Intersectional analysis of men’s masculinities and mental health-seeking behavior: A novel application of multiple correspondence analysis Supplementary Material Relational Analysis of Masculinity 2 SECTION A Original survey questions from the FiveThirtyEight.com masculinity_survey Table A.1 Original 30 Questions Used in the Masculinity Survey (Mehta, 2018) ID QUESTION QUESTION TYPE INCLUDED IN RAW DATA INCLUDED IN MCA ANALYSIS Section 1: Ideas about masculinity Q1 In general, how masculine or “manly” do you feel? Ordinal (4 pt. Likert-type) Y Y (active) Q2 How important is it to you that others see you as masculine? Ordinal (4 pt. Likert-type) Y Y (active) Q3 What ONE quality or hobby do you most associate with being masculine? Open-end qualitative N N Q4 Where have you gotten your ideas about what it means to be a good man? Nominal Y Y (active) Q5 Do you think that society puts pressure on men in a way that is unhealthy or bad for them? Nominal (binary) Y Y (active) Q6 How do you think society puts pressure on men in a way that is unhealthy or bad? (if Q5=yes) Open-end qualitative N N Section 2: Lifestyle Q7 How often would you say you do each of the following? (11 items) Matrix (5 pt. Likert-type) Y Y (active) Q8 Which of the following do you worry about on a daily or near daily basis? (12 items) Matrix (nominal binary) Y Y (active) Q9 Which of the following categories best describes your employment status? Nominal Y Y (supp) Section 3: Workplace Questions (if employed) Q10 In which of the following ways would you say it’s an advantage to be a man at your work right now? Nominal Y N Q11 In which of the following ways would you say it’s a disadvantage to be a man at your work right now? Nominal Y N Q12 Have you seen or heard of a sexual harassment incident at your work? If so, how did you respond? Nominal Y N Q13 And which of the following is the main reason you did not respond (if Q12=”did not respond at all”) Nominal Y N Section 4: #MeToo movement (if employed) Q14 How much have you heard about the #MeToo movement? Nominal Y N Q15 As a man, would you say that you think about your behavior at work differently in the wake Nominal (binary) Y N Relational Analysis of Masculinity 3 of #MeToo (if heard about #MeToo) Q16 How do you think about your behavior differently in the wake of #MeToo? (if heard about #MeToo) Open-end qualitative N N Section 5: Relationships Q17 Do you typically feel as though you’re expected to make the first move in romantic relationships Nominal (binary) Y Y (active) Q18 How often do you try to be the one who pays when on a date? Ordinal (5pt. Likert type) Y Y (active) Q19 Which of the following are reasons why you try to pay when on a date? (if Q18 = “always” or “often”) Nominal Y N Q20 When you want to be physically intimate with someone, how do you gauge their interest? Nominal Y N Q21 Over the past 12 months, when it comes to sexual boundaries, which of the following things have you done? Nominal Y N Q22 Have you changed your behavior in romantic relationships in the wake of #MeToo movement? Nominal (binary) Y N Q23 How have you changed your behavior in romantic relationships in the wake of the #MeToo movement? Open-end qualitative N N Section 6: Demographics Q24 Are you now married, widowed, divorced, separated, or have you never been married? Nominal Y Y (supp) Q25 Do you have any children? Nominal (binary) Y Y (supp) Q26 Would you describe your sexual orientation as: Nominal Y Y (supp) Q27 What is your age? Open-end quantitative reported in 4 bins: (<35, 35-64, 65+, “prefer not to answer”) Y Y (supp) Q28 Are you: [racial