status on the visibility and evaluations of politicians

Abstract

In increasingly personalised electoral contests, voters use evaluations of candidates’ personal characteristics in their vote decisions, and candidates deploy personal information about themselves which they believe convey a positive message in their communications with voters. We expand the study of candidate characteristics to include parental status, examining the public’s view of politicians with and without children and the behaviour of politicians in their communications with voters. Men and women are equally likely to refer to their children regardless of party. We find a preference for candidates who are parents and no punishment effect for women politicians with children. Our findings, from a British study, contradict some of the research from the United States which finds that voters’ reactions to candidates’ parental status vary depending on candidate gender; as such, our results suggest that political and cultural context are important factors determining the role gender plays in political behaviour.

Keywords

candidate evaluations electoral behaviour gender parenthood personalisation political behaviour

There is an increasingly large body of research examining the impact that candidate characteristics have on voters’ preferences and attitudes. This literature has focussed predominantly on the United States and on a relatively small number of characteristics, most obviously candidate sex and race/ethnicity. But more recent research has begun to extend the study further, both geographically and in terms of subject area, finding other significant characteristics, including visual image (Banducci et al., 2008, Mattes and Milazzo, 2014), occupation and wealth (Campbell and Cowley, 2014a, McDermott, 2005), age (Campbell and Cowley 2014b, Trent et al., 2010) and residency (Arzheimer and Evans, 2012, 2014).

There is, however, relatively little research that considers the impact that candidates’ parental status may have on voter evaluations. The dearth of literature on this topic is surprising given that politicians routinely use images of themselves in domestic family settings – Langer argues that politicians increasingly use aspects of their personal lives in their campaigns in order to ‘offer a “human” persona’ (Langer, 2009: 61) – and that the subject manifests itself frequently in political discussion. There is plenty of research that considers the impact of voters’ parental status on their voting behaviour and political attitudes (Elder and Greene, 2007, 2008, 2012; Greenlee, 2014, Oswald and Powdthavee, 2010) but much less that tests the impact of politicians’ parental status on vote choice and political attitudes.

Interest in the subject is however slowly growing (Bell and Kaufmann, 2015; Morin and Taylor, 2008; Stalsburg, 2010), not least because of the way that motherhood has been politicised, particularly in the United States since Sarah Palin’s candidacy for Vice President in 2008 (Deason et al., 2015; Greenlee et al., 2017), and the way that the topic of parenthood in politics is so obviously gendered (Miller, 2017; Thomas and Bittner, 2017). For example, the Australian Prime Minister Julia Gillard was variously described by some of her opponents as ‘deliberately barren’ and ‘an unproductive old cow’ – phraseology that, for obvious reasons, would never be ascribed to a man – along with the claim that because she had ‘chosen not to be a parent’, she was ‘very much a one-dimensional person’. British Prime Minister Theresa May’s lack of children was raised by one of her opponents for the Conservative leadership, who argued that her childlessness meant May lacked a ‘stake in the future’.¹ In New Zealand, Jacinda Ardern was asked about whether she intended to have children within hours of becoming leader of the Labour Party.

This article reports two studies into the effect of politicians’ parental status, one reporting on the behaviour of politicians, the other examining the reaction of voters. We test whether the findings of the extant empirical work – namely, that male politicians are more likely to publicise their parental status than women politicians and that women politicians are more likely to be negatively evaluated for their parental status than men – hold in the case of Britain. Evaluating both legislators’ behaviour and the reactions of voters allows us to assess whether politicians might perceive a bias and whether one in fact exists. We report data from Great Britain, where the issue has occasionally been one of topical political debate – as noted above – but where a smaller proportion of the public has traditional attitudes to gender roles, to several of the countries where the subject has been researched thus far. We assess whether British Members of Parliament (MPs) differ in the extent to which they display or hide details of their families from the public using observational data and whether the British public view politicians differently if they have children by using a survey experiment. We find that politicians do utilise their parental status in publicity material and find a clear public preference for candidates who are parents over those who are childless – but we find relatively few gendered or partisan effects.

Hypotheses

Whether politicians’ parental status influences voter support has, with some notable exceptions (Elder and Greene, 2012), been paid relatively little attention by political scientists. The few studies to investigate the topic have found that the impact of politicians’ parental status on candidate evaluations was mediated by gender. In an experiment conducted with an undergraduate sample in the United States, Brittany Stalsburg demonstrated that childless women were rated less favourably compared to childless men, and men and women candidates with children (Stalsburg, 2010). A bias against unmarried childless women candidates was also found by Melissa Bell and Karen Kaufmann in their survey experiment (Bell and Kaufmann, 2015). This research chimes with the above examples of both Theresa May and Julia Gillard, in which being a parent in general may be considered an electoral asset, but where childless women candidates are thought to suffer an electoral penalty.

