Gender differences in loneliness and isolation among older people: a mixed methods study into older men's propensity to acknowledge loneliness

Introduction
Loneliness is a problem 'particularly associated with old age, growing older, and later life' (Victor and Yang 2012, p85). Furthermore, studies have shown men to be relatively unwilling to acknowledge and/or seek help for a multitude of physical and psychological problems (Mackenzie et al 2007; Johnson et al 2012). Older men, then, may be particularly likely to feel lonely, but unlikely to acknowledge or seek-help for feelings of loneliness. Therefore, to assist policy-makers and service providers promote better well-being in older men, it is vital to investigate older men's propensity to acknowledge or seek help for loneliness, and the reasons behind their propensity.

Research Questions:
1. Are older men aggregately lonelier than data suggests?
2. Why are older men less likely to state they are lonely than older women?

Methodology
A mixed-methods study, following Creswell and Plano Clark's (2011) 'explanatory sequential design', will be employed. Specifically, this involves conducting an initial study, analysing it, then conducting a second study to elaborate on and 'explain' the results. In this context, this will consist of a survey, followed by semi-structured interviews, and is the best available methodology for three reasons. Firstly, a survey can test whether De Jong-Gierveld et al's (2016) hypothesis (that men indicate they are lonelier on a scale than in response to a direct question) is true within a single sample of people. Secondly, attitudinal statements, based on Ratcliffe's (2017) qualitative findings, can also be included in the survey, allowing for a broader, more generalisable investigation into the relationship between masculinities and loneliness than Ratcliffe's qualitative study (Bryman 2016). Lastly, conducting qualitative interviews with survey respondents can facilitate better contextualised and more meaningful conclusions than a questionnaire alone could, particularly in this context where the validity of statistical data is uncertain (Cresswell and Plano Clark 2011; De Jong-Gierveld et al 2016).

Survey
A cross-sectional study, of at least 100 men and 100 women, aged 65+, will be randomly sampled via some or all of the following organisations: the York NHS research and development (R&D) team; the West Yorkshire NHS R&D team; YorkCVS, an organisation that works with and promotes community services across York; and Age UK. Further detail on options for sampling and distribution is available in the 'proposed fieldwork' section. The survey will consist of four sections: demographics; a single, direct question on loneliness; a loneliness scale which was identified by De Jong-Gierveld et al (2016) as providing a smaller gender difference in loneliness than a direct question (specifically, the 20 item UCLA scale); and attitudinal statements on the extent to which loneliness is a subordinate social role, incompatible with strength of mind, caring, and responsibility. The data will be analysed in SPSS, and the key hypotheses are:

1. A man, with the same loneliness score as a woman, will be less likely to state he is lonely in response to the direct question. Specifically, in a regression, where UCLA score and gender are independent variables, and the direct question on loneliness is the dependent variable, men will show significant P value.
2. Men will be more likely to indicate that a) loneliness is a subordinate role, and that b) 'strength of mind', c) 'caring', and d) 'responsibility' are gendered requirements incompatible with the role.
3. The construction of loneliness as a subordinate role, and that 'strength of mind', 'caring', and 'responsibility' are incompatible with it, cause men's disinclination to state they are lonely in response to a direct question. As such, when the significant attitudinal statements are added to the regression used for hypothesis 1, gender will no longer be a significant variable.