Outcome variable. The 12-item version of the General Health Questionnaire (GHQ-12) [48 ], adapted into and validated for Spanish by Rocha et al. [49 ], was used to examine mental health. The purpose of this screening instrument is to detect psychological morbidity and general mental health problems among the general population [50 (link)]. It is made up of 12 items that are answered using a Likert-type scale ranging from 0 to 3 for each response option. Following Rocha et al. [49 ], scores equal to or higher than 3 meet the criterion for a potential case of impaired mental health. It is frequently used to analyze the mental health of homelessness [31 (link),51 (link),52 (link)].
An exploratory factor analysis was conducted to assess the unidimensionality of the GHQ-12 [53 (link)]. The parallel analysis suggested a single factor with good statistical values: KMO = 0.85 and UniCo = 0.95. GHQ-12 shows adequate reliability, with McDonald’s ordinal Omega = 0.86 and Standardized Cronbach’s Alpha = 0.86, revealing high internal consistency.
Exposure variables. The questions included in the Spanish Survey on Homelessness were used to measure the exposure variables (perceived discrimination and types of violence suffered) [3 ]:

Perceived discrimination. PSH were asked if they had ever felt discriminated against due to being in a situation of homelessness. The following response options were included: “yes, I have felt discriminated against”; “no, I have not felt discriminated against”; and “don’t know/no response”.

Types of violence suffered. PSH were asked if they had been the victim of any criminal offence while in a situation of homelessness. The following response categories were offered: “No, I have not suffered violence”; “I have suffered physical violence”; “I have suffered some form of sexual assault”; “I have suffered verbal assaults (insults and/or threats); and “don’t know/no response”. This question was recoded into three response categories: “No, I have not suffered violence”; “I have suffered violence” and “Don’t know/no response”. The aim was to obtain data on people who have suffered from violence and people who have not suffered from violence.

Control variables. The study includes self-rated health and sociodemographic variables as control variables:

Self-rated health. This is a significant variable insofar as it affects living conditions and health, including mental health, among PSH [54 (link)]. This variable was examined using a single question: “how would you describe your current state of health?” Five response options were offered, ranging from “very poor” (1) to “very good” (5). The variable was dichotomized into good self-perceived health (very good and good) and poor self-perceived health (average, poor, and very poor). This is a widely used way of asking about self-perceived general states of health [55 (link)].

Sociodemographic variables. Variables that are commonly used with PSH were included [24 (link)]: sex (male/female); nationality (Spanish/Latin American/African/European); age (quantitative variable, recodified into the following ranges: 35 years or younger; 36–50 years; 51 years or older); and housing situation (literally homeless/broadly homeless).

Free full text: Click here