However, the broader literature relating to this question is more divided on the consequences of parenthood, and especially motherhood. It can equally be argued that it should be women with children who would suffer electorally, based on the traditional stereotypical view that mothers should prioritise giving childcare over paid work (Douglas and Michaels, 2004; Mezey and Pillard, 2012; Norris and Lovenduski, 1995). Although attitudes to traditional gender roles have changed markedly in western democracies over the last half century, women continue to make up the overwhelming majority of carers of young children and there remains a minority of the public who believe that women’s place is in the home not the workplace (Campbell et al., 2010; Inglehart and Norris, 2000). It is not unknown for women candidates for office to be asked how they will combine elected office with family life, questions that are rarely, if ever, asked of male candidates (Dolan, 2014: 2). From this perspective, having children should have a reverse effect on support for women candidates compared with men: fatherhood might be an electoral asset, motherhood an electoral constraint.

The conflicting nature of the potential impact of stereotypes on political ambition and support for women candidates who are mothers was explored by Grace Deason et al. in their discussion of the politicisation of motherhood in the United States (Deason et al., 2015). They describe how politicians’ increasing use of their parental status to develop their political brand provides both constraints and opportunities for women candidates, and how the increased visibility of mothers in politics has potentially expanded concepts of political leadership to include feminine traits associated with motherhood. Thus, candidates who are mothers may receive an electoral penalty because voters may question how they can fulfil their domestic role while holding office but equally they may be viewed as uniquely capable of performing aspects of political leadership associated with representing the interests of children and considered to have special gifts associated with multitasking, consensus building and compassion that are associated with motherhood.

In sum, the extant literature views the potential impact of parenthood on electability as gendered but the effect on women politicians is not clear. Some suggest that motherhood, as opposed to fatherhood, may be an electoral constraint, while other literature sees it as an asset. However, all the literature argues that such effects may be contingent, mediated through either party or place. One study analysing candidates’ chances of winning in elections in the United States found that Republicans were less likely to vote for women who were the mothers of young children than men who were the fathers of young children, but the reverse was true of Democratic candidates (Morin and Taylor, 2008). Stalsburg (2010) found that Republican supporters were the least favourable towards women candidates without children. Similarly, Thomas and Lambert hypothesise that a candidate’s decision to promote their parental status will be influenced by their party membership. Male candidates from conservative parties that espouse a traditional ideological position on gender roles may be more likely to display their parental status in a bid to align themselves with the traditional family. On the contrary, conservative female politicians may be less likely to draw attention to their parental status, particularly when they are the mothers of young children, for fear of violating gender norms.² This contrast suggests that, to the extent that attitudes to traditional gender roles are correlated with partisanship, voters’ reactions to candidates’ parental status may vary according to voters’ party identification.

More broadly, however, we would also expect contextual variation in the extent to which candidates’ gender influences their willingness to reveal their parental status depending on the wider gender politics. In countries with a dominant norm that the mothers of young children should be at home (such as Germany), women politicians who are mothers may well be more inclined to hide their parental status than in countries where the traditional view has subsided (Kürschner, 2011). In countries such as the United States where attitudes to gender roles are polarised by party, there may be more complexity in the extent to which women with children are evaluated and how they present themselves depending on their party allegiance (Deason et al., 2015). However, in countries where there is currently a widespread acceptance of more equal gender roles, such as Britain, there may be little relationship between gender, parental status and political behaviour.

This article draws on data from Great Britain, where there has been a considerable shift in public attitudes to mothers and paid employment in recent years (Park et al., 2013: 115). The shift is not absolute and there remain gender differences in the division of unpaid work (Miller, 2012) and women continue to face gender discrimination in employment (Boeckmann et al., 2015).³ However, none of the major parties now hold explicitly traditional positions on gender roles (Campbell, 2016). Since 2005, which saw the election of David Cameron as leader of the Conservative party, at least six of the seven main UK political parties now espouse feminist views on gender roles, and almost all are committed to seeing an increase in the number of women MPs, even if they differ in the seriousness with which they take the issue.⁴ The 2017 election saw a record-high number of women elected to the House of Commons, although they still constitute a minority, at some 32% of MPs (Campbell and Hudson, in press). There is, in other words, a relatively liberal approach to gender, and one which is not especially polarised by party.

In this article, we test six hypotheses, drawn from the above discussion, three relating to politicians, three relating to voters.⁵ In common with the extant literature discussed above, we assume that in general British voters will prefer candidates with children, as a proxy for a connection with ordinary life and thus reduced social distance between voter and representative, and that politicians will act accordingly, not hiding their parental status – but that because of the wider political context there will be no significant differences in either voter attitudes or the behaviour of MPs by either sex or party.

Politicians

H1. Politicians with children will make reference to them in material for external consumption.

H2. Female politicians with children will be no less likely to make reference to them in material for external consumption than male politicians with children.

H3. Politicians with children from conservative parties will be no more or less likely to make reference to them in material for external consumption than politicians with children from leftist parties.

Voters

H4. Voters will react positively to politicians with children.

H5. Voters will not react negatively to women politicians with children.

H6. Supporters of conservative parties will not react negatively to women politicians with children.

