7  Congress experience

How did people with different social identities experience the congress?

7.1 Discrimination and harassment

Respondents of the post-congress survey were asked if they themselves experienced discrimination and/or harassment (of any sort) at the congress and whether they reported it to the Awareness team, or if they witnessed someone else experiencing this.

A total of 11 respondents experienced some form of discrimination or harassment, of which 2 cases were reported to the Awareness team. A total of 3 respondents witnessed somebody else receiving some form of discrimination or harassment, of which 1 case was reported to the Awareness team. Reasons of not reporting the cases to the Awareness team included that the case was unrelated to EDI issues. Even though only a few cases were reported to the Awareness team, the qualitative feedback given in the survey highlighted that the presence of this team made some people feel safe.

Due to the low number of reports of discrimination or harassment, we cannot statistically test if some social identities experienced more discrimination/harassment than others. Nevertheless, these are the summary statistics per social identity for the answer to whether respondents experienced discrimination/harassment themselves:

dis <- subset(survey, discrimination == "Yes")

table(dis$gender, dis$lgbtq, dis$nationality_subcontinent)

, ,  = West Europe

            
             No Yes
  Female      3   3
  Male        1   0
  Non-binary  0   0

, ,  = North Europe

            
             No Yes
  Female      0   1
  Male        0   0
  Non-binary  0   0

, ,  = South Europe

            
             No Yes
  Female      0   0
  Male        0   0
  Non-binary  0   0

, ,  = East Europe

            
             No Yes
  Female      0   0
  Male        0   0
  Non-binary  0   0

, ,  = North America

            
             No Yes
  Female      1   0
  Male        0   0
  Non-binary  0   1

, ,  = Oceania

            
             No Yes
  Female      0   0
  Male        0   0
  Non-binary  0   0

, ,  = South American

            
             No Yes
  Female      0   0
  Male        0   0
  Non-binary  0   0

, ,  = Asia

            
             No Yes
  Female      0   0
  Male        0   0
  Non-binary  0   0

, ,  = Dual-nationality

            
             No Yes
  Female      0   0
  Male        0   0
  Non-binary  0   0

This shows that most reports came from women, and from West Europe.

7.2 Congress experience

In the post-congress survey, we asked respondents to answer on a 7-point Likert scale (1: Strongly disagree, 7: Strongly agree) how much they agree with the following three statements: 1) “I felt heard during the conversations I had, both during Q&A sessions and social activities”; 2) “I felt comfortable being myself”; 3) “Attending the Behaviour 2023 congress helped me feel like I belong in my research field”. For each of the three statements, we fitted ordinal GLMs to identify which social identity variables (gender, LGBTQIA+, nationality, affiliation, expat status) were significantly associated with the Likert-scale response to the statement. Additionally, we controlled for the level of comfort a person had speaking English as well as their self-reported level of expertise. Only the variables that were significant were used in the final model.

We have information on three variables that roughly measure the same thing: age is indicative of career stage, career stage is indicative of expertise, and self-reported expertise rating directly measures this expertise. Since age and career stage are indicator variables of expertise, we therefore only investigate expertise rating as this will give us the signal we are interested in.

7.2.1 Feeling heard

# first, we explore the distribution of answers

# feeling heard
ggplot(survey, aes(feeling_heard_rating)) + 
  geom_histogram(stat="count", aes(y=stat(count/sum(count)*100)), fill = clrs[11]) + 
  geom_text(aes(label = paste0("N = ", ..count..), y = stat(count/sum(count)*100)), 
            stat="count", vjust=-1) +
  ylim(0, 100) + 
  labs(x = "I felt heard", y = "% of responses") 

# next: explore the relationship between age and expertise, and career stage and expertise rating

coeftest(polr(as.factor(expertise_rating) ~ age, data = survey))

t test of coefficients:

         Estimate Std. Error t value  Pr(>|t|)    
age35-50  2.02180    0.22656  8.9239 < 2.2e-16 ***
age> 50   3.43812    0.38098  9.0244 < 2.2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

coeftest(polr(as.factor(expertise_rating) ~ career_3cat, data = survey))

t test of coefficients:

                 Estimate Std. Error t value  Pr(>|t|)    
career_3catMid    2.20513    0.23164  9.5196 < 2.2e-16 ***
career_3catLate   4.16000    0.36819 11.2985 < 2.2e-16 ***
career_3catOther  1.81962    0.53727  3.3868 0.0007817 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

# as expected!

# another check: is expertise rating affected by gender?

coeftest(polr(as.factor(expertise_rating) ~ gender, data = survey))

t test of coefficients:

                  Estimate Std. Error t value Pr(>|t|)
genderMale        0.074149   0.203076  0.3651   0.7152
genderNon-binary -0.208320   0.759624 -0.2742   0.7841