identity] Nominal Y Y (supp) Q29 What is the last grade of school you completed? Ordinal Y Y (supp) Q30 What state do you live in? Nominal Y Y (supp) Legend: Y = “Yes”; N = “No”; “active” = used as an active variable in MCA analysis; “supp” = used as a supplementary variable in MCA analysis Source: Mehta, Dhrumil. (2018). “Masculinity Survey.” Accessed October 17, 2019. https://github.com/fivethirtyeight/data/blob/master/masculinity-survey/ https://github.com/fivethirtyeight/data/blob/master/masculinity-survey/ Relational Analysis of Masculinity 4 SECTION B Our method for data cleaning is outlined in Figure B1. After importing the raw data into the R environment, we observed 98 variables: (a) 85 variables representing the 26 survey questions and their associated sub-variables, (b) 7 variables engineered from the original 26 survey questions by FiveThirtyEight for the purposes of their analysis, and (c) 6 variables representing metadata generated by the survey administrator. All 7 engineered variables were dropped, as were 5 of the metadata variables (start and end date, user device, respondent ID, and region). The sixth metadata variable representing respondents’ income level was included in the analysis. We next excluded 30 variables that were only shown to a subset of respondents (Q10- 15; Q19), 11 variables that were strictly focused on respondents’ sexual behavior in the context of the #MeToo movement (Q20-22), and 1 redundant variable representing the level “None of the above” from Q8 (the other 11 levels were included in the analysis and the indication of “None reported” for these variables by definition indicated “None of the above”). After excluding these variables, 44 variables remained representing 16 survey questions (Q1-2, Q4-5, Q7-9, Q17-18, Q24-30) and 1 metadata variable (Income). Our next step was to engineer 6 features from the intermediate dataset: (a) the 6 variables relating to Q4 (origin of respondents’ ideas of masculinity) were recoded into a single variable, MascOrigin; (b) the 11 variables relating to Q8 (daily or near daily worry about [feature]) were recoded into 4 variables (HealthWorry, AppWorry, FinWorry, and SexWorry); and (c) the 3 variables related to Q25 (whether the respondent has children) were recoded into a single variable, Children. Relational Analysis of Masculinity 5 MasculinitySurvey Raw Dataset: 98 variables • 26 survey questions • 7 engineered variables • 6 metadata variables Intermediate Dataset: 44 variables • 16 survey questions • 1 metadata variables Cleaned Dataset: 30 variables • 16 survey questions • 1 metadata variables Feature Engineering Data Exclusion • 7engineered variables • 5 metadata variables • 30 variables shown only to a subset of respondents • 11 variables strictly related to the #MeToo movement • 1 redundant variable • CreatedMascOrigin from the 6 variables related to Q4 • Created Children from the 3 variables related to Q25 • Created HealthWorry, AppWorry, FinWorry, and SexWorry from the 11 variables related to Q8 Figure B1 – Data cleaning workflow Relational Analysis of Masculinity 6 SECTION C Treatment of missing data The presence of missing data is ubiquitous in survey data, and one that cannot be ignored. The treatment of missing data (whether affected cases are deleted, or imputed with some estimator) can have substantial and negative impacts on the quality of an analysis potentially introducing bias or reducing the power of the results depending on the technique used. There is no one panacea for handling missing data, however, and researchers must select the best technique based on the quantity