We test these hypotheses by drawing on two studies that examine the relationship between parenthood and politicians. Study 1 utilises observational data to examine which MPs are more or less likely to hide details of their families from the public. Study 2 is based on experimental survey data to assess whether the public view politicians differently if they have children.

Study 1: Politicians

Our first study examines how British MPs presented themselves, and the extent to which they did or did not publicise information about their children. For data, we utilised British MPs’ websites. Almost all British MPs now have their own website and almost all of these have a section entitled ‘About’ or ‘Biography’ or similar, in which the MP provides information about themselves, their background, their beliefs and so on. There is no standard format to this material. Some MPs provide only very cursory information, others are much more detailed. Some talk solely about their political beliefs or careers, some focus on their personal background; most talk about some mix of the two. Importantly for our purposes, some talk about their families, others do not; some utilise photographs of their families, others do not. The most important point about such websites is that the MP can choose how they present this information. Subject to almost no constraints, they can choose what to reveal and what to omit. Their websites therefore present the image that the MP wishes to project to the voter.

We have chosen to compare MPs’ rather than candidates’ websites because candidates are only in place during election campaigns and their websites vary considerably in quality based on seat marginality and the likelihood of the candidate winning the seat as more resources tend to be expended by parties in its target seats. Moreover, women candidates are also more often placed in unwinnable seats which would introduce bias into our data. By comparing existing MPs, we are therefore considering a more homogeneous group.

Of the 650 MPs in the House of Commons, in April 2014, we found 604 (93%) who had their own websites.⁶ We include in this group MPs who did not have a personal site but where there was a considerable section about the MP on a local party site (and where in many such cases, it was fairly obvious the site was essentially focussed on the MP).⁷ In another 27 (4%) cases, we found MPs who had no individual or local website, but where there was a profile hosted on a national or regional party website.⁸ Such profiles still exhibited considerable variation in content, but because it is possible that MPs have less freedom over the content of such sites (and certainly less control over issues of presentation), we analysed these separately (although, in practice, the differences appear to be negligible). Below we report findings from the 97% of MPs with some web presence, but the difference between the 97% group and the 94% was never larger than 1 percentage point in any of the findings reported below.⁹

Of the 631 MPs with some web presence, 292 (46%) made some reference to their own children, 339 (54%) did not. Indicative examples would include the following:

‘John and his wife Susan live in his Lincolnshire constituency and have two young sons’.

‘Heather has lived in Bretby for the last 22 years with her husband and daughter’.

‘Elizabeth is married with two daughters’.

‘He is married to Michelle and is the proud father of three daughters’.

‘I live in South Devon with my husband Adrian and we have 3 children, all at university’.

Another 13 had some additional reference to parenthood, but with no explicit reference to children. Photographs, however, were much less common: just 27 (4%) had pictures of their children (where their identity was either explicitly labelled or obvious from the context).

A basic descriptive analysis of the data showed that of those MPs with websites, 34% of women and 50% of men mentioned their children in their personal website, and 1% of women compared with 5% of men included pictures of their children. At first sight, therefore, this appeared to reject both H1 and H2, since only a minority of MPs were displaying information about children, with male MPs being much more likely to do so than female MPs. However, not all MPs will have children, and while this apparent sex gap may occur because women disproportionately ‘hide’ their children, it is equally plausible that women MPs simply have fewer children. Recent British research has demonstrated that women MPs are more often childless than their male colleagues: 45% of women sitting in the House of Commons in 2013 had no children compared to 28% of men (Campbell and Childs, 2014).

In order to control for this, we draw on a 2013 survey of MPs which identified 426 MPs with children, of which 403 had their own website.¹⁰ We merged these data with the data on websites and re-examined the self-presentation of MPs, this time focussing just on those MPs that we knew had children. This exercise produces very different results. Of those MPs who we knew had children, a clear majority, 66%, had some mention of those children on their website, and there was now no statistically significant difference between men (67%) and women MPs (62%) (thus supporting both H1 and H2).¹¹ There was, however, still a difference in the proportion of men and women displaying a picture of their children. Of those MPs with children, some 6% had a photograph of one or more of their children on their website. Of women MPs, the figure was 1%, of men MPs it was 8%, a statistically significant difference (p < 0.05).¹² Even this difference still meant that while male MPs more often used a photograph of their children on their website, very few did so.¹³

Given that some extant studies had found differences depending on the age of children, again mediated by sex, we also attempted to examine this difference. The 2013 survey gave us the date of birth of the eldest child of 201 MPs, which serves as a proxy for having young children (albeit with some error where there is a substantial age gap between an MP’s children). Table 1 demonstrates that both men and women MPs were more likely to mention their children on their website when their eldest child was under 16 years old than when their eldest child was over 16.¹⁴ Men with children under 16 more often mentioned their children than women with young children but (albeit with a relatively small sample size) the difference is not statistically significant. Here too, therefore, we find no significant differences by sex. The data therefore appear to confirm H2, with the caveat that it does not apply to the small number of politicians, overwhelmingly male, who displayed photographs of their children.