# answer = no

# then we test the effect of each of the social identity variables

# gender
survey$gender <- factor(survey$gender, levels = c("Male", "Female", "Non-binary"))

m_heard_gender_null <- polr(feeling_heard_rating ~ 1, data=subset(survey, !is.na(gender)))

m_heard_gender <- polr(feeling_heard_rating ~ gender,
                   data=survey) 

m_heard_gender_out <- collect_out(model = m_heard_gender, null = m_heard_gender_null, name = "feeling_heard_gender", n_factors = 2, type = "likert", save = "yes",  dir = "../results/survey") 

m_heard_gender_out %>% t() %>% kbl() %>%
  kable_classic_2() # not significant

model_name	feeling_heard_gender
AIC	1086.637
n_obs	373
lrt_pval	0.112
lrt_chisq	4.378
intercept_12	-3.623
intercept_23	-3.302
intercept_34	-2.95
intercept_45	-1.997
intercept_56	-1.001
intercept_67	0.355
n_factors	2
est_genderFemale	-0.113
lowerCI_genderFemale	-0.525
higherCI_genderFemale	0.3
se_genderFemale	0.21
tval_genderFemale	-0.537
pval_genderFemale	0.591
est_genderNon-binary	-1.348
lowerCI_genderNon-binary	-2.576
higherCI_genderNon-binary	-0.12
se_genderNon-binary	0.624
tval_genderNon-binary	-2.158
pval_genderNon-binary	0.032

# lgbtqia 

m_heard_lgbtq_null <- polr(feeling_heard_rating ~ 1, data=subset(survey, !is.na(lgbtq)))

m_heard_lgbtq <- polr(feeling_heard_rating ~ lgbtq,
                   data=subset(survey, !is.na(lgbtq))) 

m_heard_lgbtq_out <- collect_out(model = m_heard_lgbtq, null = m_heard_lgbtq_null, name = "feeling_heard_lgbtq", n_factors = 1, type = "likert", save = "yes",  dir = "../results/survey") 

m_heard_lgbtq_out %>% t() %>% kbl() %>%
  kable_classic_2() # almost significant, note that the baseline is answer "No" to lgbtq identity

model_name	feeling_heard_lgbtq
AIC	1049.062
n_obs	360
lrt_pval	0.059
lrt_chisq	3.567
intercept_12	-3.658
intercept_23	-3.31
intercept_34	-2.935
intercept_45	-1.968
intercept_56	-0.946
intercept_67	0.353
n_factors	1
est_lgbtqYes	-0.487
lowerCI_lgbtqYes	-0.992
higherCI_lgbtqYes	0.018
se_lgbtqYes	0.257
tval_lgbtqYes	-1.895
pval_lgbtqYes	0.059

# nationality
survey$nationality_continent <- factor(survey$nationality_continent, levels = c("Europe", "Asia", "North America", "Oceania", "South America"))

m_heard_nat_null <- polr(feeling_heard_rating ~ 1, data=subset(survey, !is.na(nationality_continent)))

m_heard_nat <- polr(feeling_heard_rating ~ nationality_continent,
                   data=survey) 

m_heard_nat_out <- collect_out(model = m_heard_nat, null = m_heard_nat_null, name = "feeling_heard_nat", n_factors = 4, type = "likert", save = "yes", 
            dir = "../results/survey") 

m_heard_nat_out %>% t() %>% kbl() %>%
  kable_classic_2() # almost significant, but Asia is borderline sig with lower values

model_name	feeling_heard_nat
AIC	1082.951
n_obs	374
lrt_pval	0.055
lrt_chisq	9.247
intercept_12	-3.58
intercept_23	-3.258
intercept_34	-2.901
intercept_45	-1.953
intercept_56	-0.961
intercept_67	0.408
n_factors	4
est_nationality_continentAsia	-0.729
lowerCI_nationality_continentAsia	-1.393
higherCI_nationality_continentAsia	-0.065
se_nationality_continentAsia	0.338
tval_nationality_continentAsia	-2.159
pval_nationality_continentAsia	0.032
est_nationality_continentNorth America	0.435
lowerCI_nationality_continentNorth America	-0.267
higherCI_nationality_continentNorth America	1.136
se_nationality_continentNorth America	0.357
tval_nationality_continentNorth America	1.218
pval_nationality_continentNorth America	0.224
est_nationality_continentOceania	-0.93
lowerCI_nationality_continentOceania	-2.234
higherCI_nationality_continentOceania	0.374
se_nationality_continentOceania	0.663
tval_nationality_continentOceania	-1.402
pval_nationality_continentOceania	0.162
est_nationality_continentSouth America	0.602
lowerCI_nationality_continentSouth America	-0.905
higherCI_nationality_continentSouth America	2.109
se_nationality_continentSouth America	0.766
tval_nationality_continentSouth America	0.785
pval_nationality_continentSouth America	0.433