and pattern of missingness, and the nature of the affected variables (e.g. continuous, ordered categorical, unordered categorical, etc.) (Shlomer et al., 2010; Allison, 2001). There were 466 missing responses across 21 variables in the intermediary data set, representing a 16.6% of participants with some information loss (Table C.1). Initial visualizations of the missingness revealed apparent “randomness” in the data, with only 18 individuals skipping more than 3 questions. Through the application of Little’s MCAR test (Little, 1988) we were unable to reject the null hypothesis (χ2 = 1,966.73, df = 1038, p = 0.3192) and so could not discount the possibility that the data was missing completely at random. A data set with cases MCAR is unlikely in practice (Shlomer et al. 2010). However, based on the sparseness of the missing responses (466 out of the 33,915 complete set of responses, representing 1.37%) and unobservable patterns (see Figure C.1), we concluded that the data was MCAR for the purposes of treatment. There are many methods for treating missing cases discussed in the literature. Methods include listwise deletion (deletion of cases), pairwise deletion (deletion in situ), single-value imputation, and stochastic imputation. Treating missing cases in purely categorical datasets, however, (and particularly unordered categories) presents many challenges (Shlomer, et al. Relational Analysis of Masculinity 7 2010). Various methods have been discussed and compared in the literature including listwise deletion, Bayesian imputation, and various multiple imputation techniques (e.g. Graham, 2009; Allison, 2001; Chen & Åstebro, 2003). A recent simulation has compared three multiple imputation techniques for missing cases in fully categorical datasets (Akande et al., 2017, Breiman et al., 1984). Akande et al. found that a CART approach (classification and regression trees) proved to be a superior model compared to Dirichlet process mixture of products of multinomial distributions (DPMPM) and generalized linear model GLM approaches. Multiple imputation with CART can be readily applied in R using the mice package (Van Buuren, & Groothuis-Oudshoorn, 2011) and has the benefit that it can treat several multiple variables simultaneously, including variables of different data types (e.g. binary variables treated with logistic regression trees, and unordered categorical variables with a classification algorithm). Table C.1 Frequency statistics for missing data (generated in naniar; Tierney & Cook, 2020) Variable-level summary Variable name Number missing Percent missing Cry 91 5.63% WorkOut 41 2.54% PayDate 41 2.54% FSex 35 2.17% Therapy 35 2.17% Rom.Lead 31 1.92% Lonely 28 1.73% MSex 21 1.30% AdvPers 19 1.18% AdvProf 16 0.99% MascFeel 14 0.87% MAffect 14 0.87% Sports 14 0.87% Pressure 13 0.81% Orientation 13 0.81% MascApp 9 0.56% Fight 9 0.56% Children 9 0.56% Relational Analysis of Masculinity 8 Marital 8 0.50% Employ 3 0.19% Income 2 0.12% No missing cases: MascOrigin, HealthWorry, FinWorry, AppWorry, SexWorry, Education, Race, Age, Region Observation-level summary Number of missing responses Number of participants Percent of participants 0 1,347 83.4% 1 199 12.3% 2 33 2.04% 3 18 1.11% 4 4 0.25% 5 1 0.06% 6 4 0.25% 7 2 0.12% 9 1 0.06% 10 2 0.12% 11 1 0.06% 14 2 0.12% 20 1 0.06% Relational Analysis of Masculinity 9 Figure C.1. Patterns of Missing Data using vis_miss() in naniar (Tierney & Cook, 2020) Additional References Allison, Paul D. (2001). Missing Data. SAGE Publications: Thousand Oaks, CA. Breiman, L., Friedman, J., Stone, C. J., & Olshen, R. A. (1984). Classification