Table 1.

Number (%) of mentions of children by MP’s sex and age of their eldest child (N = 201).

	Age of eldest child	Children not mentioned	Children mentioned	Total
Women	Over 16	16 (46%)	19 (54%)	35 (100%)
	Under 16	6 (35%)	11 (65%)	17 (100%)
Men	Over 16	30 (32%)	64 (68%)	94 (100%)
	Under 16	11 (20%)	44 (80%)	55 (100%)

MPs: Members of Parliament.

Last, we consider H3 that politicians with children from conservative parties will be no more or less likely to make reference to them in material for external consumption than politicians with children from leftist parties. We find that of MPs with children 79% of Conservative MPs compared with 55% of Labour MPs made some reference to parenting in their website; Table 2 shows the percentages broken down by sex and party.¹⁵ Contrary to H3, however, we find that Conservative MPs were noticeably more likely to refer to their children than were Labour MPs. However, the within-party sex differences are not statistically significant, with approximately the same proportion of Conservative women MPs making mentions of their children as Conservative men (81% of women compared to 78% of men), with a similarly small difference for Labour MPs (52% compared to 56%).

Table 2.

Frequency of mentions of parenting by MP’s sex and party (N = 360).

Party	Sex	No reference to parenting (%)	At least one reference (%)	Total (%)
Conservative	Women	19	81	100
Conservative	Men	22	78	100
Labour	Women	48	52	100
Labour	Men	44	56	100

MPs: Members of Parliament.

Study 2: Voters

Having demonstrated that British politicians mostly do make of mention any children they have, in our second study we examined the public’s reaction to politicians depending on their parental status. We used a survey experiment to create a low-information environment where respondents had to compare two politicians and choose which one they would prefer to be their representative. Experimental methods are becoming increasingly popular in political science (Birch and Allen, 2011; Druckman et al., 2006; Huddy and Terkildsen, 1993; Rosenberg and McCafferty, 1987; Sanbonmatsu, 2002). They offer the opportunity to model hypothetical scenarios giving us insights into the priorities of citizens not possible with conventional survey or observational data. In this study, we use a sample of the adult British population which allows us to test elements of Stalsburg’s US undergraduate sample study in the United Kingdom on a wider cross-section of society.

We ran a split-sample Internet survey with the survey company YouGov.¹⁶ Each survey involved respondents reading two short profiles about hypothetical politicians, and then deciding which of the two politicians they preferred. The context was pared back to one where biographical information about the politician was the only material available to respondents. We sought to give each characteristic the maximum chance of having an impact on preference without introducing another layer of complexity by interacting with political party. Following Sanbonmatsu (2002), our research design included profiles of two politicians, initially described as follows:

Please read these two short profiles of potential parliamentary candidates.

John Burns is 48 years old, and was born and brought up in your area, before going to University to study for a degree in Physics. After university John trained as an accountant, and set up a company ten years ago; it now employs seven people. John has interests in the health service and the environment. He is married.

George Mountford is 45 years old. He lives in the constituency and studied English at University. He is a solicitor and runs a busy local practice. George is passionate about education and pensions. He is married.

We then manipulated the biographical information in two ways, changing both the sex and the number of children involved for both candidates; this resulted in eight treatments in total. Approximately half of respondents saw ‘John’ and ‘George’ (as above); in the remainder, George became ‘Sarah’. The change in name (and consequential changes, such as pronouns) aside, the profiles remained otherwise identical. We also changed the number of children that each of our hypothetical candidates had. The four variants were as follows: no mention of children, both with two children, John with no mention of children and George/Sarah with two children, and John with two children and George/Sarah with no mention of children.¹⁷ The experiment is thus constructed to allow us to compare gender effects, parenting effects, and the interaction between the two.

We asked respondents which candidates they would prefer as their MP, and to compare the candidates on three traits. There are a large number of candidate traits used in the academic literature, including (but not limited to) ‘competence’, ‘experience’, ‘strength’, ‘leadership ability’, ‘effectiveness’, ‘integrity’, ‘honesty’, ‘morality’ ‘trustworthiness’, ‘compassion’, ‘warmth’, ‘approachability’ and ‘likeableness’ (Bartels, 2002; Johns and Shephard, 2008; McDermott, 1998; Miller et al., 1966; Miller and Shanks, 1996; Peterson, 2005; Rosenberg and McCafferty, 1987). We examined the impact of cues on three traits: approachability, experience and effectiveness. Approachability was selected both to tap into feelings of ‘compassion’ (commonly associated with femininity and which we expected could trigger gender stereotypes where they exist); approachability evokes an element of commonality or shared understanding between voter and candidate rather than simply suggesting an agreeable glow emanating from one to the other. We selected experience as a measure of competence rather than strength or leadership ability based on the same logic; strength is associated with masculinity and would most likely yield gender effects and bias results. Finally, we chose effectiveness because it is a measure of competence that looks at potential outcomes which should be the most important to voters.¹⁸

Each screen concluded with the following questions:

Without knowing which party they stand for, which of them do you think would be:

More approachable as an MP: [Response options: John Neither George]

More experienced as an MP: [Response options: John Neither George]

More effective as an MP: [Response options: John Neither George]

Which would you prefer as your MP: [Response options: John Neither George]

Respondents were surveyed on 8–11 June 2014. Randomisation was conducted by the survey company. Total sample size, across the 4 days, was 5816, with sub-samples ranging in size between 700 and 758 respondents, as listed in Table 3. Any comparison of two sub-samples thus draws on a sample of more than 1400, easily large enough to draw robust conclusions.