# affiliation
survey$affiliation_continent <- factor(survey$affiliation_continent, levels = c("Europe", "Asia", "Africa", "North America", "Oceania", "South America"))

m_heard_aff_null <- polr(feeling_heard_rating ~ 1, data=subset(survey, !is.na(affiliation_continent)))

m_heard_aff <- polr(feeling_heard_rating ~ affiliation_continent,
                   data=subset(survey, !is.na(affiliation_continent))) 

m_heard_aff_out <- collect_out(model = m_heard_aff, null = m_heard_aff_null, name = "feeling_heard_aff", n_factors = 5, type = "likert", save = "yes", dir = "../results/survey") 

m_heard_aff_out %>% t() %>% kbl() %>%
  kable_classic_2() # not significant

model_name	feeling_heard_aff
AIC	1094.802
n_obs	377
lrt_pval	0.122
lrt_chisq	8.695
intercept_12	-3.528
intercept_23	-3.206
intercept_34	-2.85
intercept_45	-1.886
intercept_56	-0.913
intercept_67	0.449
n_factors	5
est_affiliation_continentAsia	-0.667
lowerCI_affiliation_continentAsia	-1.686
higherCI_affiliation_continentAsia	0.351
se_affiliation_continentAsia	0.518
tval_affiliation_continentAsia	-1.288
pval_affiliation_continentAsia	0.199
est_affiliation_continentAfrica	15.592
lowerCI_affiliation_continentAfrica	15.592
higherCI_affiliation_continentAfrica	15.592
se_affiliation_continentAfrica	0
tval_affiliation_continentAfrica	6117895554
pval_affiliation_continentAfrica	0
est_affiliation_continentNorth America	0.438
lowerCI_affiliation_continentNorth America	-0.454
higherCI_affiliation_continentNorth America	1.329
se_affiliation_continentNorth America	0.453
tval_affiliation_continentNorth America	0.965
pval_affiliation_continentNorth America	0.335
est_affiliation_continentOceania	-0.369
lowerCI_affiliation_continentOceania	-1.606
higherCI_affiliation_continentOceania	0.867
se_affiliation_continentOceania	0.629
tval_affiliation_continentOceania	-0.587
pval_affiliation_continentOceania	0.557
est_affiliation_continentSouth America	15.592
lowerCI_affiliation_continentSouth America	15.592
higherCI_affiliation_continentSouth America	15.592
se_affiliation_continentSouth America	0
tval_affiliation_continentSouth America	3058988690
pval_affiliation_continentSouth America	0

# expat

survey$expat <- factor(survey$expat, levels = c("No expat", "Expat"))

m_heard_expat_null <- polr(feeling_heard_rating ~ 1, data=subset(survey, !is.na(expat)))

m_heard_expat <- polr(feeling_heard_rating ~ expat,
                   data=subset(survey, !is.na(expat))) 

m_heard_expat_out <- collect_out(model = m_heard_expat, null = m_heard_expat_null, name = "feeling_heard_expat", n_factors = 1, type = "likert", save = "yes",  dir = "../results/survey") 

m_heard_expat_out %>% t() %>% kbl() %>%
  kable_classic_2() # not significant

model_name	feeling_heard_expat
AIC	1077.053
n_obs	371
lrt_pval	0.197
lrt_chisq	1.661
intercept_12	-3.381
intercept_23	-3.06
intercept_34	-2.706
intercept_45	-1.775
intercept_56	-0.806
intercept_67	0.561
n_factors	1
est_expatExpat	0.245
lowerCI_expatExpat	-0.129
higherCI_expatExpat	0.619
se_expatExpat	0.19
tval_expatExpat	1.287
pval_expatExpat	0.199

# english level 
survey$english_comfort_rating <- as.numeric(as.character(survey$english_comfort_rating))

m_heard_english_null <- polr(feeling_heard_rating ~ 1, data=subset(survey, !is.na(english_comfort_rating)))

m_heard_english <- polr(feeling_heard_rating ~ english_comfort_rating,
                   data=subset(survey, !is.na(english_comfort_rating))) 

m_heard_english_out <- collect_out(model = m_heard_english, null = m_heard_english_null, name = "feeling_heard_english", n_factors = 1, type = "likert", save = "yes",  dir = "../results/survey") 

m_heard_english_out %>% t() %>% kbl() %>%
  kable_classic_2() # significant

model_name	feeling_heard_english
AIC	1098.731
n_obs	384
lrt_pval	0
lrt_chisq	14.376
intercept_12	-1.169
intercept_23	-0.845
intercept_34	-0.48
intercept_45	0.5
intercept_56	1.517
intercept_67	2.903
n_factors	1
est_english_comfort_rating	0.381
lowerCI_english_comfort_rating	0.186
higherCI_english_comfort_rating	0.575
se_english_comfort_rating	0.099
tval_english_comfort_rating	3.841
pval_english_comfort_rating	0