and regression trees. Chapman & Hall/CRC Press: Boca Raton, FL. Chen, G., & Åstebro, T. (2003). How to deal with missing categorical data: Test of a simple Bayesian method. Organizational Research Methods, 6, 309-327. DOI: 10.1177/1094428103254672. Graham, J. W. (2009). Missing data analysis: Making it work in the real world. Annual Review of Psychology, 60: 549-576. DOI:10.1146/annurev.psych.58.110405.085530. Schlomer, G. L., Bauman, S., & Card, N. A. (2010). Best practices for missing data management in counseling psychology. Journal of Counseling Psychology, 57(1), 1-10. DOI: 10.1037/a0018082. Relational Analysis of Masculinity 10 Tierney, N., & Cook, D. (2020). Expanding tidy data principles to facilitate missing data exploration, visualization and assessment of imputations. Journal of Statistical Software, arXiv:1809.02264v2. DOI :10.18637/jss.v00.i00. Relational Analysis of Masculinity 11 SECTION D Descriptive statistics for the 21 active and 9 supplementary variables included in the MCA. TABLE D.1 Descriptive Statistics and Frequency Distributions Active Variables Category Count Percentage Category Count Percentage MascFeel: how masc. a respondent feels HealthWorry: health-related worry m.feel_high 619 38.33% h.worry_ment 97 6.01% m.feel_med 831 51.46% h.worry_phys 544 33.68% m.feel_low 133 8.24% h.worry_both 291 18.02% m.feel_no 32 1.98% h.worry_nr 683 42.29% MascApp: importance of appearing masc. FinWorry: finance-related worry m.app_high 197 12.20% f.worry_personal 388 24.02% m.app_med 630 39.01% f.worry_providing 63 3.90% m.app_low 546 33.81% f.worry_both 381 23.59% m.app_no 242 14.98% f.worry_nr 783 48.48% MascOrigin: origin of masculine ideals AppWorry: appearance-related worry o_family 675 41.80% a.worry_phys 701 43.41% o_society 286 17.71% a.worry_style 50 3.10% o_both 640 39.63% a.worry_both 287 17.77% o_none 14 0.87% a.worry_nr 577 35.73% Rom.Lead: take lead in romantic situations SexWorry: sex-related worry r.lead_yes 1,035 64.09% s.worry_perf 254 15.73% r.lead_no 580 35.91% s.worry_gen 48 2.97% s.worry_both 100 6.19% AdvProf: seek professional advise from friends s.worry_nr 1,213 75.11% Often 202 12.51% Pressure: unhealthy pressure on men? Sometimes 762 47.18% press_yes 963 59.63% Rarely 423 26.19% press_no 652 40.37% Never, but open to it 158 9.78% Never, not open to it 70 4.33% Relational Analysis of Masculinity 12 AdvPers: seek personal advise from friends MAffect: express affection to male friends Often 198 12.26% Often 228 14.12% Sometimes 705 43.65% Sometimes 499 30.90% Rarely 518 32.07% Rarely 459 28.42% Never, but open to it 124 7.68% Never, but open to it 125 7.74% Never, not open to it 70 4.33% Never, not open to it 304 18.82% Cry: cry Fight: engage in physical fights Often 69 4.27% Often 11 0.68% Sometimes 553 32.24% Sometimes 32 1.98% Rarely 728 45.08% Rarely 324 20.06% Never, but open to it 160 9.91% Never, but open to it 317 19.63% Never, not open to it 105 6.50% Never, not open to it 931 57.65% FSex: sexual relations with women MSex: sexual relations with men Often 623 38.58% Often 60 3.72% Sometimes 514 31.83% Sometimes 56 3.47% Rarely 212 13.13% Rarely 42 2.60% Never, but open to it 125 7.74% Never, but open to it 82 5.08% Never, not open to it 141 8.73% Never, not open to it 1,375 85.14% Sports: watch sports WorkOut: work out Often 663 41.05% Often 415 25.70% Sometimes 477 29.54% Sometimes 511 31.64% Rarely 287 17.77% Rarely 443 27.43% Never, but open to it 62 3.84% Never, but open to it 135 8.36% Never, not open to it 126 7.80% Never, not open to it 111 6.87% Therapy: see a therapist Lonely: feel lonely/isolated Often 