Table 3.

Experimental manipulations.

Variant	Sex and parenthood information	N
1	Both men	720
2	Both men with two children	715
3	Both men; George with two children	700
4	Both men; John with two children	747
5	Man and woman	729
6	Man and woman, with two children	739
7	Man and woman; woman with two children	708
8	Man and woman; man with two children	758

Table 4 shows the scores from the question about which candidate respondents preferred overall. The percentage selecting John varied between 29% and 41%, depending on the biographical information shown, with the percentage selecting George/Sarah ranging between 23% and 34%. Whatever the variant, there were a sizable number who were unable to choose (of between 36% and 41%). In general, John was the more popular of the candidates, usually being preferred to George/Sarah, but not always, and the size of the lead varied from 18 percentage points down to one scenario where Sarah was the favoured candidate by five points. We are not interested here in why John is broadly the more popular candidate; what matters to us are the variations that occur when we alter the biographical information shown to respondents.

Table 4.

Percentage of participants who preferred a candidate, by treatment.

Variant		John	George	Neither	John lead	Net effect of having children
1	No mention of children	37	23	40	+14
2	Both have two children	41	23	36	+18
3	George has two children	33	27	41	+6	+8
4	John has two children	41	23	36	+18	+4
			Sarah
5	No mention of children	30	32	39	−2
6	Both have two children	31	30	39	+1
7	Sarah has two children	29	34	38	−5	+3
8	John has two children	37	26	37	+11	+13^a

Data weighted by YouGov’s standard survey weight.

Difference between sub-samples significant at the 0.01 level chi-square test.

In each of the four scenarios above, John’s lead was smaller when facing Sarah than when facing George (for example, comparing variants 1 vs 5, 2 vs 6, 3 vs 7 and 4 vs 8), but the effect on that lead of any or all of the candidates having children was mostly not statistically significant. The net effect of having children on candidate preference is calculated by using the scenario where neither candidate has children as the baseline; for example, when George is described as having two children and John no children (variant 3), George gains 8 percentage points when compared to the scenario where neither have children (variant 1). Using this method, and averaging across the four scenarios with children, the average net gain from having children is a non-trivial 7 percentage points. The biggest change from the baseline – and the only statistically significant effect – occurred when the male candidate had children and the female candidate did not.

We now turn to the three underlying traits: approachability, experience, effectiveness. Table 5 shows the results from the question about how approachable the candidates seem. It, and Tables 6 and 7, is calculated in an identical way to Table 4. It shows a similar pattern of findings to Table 4. The total net effect of having children on approachability was 9 percentage points. This time, however, there were two statistically significant differences – in both cases where one candidate had children and the other did not. And again, the biggest single effect was when John has children and was facing a female candidate (variant 8).

Table 5.

Percentage of participants who rated a candidate as most approachable, by treatment.

Variant		John	George	Neither	John lead	Net effect of having children
1	No mention of children	38	22	41	+16
2	Both have two children	42	21	37	+21
3	George has two children	34	27	40	+7	+9
4	John has two children	45	19	36	+26	+10^a
			Sarah
5	No mention of children	27	33	40	−6
6	Both have two children	29	30	41	−1
7	Sarah has two children	26	37	37	−11	+5
8	John has two children	35	28	37	+7	+13^b

Data weighted by YouGov’s standard survey weight.

Difference between sub-samples significant at the 0.05 level chi-square test.

Difference between sub-samples significant at the 0.01 level chi-square test.

Table 6.

Percentage of participants who rated a candidate as most experienced, by treatment.

Variant		John	George	Neither	John lead	Net effect of having children
1	No mention of children	21	24	54	−3
2	Both have two children	25	23	52	+2
3	George has two children	23	23	53	0	−3
4	John has two children	28	23	50	+5	+8^a
			Sarah
5	No mention of children	26	21	53	+5
6	Both have two children	27	20	53	+7
7	Sarah has two children	25	22	53	+3	+2
8	John has two children	33	17	50	+16	+11^b

Data weighted by YouGov’s standard survey weight.

Difference between sub-samples significant at the 0.05 level chi-square test.

Difference between sub-samples significant at the 0.01 level chi-square test.

Table 7.

Percentage of participants who rated a candidate as most effective, by treatment.

