Participants
Participants were recruited from an online research panel by Qualtrics, a survey company that used stratified quota sampling to gather a diverse sample of the United States. All participants volunteer for an opt-in panel with the intent to participate in research surveys, but the survey invitation did not include specific details about the nature of the study to avoid self-selection bias. All surveys included in the online research panel are offered equally to potential participants, and no one survey is advertised over another. Data were derived from a larger study in which the purpose was to test the usefulness of several measurement instruments. Eligibility requirements included living in the United States and being 18 years of age or older. Data was collected between April 2020 and May 2020, and consent was obtained from all participants. All study procedures were sanctioned by an institution review board (IRB) at a university located in the southern area of the United States.
Three-thousand seven-hundred fifty individuals expressed interest through the online research portal, of which 1,987 were excluded. Qualtrics is unable to report on the exact number of people who saw a survey but did not partake, but estimates that the number of people invited to complete a survey is nine to 10 times the amount of people who complete the survey. Thus, it can be estimated that approximately 33,750 to 37,500 people were invited to participate in the study. Exclusions were based on failing to meet eligibility requirements, refusals, and missing data. The final analytic sample consisted of 1,763 participants. The average age was 48.07 years old (SD = 17.1) and the gender breakdown of this sample consisted of 50.9% women, 47.9% men, and 1.2% “other”. Further demographic information for this sample can be found in Table 1. Participants were compensated $4.80 for their participation.
Measures
Plant-Based Diet Assessment. To assess whether participants followed a plant-based diet, they were asked to endorse the best option reflecting their diet among three choices: (1) “I exclusively eat a plant-based diet (no animal foods or ingredients)”; (2) “I eat a vegetarian diet that includes eggs and/or dairy but no meat”; (3) “I eat a diet that includes meat”. Endorsements on this measure were collapsed to reflect a dichotomous plant-based (item 1)/not plant-based (items 2 and 3) variable.
International Trauma Exposure Measure (ITEM; Hyland et al., 2021). This measure consists of 22 items that assess traumatic life events across three developmental periods: childhood, adolescence, and adulthood. Example items from the ITEM include “You were diagnosed with a life threating illness,” “Someone close to you died in an awful manner,” “You were physically assaulted by a parent or guardian,” and “You were exposed to a natural disaster where your life was in danger.” Psychologically threatening events were excluded from the measure. An additional two items were added to assess a broader range of trauma exposure: “You were made fun of, picked on, pushed, shoved, hit, or threatened with harm because of your race or religion, or ethnicity” and “You were discriminated against, treated with disrespect, called names, heard negative comments because of your race or religion or ethnic group.” This resulted in a total of 19 items. Participants determined if they experienced each event “up to the age of 12,” “between ages 13-18,” and/or “after the age of 18.” Only events that occurred during childhood were examined for the purposes of this study. Childhood trauma was scored by summing the events that occurred “up to the age of 12.” Responses to the open-ended item “Any other event not listed (please specify)” were not interpreted, so they were not included in the sum scores.
(Modified) Extended-Hurt, Insult, Threaten, Scream (E-HITS; Chan et al., 2010; Modified E-HITS; Portnoy et al., 2018). The E-HITS and Modified E-HITS consist of 5 items that assess relationship aggression perpetration (Modified E-HITS) and 5 items that assess relationship aggression victimization (E-HITS) in the last six months. Items include “Screamed or cursed at partner,” “Insulted or talked down to partner,” “Threatened partner with harm,” “Physically hurt partner” and “Forced or pressured partner to have sexual contact against their will or when they were unable to say no.” Participants determined the frequency of each behavior by responding on a Likert scale ranging from 1 (Never) to 5 (Frequently). Perpetration and victimization items were dichotomized into 1 (occurred in the last 6 months) or 0 (did not occur in the last 6 months) and then summed to represent a count of the behaviors (ranging from 0-5 for each score). The E-HITS has exhibited strong psychometric properties, including good concurrent and discriminant validity, internal consistency, and reliability (Chan et al., 2010.) In a study examining the accuracy and acceptability of the Modified E-HITS in assessing IPV perpetration in comparison to the CTS-2, the gold-standard for assessing IPV perpetration, the measure was deemed a promising screening tool for IPV perpetration (Portnoy et al., 2018).
Revised Conflict Tactics Scale Short Form (CTS2S; Straus & Douglas, 2004; Straus et al, 1996). The CTS2S consists of five subscales: assault, injury, psychological aggression, sexual coercion, and negotiation. Each subscale consists of 2 behaviors, which are repeated to assess the participants and their partners’ behaviors in the past six months. Only the assault and psychological aggression subscales were examined for the purposes of this study, resulting in a total score made up of 4 items assessing perpetration behaviors and 4 items assessing victimization behaviors. Participants determined the frequency of each behavior by responding on a Likert scale that was modified from 1 (This has never happened) to 6 (More than 20 times in the past year) to range from 1 (Never) to 5 (Frequently), with an option to select “This has happened before but not in the last 6 months.” Scores were recoded to reflect perpetration that occurred within the last six months such that 1 = occurred in the last 6 months and 0 = did not occur in the last 6 months. Next, total relationship aggression perpetration and victimization in the last six months was scored by summing the count of the behaviors (ranging from 0-4 for each score). The CTS2S has demonstrated concurrent and construct validity similar to that of the CTS2, which has exceptional construct validity, content validity, internal consistency, and reliability (Straus and Douglas, 2004; Straus et al., 1996; Newton et al., 2001).
Data Analysis
Analyses were conducted with IBM SPSS for Windows, Version 27.0. Skewness and kurtosis of the dependent variables were within normal limits (i.e., less than +/- 3 and +/-10 respectively; Brown, 2006) and there was no evidence of multicollinearity among predictor variables (all r’s < .70). Given that less than 5% of data were missing across all variables, pairwise deletion was used to incorporate all available data. We conducted hierarchical linear multivariate regressions to examine associations between diet and relationship aggression perpetration. We ran separate regressions for EHITS perpetration and CTS-2 perpetration. Specifically, Step 1 included diet (plant-based vs. non-plant based) and demographic factors (participant age, participant sex). Step 2 included victimization count scores (either EHITS victimization or CTS-2 victimization) and childhood trauma frequency (continuous).