78 4.83% Often 182 11.27% Sometimes 141 8.73% Sometimes 468 28.98% Rarely 272 16.84% Rarely 627 38.82% Never, but open to it 652 40.37% Never, but open to it 94 5.82% Never, not open to it 472 29.23% Never, not open to it 244 15.11% PayDate: pay for a date Always 813 50.34% Often 470 29.10% Sometimes 230 14.24% Rarely 24 1.49% Never 78 4.83% Relational Analysis of Masculinity 13 Supplementary Variables Category Count Percentage Category Count Percentage Employ: employment status Age: respondents’ age group emp_ft 738 45.70% a.18-34 133 8.24% emp_pt 144 8.85% a.35-64 855 52.94% emp_u.look 68 4.21% a.65+ 627 38.82% emp_u.nolook 110 6.81% Race: respondents’ stated race/ethnicity emp_u.ret 525 32.57% race_asian 36 2.23% emp_u.stud 30 1.86% race_black 72 4.46% Education: educational attainment race_hisp 71 4.40% ed_no.hs 21 1.30% race_other 85 5.26% ed_hs 157 9.72% race_white 1,351 83.65% ed_p.ugrad 292 18.08% Marital: respondents’ marital status ed_assoc 148 9.16% m_div 220 13.62% ed_ugrad 515 31.89% m_mar 1,000 61.80% ed_pgrad 482 29.85% m_nev 286 17.77% Income: respondents’ income level m_sep 25 1.55% inc_high 489 30.28% m_wid 84 5.26% inc_mid 493 30.65% Region: respondents’ geographic location inc_low 421 26.01% reg_n.atlantic 76 4.71% inc_nr 212 13.07% reg_mid.atlantic 255 15.79% Children: does respondent have children reg_s.atlantic 276 17.09% child_yes 1,069 66.19% reg_ne.central 236 14.61% child_no 546 33.81% reg_se.central 58 3.59% Orientation: sexual orientation reg_nw.central 108 6.69% so_bi 52 3.22% reg_sw.central 157 9.72% so_gay 112 6.93% reg_mountain 161 9.97% so_other 31 1.98% reg_west 288 17.83% so_straight 1,420 87.86% Relational Analysis of Masculinity 14 SECTION E Additional graphs and figures related to the MCA. FIGURE E.1: Scree plot describing the % of inertia (variance) explained in the first 10 dimensions Note: Color to be included online only Relational Analysis of Masculinity 15 TABLE E.1 % of Inertia (Variance) Explained in First 10 Dimensions Dimension Eigenvalue % Variance Cumulative % Variance Benzécri’s Modified Rate (%) Cumulative Modified Rate (%) 1 0.16883 5.0% 5.0% 44.31% 44.31% 2 0.137716 4.1% 9.1% 24.48% 68.79% 3 0.106528 3.2% 12.2% 10.47% 79.26% 4 0.098985 2.9% 15.1% 7.96% 87.22% 5 0.085005 2.5% 17.7% 4.22% 91.43% 6 0.071354 2.1% 19.8% 1.70% 93.13% 7 0.067178 2.0% 21.8% 1.15% 94.29% 8 0.066659 2.0% 23.7% 1.09% 95.38% 9 0.064569 1.9% 25.6% 0.87% 96.25% 10 0.064444 1.9% 27.5% 0.85% 97.10% Note: 61 dimensions not shown. FIGURE E.2: Contribution of variable categories to dimensions 1 and 2 with average contribution line Note: Color to be included online only TABLE E.2 Relational Analysis of Masculinity 16 Contribution of Each Variable Level in Dimensions 1 and 2 Variable Level Dimension 1 Dimension 2 a.worry_both 1.65%* 1.04% a.worry_nr 3.61%* 0.38% a.worry_phys 0.74% 0.05% a.worry_style 0.03% 0.22% AdvPers_Never_not_open 8.62%* 1.60%* AdvPers_Never_open 1.54%* 0.05% AdvPers_Often 1.53%* 1.95%* AdvPers_Rarely 0.03% 0.86% AdvPers_Sometimes 0.90% 0.19% AdvProf_Never_not_open 8.37%* 1.47%* AdvProf_Never_open 0.91% 0.31% AdvProf_Often 1.36%* 1.43%* AdvProf_Rarely 0.12% 0.47% AdvProf_Sometimes 0.97% 0.52% Cry_Never_not_open 5.99%* 0.17% Cry_Never_open 0.26% 0.03% Cry_Often 0.46% 3.70%* Cry_Rarely 0.00% 1.61%* Cry_Sometimes 1.16% 0.47% f.worry_both 1.82%* 0.08% f.worry_nr 1.58%* 0.35% f.worry_personal 0.27% 0.57% f.worry_providing 0.03% 0.23% Fight_Never_not_open 0.39% 0.22% Fight_Never_open 0.12% 0.44% Fight_Often 0.01% 0.38% Fight_Rarely 0.76% 0.15% Fight_Sometimes 0.18% 0.19% FSex_Never_not_open 0.50% 6.69%* FSex_Never_open 0.06% 