Variant		John	George	Neither	John lead	Net effect of having children
1	No mention of children	28	25	47	+3
2	Both have two children	33	23	44	+10
3	George has two children	26	25	49	+1	+2
4	John has two children	31	24	45	+7	+4
			Sarah
5	No mention of children	25	30	46	−5
6	Both have two children	26	23	50	+3
7	Sarah has two children	28	28	45	0	−5
8	John has two children	32	24	44	+8	+13^a

Data weighted by YouGov’s standard survey weight.

Difference between sub-samples significant at the 0.01 level chi-square test.

Respondents clearly found it harder to choose which of the candidates was more experienced: for all eight variants in Table 6, the majority selected the ‘neither’ option. Yet of those who were able to select a candidate, the pattern was broadly similar, if smaller in magnitude, to that seen in Tables 4 and 5. The point at which John did best was when he had children and his rival did not. Sarah did best when she had two children and John did not. The average net effect of having children was 4.5 percentage points, just over half the average net effect on approachability. And again, the largest deviation from the baseline comes with variant 8, when the male candidate with children was facing a female candidate without children. Note that for each variant John was considered more experienced when compared to Sarah than when compared to George, in the same way (in Table 5) that John is considered less approachable when compared to Sarah rather than George. This pattern is identical to that noted by Campbell and Cowley (2014b), in which otherwise identical candidates are considered less experienced but more approachable if they are women than if they are men.

Next we consider effectiveness (Table 7). We again find high levels of respondents who selected neither candidate, with a net effect of approximately 3.5 percentage points, and no statistically significant difference between several of the sub-samples. Again, however, the largest net effect (and the only statistically significant one) is when the male candidate had children and was facing a female candidate without children.

We thus find clear evidence that voters think more highly of politicians with children when compared to politicians who do not (H4). Of the 16 results testing a candidate with children against one without, there is a positive effect in 14 cases, which was statistically significant in six cases. We found no evidence that voters reacted negatively to women politicians with children (also confirming H5), but we did find some evidence that women without children are less attractive when compared to a male candidate with children.¹⁹ In all four tables – measuring approachability, experience, effectiveness and overall preference – the result for the male candidate with children versus the female candidate without shows a statistically significant advantage for the man.

Finally, we examine whether these findings vary by the respondent’s ideological position. We split respondents into two broad groups: those who intended to vote for parties seen as being on the left of the mean point on the ideological spectrum, and those on the right. We excluded non-GB wide parties as well as those on the ideological extremes. Our left group therefore includes voters who supported Labour, the Liberal Democrats and the Greens, while our right group includes those who supported either the Conservatives or the United Kingdom Independence Party (UKIP). This split creates two broadly equal-sized groups, of just over 2000 respondents each.²⁰

As is clear from Table 8, these two groups behave differently to one another. For one thing, all four of the ‘Sarah’ options are more popular with those on the left than those on the right. Indeed, although Sarah often led John among the full sample, among those on the right she is behind in all four variants of the profiles. These hypothetical female candidates do less well with voters on the right, whatever their parental status. Of more interest to us here, however, are the variants within each group. We find no evidence that voters on the right are less likely to prefer women politicians without children than voters on the left. The net positive effect of having children is slightly smaller for the woman candidate among both left and right leaning voters. When George has children, he gains 5 percentage points over John, and when Sarah has children, she gains 1 percentage point over John among left leaning voters – a difference of 4 percentage points. Among the right-leaning voters, George gains 6 percentage points over John when he has children, and Sarah gains 3 percentage points; a difference of 3 percentage points.

Table 8.

Percentage of participants who preferred the candidate, by treatment and left–right position.

Variant		John	George	Neither	John lead	Net effect of having children
Left
1	No mention of children	43	26	31	+17
2	Both have two children	41	26	34	+15
3	George has two children	37	25	38	+12	+5
4	John has two children	40	24	36	+16	−1
			Sarah
5	No mention of children	28	32	40	−4
6	Both have two children	30	36	34	−6
7	Sarah has two children	31	36	34	−5	+1
8	John has two children	41	28	32	+13	+17^a
Right			George
1	No mention of children	36	26	38	+10
2	Both have two children	47	23	30	+24
3	George has two children	37	33	31	+4	+6
4	John has two children	43	26	31	+17	+7
			Sarah
5	No mention of children	36	32	32	+4
6	Both have two children	38	27	35	+11
7	Sarah has two children	32	33	35	+1	+3
8	John has two children	41	28	31	+13	+9

Data weighted by YouGov’s standard survey weight.

Difference between sub-samples significant at the 0.01 level chi-square test.

Discussion and conclusion

Our research shows that children are an electoral asset – and it would appear that British politicians know that. In study 1, we found that politicians with children overwhelmingly make mention of their children. In study 2, we found clear evidence that politicians with children tended to receive higher evaluations than those without. The boost the hypothetical candidates in our survey received was not massive, but neither was it trivial.

