Does it pay to break male gender stereotypes in advertising? A comparison of advertisement effectiveness between the United Kingdom, Poland and South Africa

Abstract

Advertisers shy away from using non-traditional (vs. traditional) male gender portrayals even though theory suggests they may be more effective cross-nationally. Two main hypotheses were tested cross-nationally for the first time. H1: ‘paternalistic’ male stereotypes (e.g. Househusband) would be more effective than ‘envious’ male stereotypes (e.g. Businessman) across countries confirming the stereotype content model (SCM). H2: the match between initial male gender role attitudes and advertisement type would increase advertisement effectiveness only in countries with relatively low egalitarian norms (i.e. Poland and South Africa). A cross-national study was conducted through the use of student samples following a 3(country: United Kingdom, Poland and South Africa) × 2(advertisement type) × (gender attitude) mixed design (N = 373). A three-way multivariate analysis of variance showed support for H1 and partial support for H2 (i.e. the second hypothesis held on purchase intent and for South Africa). The study provides evidence for the cross-national applicability of the SCM to advertising and the limited predictive value of gender attitudes for purchase intent depending on country. Thus, contrary to mainstream advertising practices, breaking male gender stereotypes does appear to pay cross-nationally. Theoretical and practical implications alongside the potential for change in practices are discussed.

Keywords:

• Advertising
• gender attitudes
• gender portrayal
• gender stereotypes
• sex roles
• cross-cultural

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This work was supported by the Interinstitutional funds.

Notes

1. A comparison of the key demographics is reported in Zawisza et al. (2012) – a subsample of the data is used here.

2. Attractiveness was tested in a separate pilot study (n = 18 students from a high school in Marlow, London: 6 men and 12 women, averaging 17 years old, the majority of whom were British – 83.3%). They evaluated the adult models in the photographs on a seven-point semantic differential scales (–3 = very unattractive and 3 = very attractive). An independent t test showed that the two sets of pictures did not differ in attractiveness: M _Tr = .9, SD = 1.21 vs. M _nTr = 1.63, SD = 1.36, t(14) = 1.59, p = .135, d = −.568.

3. The traditional Bm1 advertisement depicted a smartly dressed man in his early 30s in an outdoor setting, standing on a busy city centre pavement and talking on a mobile phone. The Bm2 advertisement depicted another smartly dressed man in his early 30s in an office setting, standing next to his desk with a laptop on it, holding a newspaper and talking on a mobile phone. The non-traditional Hh1 advertisement portrayed a casually dressed man in his mid-30s in a home setting, standing next to an ironing board with a pile of clothes on it, and ironing a T-shirt. The non-traditional Hh2 advertisement portrayed the same man, in the same setting, performing the same activities. In this instance, however, he was also holding a newborn baby. Other features of the advertisements were kept constant: in all cases the men were depicted from their waist up, they were smiling, and were presented in frontal view. The heading variably read: ‘Fathers/Professional men agree: until you try new X/Y orange juice you will never know what a real orange juice tastes like’. The product (i.e. a glass of orange juice surrounded by sliced oranges) was positioned in the middle right section of the advertisements.

4. The pilot study showed that the orange juice received a mean score of −.78 and a modal score of 0 on the Product Gender Scale [anchored: ‘feminine’ (−3) and ‘masculine’ (3)]. It also received a mean score of 2.56 and modal score of 2 on the Product Involvement Scale [anchored: ‘product requires little thought when purchasing’ (1) vs. ‘a lot of thought’ (7)].

5. All effect size calculations for paired t tests reported here were corrected for dependence between means for paired t tests, using Morris and DeShon’s (2002) equation 8 (calculator available at http://www.cognitiveflexibility.org/effectsize/).