2.48%* FSex_Often 0.03% 1.82%* FSex_Rarely 0.00% 0.36% FSex_Sometimes 0.09% 1.08% h.worry_both 1.85%* 2.67%* h.worry_ment 0.38% 1.58%* h.worry_nr 2.40%* 0.55% h.worry_phys 0.23% 0.80% Lonely_Never_not_open 6.38%* 0.10% Lonely_Never_open 0.26% 0.13% Lonely_Often 0.96% 5.51%* Lonely_Rarely 0.15% 2.28%* Relational Analysis of Masculinity 17 Lonely_Sometimes 1.00% 0.47% m.app_high 0.13% 0.04% m.app_low 0.03% 0.00% m.app_med 0.54% 0.49% m.app_no 1.29%* 1.97%* m.feel_high 0.30% 1.49%* m.feel_low 0.06% 2.98%* m.feel_med 0.62% 0.00% m.feel_no 1.20%* 2.94%* MAffect_Never_not_open 5.30%* 0.05% MAffect_Never_open 0.11% 0.08% MAffect_Often 0.70% 1.29%* MAffect_Rarely 0.24% 0.68% MAffect_Sometimes 0.86% 0.00% MSex_Never_not_open 0.11% 1.58%* MSex_Never_open 0.11% 1.38%* MSex_Often 0.34% 4.26%* MSex_Rarely 0.09% 1.80%* MSex_Sometimes 0.15% 2.15%* o_both 1.15%* 0.00% o_family 0.37% 0.93% o_none 0.98% 0.23% o_society 0.21% 1.61%* PayDate_Always 0.01% 1.13%* PayDate_Never 3.62%* 1.40%* PayDate_Often 0.53% 0.03% PayDate_Rarely 0.04% 0.53% PayDate_Sometimes 0.08% 1.73%* press_no 1.17%* 1.57%* press_yes 0.79% 1.06% r.lead_no 1.39%* 0.20% r.lead_yes 0.78% 0.11% s.worry_both 0.99% 1.55%* s.worry_gen 0.22% 0.61% s.worry_nr 0.74% 0.69% s.worry_perf 1.11%* 0.48% Sports_Never_not_open 2.06%* 3.73%* Sports_Never_open 0.13% 0.79% Sports_Often 0.09% 1.57%* Sports_Rarely 0.07% 1.12%* Sports_Sometimes 0.10% 0.43% Therapy_Never_not_open 5.73%* 0.10% Therapy_Never_open 0.46% 0.61% Therapy_Often 0.72% 2.25%* Relational Analysis of Masculinity 18 Therapy_Rarely 0.92% 0.02% Therapy_Sometimes 0.90% 1.81%* WorkOut_Never_not_open 4.12%* 2.87%* WorkOut_Never_open 0.07% 1.22%* WorkOut_Often 0.10% 0.45% WorkOut_Rarely 0.04% 0.02% WorkOut_Sometimes 0.38% 0.37% Note: Variable levels with a contribution greater than the average for each dimension denoted with an asterisk (‘*’) FIGURE E.3: Correlation between variables and dimensions 1 and 2 Note: Color to be included online only FIGURE E.4: Relational Analysis of Masculinity 19 Plots of active and supplementary variables in Dimensions 1 & 2 *Supplementary categories related to region excluded as no clear trend emerged and inclusion made visibility of other categories challenging. Note: Color to be included online only FIGURE E.5: Relational Analysis of Masculinity 20 Hierarchical cluster tree Note: Color to be included online only FIGURE E.6: Bar Plot representing cluster size Note: Color to be included online only Relational Analysis of Masculinity 21 SECTION F Additional details on the defining features of the 6 clusters. Table F.1 Defining features of the six clusters Beliefs Behaviors Demographics Cluster 1: (n = 335, 20.7%) MascFeel: Low (11.9%) AdvPers: Sometimes (54.9%) Employ: Full time (51.3%) Medium (67.2%) FSex: Rarely (21.8%) Marital: Divorced (17.9%) MascApp: No (4.5%) MSex: Open (7.8%) Medium (54.3%) Therapy: Often (10.1%) Pressure: Yes (74.3%) Sometimes (14.3%) HealthWorry: Both (45.4%) Lonely: Often (27.2%) FinWorry: Both (41.8%) Sometimes (47.2%) None (24.5%) AppWorry: Both (38.2%) None (8.4%) SexWorry: Both (17.3%) Genitalia (8.1%) None (40.3%) Performance (34.3%) Cluster 2: (n = 185, 11.5%) MascFeel: High (58.4%) AdvProf: Often (75.7%) Employ: Full time (61.6%) MascApp: High (29.2%) AdvPers: Often (78.9%) HealthWorry: Mental (13.0%) MAffect: Often (39.5%) FinWorry: Both (41.1%) Cry: Often (10.3%) AppWorry: Both (25.9%) Fight: Often (3.8%) Style (5.9%) FSex: Often (65.4%) SexWorry: Both (9.7%) Sports: Often (51.9%) Therapy: Often (10.3%) Cluster 3: (n = 188, 11.5%) MFeel: Low (20.2%) AdvProf: Sometimes (59.0%) Employ: Unemployed, not