Moreover, we find only small differences between men and women. We find no general punishment effect for women politicians with children, nor do we find voters from the left or the right behaving very differently. Moreover, in terms of gender, both male and female politicians are equally likely to refer to their children, regardless of party. Initial appearances of gender differences are due to the differential nature of politicians’ parental status. Once that is controlled for, almost no differences remain, aside from in the use of photographs but this difference relates only to a very small sample of both men and women. However, we did find that there was a large gap between Conservative and Labour politicians overall with Conservatives more likely to refer to their children than Labour politicians. Nothing in the literature suggested this finding and further work is therefore required to see whether this effect is a more common phenomenon and if so to establish an explanation. Perhaps the most significant difference was that women politicians without children were punished electorally more for their lack of children than male politicians of a similar parental status. Again, the effect is not massive, but it is consistent, and could matter in close electoral races.

The disadvantage of experimental methods, as in study 2, is the artificiality of the setting. They can only offer us an insight into how voters might respond in reality that must be further tested with observational data. There is an inevitable trade-off between internal and external validity when using experimental methods. The advantage of experimental data is, however, that it allows a very clean test of the research hypothesis which excludes possible confounding factors. We are aware of Kathleen Dolan’s (2014) note of caution on the use of survey experiments to study the impact of gender stereotypes on candidate evaluations – namely, that they are more likely to be evident in experimental settings where the candidates are not known to the voter than in real-world elections. Likewise Deborah Brooks (2013) provides evidence that gender stereotypes do not damage the electoral chances of experienced women candidates. Thus survey experiments that do not adequately control for experience might inflate gender differences in candidate popularity. However, survey experiments can be a useful first step in the comparative investigation of whether there is cross-national variation in underlying attitudes towards the parental status of candidates. Furthermore, in contexts where there are a relatively small number of women candidates, fewer women candidates with children and party variation in the number of women candidates elected, survey experiments allow us to overcome confounding bias.

Moreover, both for this reason and in general, this experimental effect is likely to be maximal given that our participants lacked other important information about the candidates, such as their party allegiance (and that, in real-world contests, many voters will not know about the parenting status of their candidates). Future research will also want to include the impact of candidates’ party but given the paucity of research in this area the first step is to establish whether an aggregate level effect of parenthood on candidate preference might exist.

However, in tight electoral districts or leadership contexts, this small advantage may be electorally significant. Given the high proportion of MPs who are parents who make reference to their children in their websites, as we showed in study 2, we suspect they are intuitively aware of the advantage that this gives them.

It is plausible that in other contexts, where gender stereotypes and attachment to traditional gender roles are more prevalent in society, a gap between men and women politicians’ willingness to identify themselves as parents might exist. But there is no evidence that there is such a gender divide currently in Britain.

Footnotes

Funding

The author(s) received no financial support for the research, authorship and/or publication of this article.

Notes

References

Arzheimer

Evans

(2012) Geolocation and voting: Candidate-voter distance effects on party choice in the 2010 General Election in England. Political Geography 31: 301–310.

Arzheimer

Evans

(2014) Candidate geolocation and voter choice in the 2013 English county council elections. Research and Politics 1: 1–9.

Banducci

Karp

Thrasher

et al . (2008) Ballot photographs as cues in low-information elections. Political Psychology 29: 903–917.

Bartels

(2002) The impact of candidate traits in American Presidential Elections. In: King

(ed.) Leaders’ Personalities and the Outcomes of Democratic Elections. Oxford: Oxford University Press, pp.44–69.

Bell

Kaufmann

(2015) The electoral consequences of marriage and motherhood: How gender traits influence voter evaluations of female candidates. Journal of Women, Politics & Policy 36: 1–21.

Birch

Allen

(2011) Framing ethical judgements about politicians: The role of informational cues and political sophistication in shaping how people evaluate the ethical behaviour of their political leaders. In: ECPR General Conference, Reykjavik, 25–27 August.

Boeckmann

Misra

Budig

(2015) Cultural and institutional factors shaping mothers’ employment and working hours in postindustrial countries. Social Forces 93: 1301–1333.

Brooks

(2013) He Runs, She Runs: Why Gender Stereotypes Do Not Harm Women Candidates. Princeton, NJ: Princeton University Press.

Campbell

(2016) Representing women voters: The role of the gender gap and the response of political parties. Party Politics 22: 587–597.

10.

Campbell

Childs

(2014) Parents in parliament: ‘Where’s mum?’ Political Quarterly 85: 487–492.

11.

Campbell

Cowley

(2014a) Rich man, poor man, politician man: Wealth effects in a candidate biography survey experiment. British Journal of Politics and International Relations 16: 56–74.

12.

Campbell

Cowley

(2014b) What voters want: Reactions to candidate characteristics in a survey experiment. Political Studies 62: 745–765.

13.

Campbell

Hudson

(in press) Candidates and MPs. In: Cowley

Kavanagh

(eds) The British General Election of 2017. London: Palgrave.

14.

Campbell

Childs

Lovenduski

(2010) Do women need women MPs. British Journal of Political Science 40: 171–194.

15.

Deason

Greenlee

Langner

(2015) Mothers on the campaign trail: Implications of politicized motherhood for women in politics. Politics, Groups, and Identities 3: 133–148.

16.

Dolan

(2014) When Does Gender Matter? Women Candidates and Gender Stereotypes in American Elections. New York: Oxford University Press.

17.

Douglas

Michaels

(2004) The Mommy Myth: The Idealization of Motherhood and How It Has Undermined All Women. New York: Simon & Schuster.

18.

Druckman

Green

Kuklinski

et al . (2006) The growth and development of experimental research in political science. American Political Science Review 100: 627–635.

19.

Elder

Greene

(2007) The myth of ‘security moms’ and ‘NASCAR dads’: Parenthood, political stereotypes, and the 2004 election. Social Science Quarterly 88: 1–19.

20.

Elder

Greene

(2008) Parenthood and the gender gap. In: Whitaker

(ed.) Voting the Gender Gap. Chicago, IL: University of Illinois Press, pp.119–140.

21.

Elder

Greene

(2012) The politics of parenthood: Parenthood effects on issue attitudes and candidate evaluations in 2008. American Politics Research 40: 419–449.

22.

Greenlee

(2014) The Political Consequences of Motherhood. Ann Arbor, MI: University of Michigan Press.

23.

Greenlee

Deason

Langner

(2017) The impact of motherhood and maternal messages on political candidacies. In: Bos

Schneider

(eds) The Political Psychology of Women in U.S. Politics. New York: Routledge, pp. 133–148.

24.

Huddy

Terkildsen

(1993) Gender stereotypes and the perception of male and female candidates. American Journal of Political Science 37: 119–147.

25.

Inglehart

Norris

(2000) The developmental theory of the gender gap: Women and men’s voting behaviour in global perspective. International Political Science Review 21: 441–462.

26.

Johns

Shephard

(2008) Candidate image and electoral preference in Britain. British Politics 3: 324–349.

27.

Kürschner

(2011) Good mother or good politician? Politics and motherhood in Germany and the United States. In: Eighteenth International Conference of Europeanists, Barcelona, 20–22 June.

28.

Langer

(2009) The politicization of private persona: Exceptional leaders or the new rule? The case of the United Kingdom and the Blair effect. International Journal of Press/Politics 15: 60–76.

29.

McDermott

(1998) Race and gender cues in low-information elections. Political Research Quarterly 51: 895–918.

30.

McDermott

(2005) Candidate occupations and voter information shortcuts. Journal of Politics 67: 201–219.

31.

Mattes

Milazzo

(2014) Pretty faces, marginal races: Predicting election outcomes using trait assessments of British parliamentary candidates. Electoral Studies 34: 177–189.

32.

Mezey

Pillard

(2012) Against the new maternalism. Michigan Journal of Gender and Law 18: 229–296.

33.

Miller

Wattenberg

Malanchuk

(1966) Schematic assessments of presidential candidates. American Political Science Review 80: 521–540.

34.

Miller

(2017) Mothers and the media on the campaign trail. In: Thomas

Bittner

(eds) Mothers and Others: The Role of Parenthood in Politics. Vancouver, BC, Canada: UBC Press, pp. 155–177.

35.

Miller

(2012) Balancing caring and paid work in the UK: Narrating ‘choices’ as first-time parents. International Review of Sociology 22: 39–52.

36.

Miller

Shanks

(1996) The New American Voter. Cambridge, MA: Harvard University Press.

37.

Morin

Taylor

(2008) Revisiting the Mommy Wars: Politics, Gender and Parenthood. Washington, DC: Pew Research Centre.

38.

Norris

Lovenduski

(1995) Political Recruitment. Cambridge: Cambridge University Press.

39.

Oswald

Powdthavee

(2010) Daughters and Left-Wing Voting. Review of Economics and Statistics 92: 213–227.

40.

Park

Bryson

Clery

et al . (2013) British Social Attitudes: The 30th Report. London: National Centre for Social Research.

41.

Peterson

(2005) Heterogeneity and Certainty in Candidate Evaluations. Political Behavior 27: 1–24.

42.

Rosenberg

McCafferty

(1987) The image and the vote manipulating voters preferences. Public Opinion Quarterly 51: 31–47.

43.

Sanbonmatsu

(2002) Gender stereotypes and vote choice. American Journal of Political Science 46: 20–34.

44.

Stalsburg

(2010) Voting for mom: The political consequences of being a parent for male and female candidates. Politics and Gender 6: 373–404.

45.

Thomas

Bittner

(2017) Mothers and Others: The Role of Parenthood in Politics. Vancouver: UBC Press.

46.

Thomas

Lambart

(2017) Private Mom Versus Political Dad? Communications of Parental Status in the 41st Canadian Parliament. In: Thomas

Bittner

(eds) Mothers and Others. The Role of Parenthood in Politics. Vancouver: UBC Press, pp. 135–54.

47.

Trent

Short-Thompson

Mongeau

et al . (2010) Cracked and shattered ceilings: Gender, race, religion, age, and the ideal candidate. American Behavioural Scientist 54: 163–183.