looking (10.1%) MascApp: No (29.3%) AdvPers: Sometimes (59.0%) Unemployed, student (5.9%) MascOrigin: Society (31.4%) MAffect: Often (36.7%) Income: Low (39.9%) Pressure: Yes (85.1%) Sometimes (48.4%) Child: No (68.6%) Rom.Lead: No (56.9%) Cry: Often (16.0%) Marital: Never (45.2%) Relational Analysis of Masculinity 22 HealthWorry: Mental (12.2%) Sometimes (54.8%) Widowed (9.0%) FinWorry: Personal (37.2%) FSex: Never not open (48.4%) Orientation : Gay (47.3%) MSex: Never not open (29.3%) Bisexual (10.6%) Never but open (11.7%) Other (5.3%) Often (26.6%) Age: 18-34 (16.5%) Rarely (12.8%) Sometimes (19.7%) Sports: Never not open (18.6%) Never but open (9.6%) Often (13.8%) Rarely (34.6%) Workout: Never but open (14.9%) Therapy: Often (9.6%) Sometimes (21.3%) Lonely: Often (14.9%) Sometimes (45.7%) PayDate: Always (13.3%) Sometimes (41.0%) Cluster 4: (n = 655, 40.6%) MascApp: No (9.6%) AdvProf: Sometimes (61.2%) Marital: Never (8.2%) HealthWorry: None (54.8%) AdvPers: Sometimes (61.2%) FinWorry: None (60.2%) Cry: Rarely (56.3%) AppWorry: Both (8.4%) Lonely: Rarely (55.9%) Cluster 5: (n = 193, 12.0%) MascApp: No (34.2%) AdvProf: Never but open (48.7%) Marital: Widowed (9.8%) HealthWorry: None (50.3%) Sometimes (11.9%) Age: 65+ (52.8%) FinWorry: None (60.1%) AdvPers: Never but open (47.2%) AppWorry: None (52.3%) Sometimes (11.9%) MAffect: Never not open (26.4%) Never but open (34.7%) Cry: Never but open (38.3%) Fight: Rarely (7.3%) FSex: Never but open (22.2%) MSex: Never but open (7.8%) Sports: Never not open (15.0%) Never but open (12.4%) WorkOut: Never not open (14.5%) Never but open (24.4%) Therapy: Never not open (44.6%) Relational Analysis of Masculinity 23 Lonely: Never not open (23.8%) Never but open (18.1%) Cluster 6: (n = 59, 3.7%) MascFeel: High (52.5%) AdvProf: Never not open (81.4%) Employ: Unemployed, not looking (16.9%) MascApp: No (28.8%) AdvPers: Never not open (81.4%) Education: High school (25.4%) : High (32.2%) MAffect: Never not open (91.5%) MascOrigin: Both (11.9%) Cry: Never not open (64.4%) None (8.5%) Fight: Often (3.4%) Society (33.9%) FSex: Never not open (32.2%) Pressure: No (61.0%) Sports: Never not open (33.9%) Rom.Lead: No (59.3%) WorkOut: Never not open (49.2%) HealthWorry: None (79.7%) Sometimes (10.2%) FinWorry: None (71.2%) Therapy: Never not open (88.1%) AppWorry: None (66.1%) Lonely: Never not open (62.7%) PayDate: Never (27.1%) Often (10.2%) Relational Analysis of Masculinity 24 SECTION G Fisher’s Exact Tests validating distribution of categories for Cluster 6 are not due to random chance. Table G.1 Outcome of Fisher’s Exact Tests Variable Alternative Hypothesis Results MascFeel Two-sided p-value < 0.001 MascApp Two-sided p-value < 0.001 MascOrigin Two-sided p-value < 0.001 Pressure True odds ratio is not equal to 1 95% confidence interval: (1.32, 4.12) OR = 2.31 p-value = 0.002 Rom.Lead True odds ratio is not equal to 1 95% confidence interval: (1.49, 4.62) OR = 2.60 p-value < 0.001 h.worry Two-sided p-value < 0.001 f.worry Two-sided p-value < 0.001 a.worry Two-sided p-value < 0.001 AdvProf Two-sided p-value* < 0.001 AdvPers Two-sided p-value* < 0.001 MAffect Two-sided p-value* < 0.001 Cry Two-sided p-value* < 0.001 Fight Two-sided p-value < 0.001 FSex Two-sided p-value < 0.001 Sports Two-sided p-value < 0.001 WorkOut Two-sided p-value* <0.001 Therapy Two-sided p-value < 0.001 Lonely Two-sided p-value < 0.001 PayDate Two-sided p-value < 0.001 Employ Two-sided p-value = 0.040 Education Two-sided p-value = 0.006 *Due to small size of certain categories, Fisher’s Exact Test was run with simulated p-value based on 2,000 replicates.