# expert rating
survey$expertise_rating <- as.numeric(as.character(survey$expertise_rating))

m_heard_expert_null <- polr(feeling_heard_rating ~ 1, data=subset(survey, !is.na(expertise_rating)))

m_heard_expert <- polr(feeling_heard_rating ~ expertise_rating,
                   data=subset(survey, !is.na(expertise_rating))) 

m_heard_expert_out <- collect_out(model = m_heard_expert, null = m_heard_expert_null, name = "feeling_heard_expert", n_factors = 1, type = "likert", save = "yes",  dir = "../results/survey") 

m_heard_expert_out %>% t() %>% kbl() %>%
  kable_classic_2() # significant

model_name	feeling_heard_expert
AIC	1091.194
n_obs	384
lrt_pval	0
lrt_chisq	21.914
intercept_12	-2.285
intercept_23	-1.962
intercept_34	-1.605
intercept_45	-0.636
intercept_56	0.383
intercept_67	1.788
n_factors	1
est_expertise_rating	0.274
lowerCI_expertise_rating	0.158
higherCI_expertise_rating	0.391
se_expertise_rating	0.059
tval_expertise_rating	4.622
pval_expertise_rating	0

### build final model only with significant variables

m_heard_null <- polr(feeling_heard_rating ~ 1, data=subset(survey, !is.na(expertise_rating) & !is.na(english_comfort_rating)))

m_heard <- polr(feeling_heard_rating ~english_comfort_rating + expertise_rating, data=subset(survey, !is.na(expertise_rating) & !is.na(english_comfort_rating)))

drop1(m_heard, test = "Chisq")

Single term deletions

Model:
feeling_heard_rating ~ english_comfort_rating + expertise_rating
                       Df    AIC     LRT Pr(>Chi)    
<none>                    1085.7                     
english_comfort_rating  1 1091.2  7.4904 0.006203 ** 
expertise_rating        1 1098.7 15.0275 0.000106 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

coeftest(m_heard)

t test of coefficients:

                       Estimate Std. Error t value  Pr(>|t|)    
english_comfort_rating 0.283610   0.102551  2.7656 0.0059628 ** 
expertise_rating       0.235642   0.061224  3.8488 0.0001394 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

m_heard_out <- collect_out(model = m_heard, null = m_heard_null, name = "feeling_heard_final", n_factors = 2, type = "likert", save = "yes",  dir = "../results/survey") 

m_heard_out %>% t() %>% kbl() %>%
  kable_classic_2() 

model_name	feeling_heard_final
AIC	1085.704
n_obs	384
lrt_pval	0
lrt_chisq	29.404
intercept_12	-0.706
intercept_23	-0.38
intercept_34	-0.015
intercept_45	0.972
intercept_56	2.009
intercept_67	3.436
n_factors	2
est_english_comfort_rating	0.284
lowerCI_english_comfort_rating	0.082
higherCI_english_comfort_rating	0.485
se_english_comfort_rating	0.103
tval_english_comfort_rating	2.766
pval_english_comfort_rating	0.006
est_expertise_rating	0.236
lowerCI_expertise_rating	0.115
higherCI_expertise_rating	0.356
se_expertise_rating	0.061
tval_expertise_rating	3.849
pval_expertise_rating	0

The results indicate that none of the social identity variables (gender, lgbtq, nationality, affiliation (but this model did not converge)) affected whether a person felt heard during the congress. However, people who are more comfortable speaking English felt heard more, and people that rated their expertise in the field higher also felt heard more.

7.2.2 Feeling comfortable being yourself

# first, we explore the distribution of answers

ggplot(survey, aes(comfort_being_yourself_rating)) + 
  geom_histogram(stat="count", aes(y=stat(count/sum(count)*100)), fill = clrs[11]) + 
  geom_text(aes(label = paste0("N = ", ..count..), y = stat(count/sum(count)*100)), 
            stat="count", vjust=-1) +
  ylim(0, 100) + 
  labs(x = "I felt comfortable being myself", y = "% of responses") 

# then we test the effect of each of the social identity variables

# gender
m_yourself_gender_null <- polr(comfort_being_yourself_rating ~ 1, data=subset(survey, !is.na(gender)))

m_yourself_gender <- polr(comfort_being_yourself_rating ~ gender,
                   data=survey) 

m_yourself_gender_out <- collect_out(model = m_yourself_gender, null = m_yourself_gender_null, name = "comf_yourself_gender", n_factors = 2, type = "likert", save = "yes",  dir = "../results/survey") 

m_yourself_gender_out %>% t() %>% kbl() %>%
  kable_classic_2() # significant

model_name	comf_yourself_gender
AIC	982.799
n_obs	374
lrt_pval	0.001
lrt_chisq	13.295
intercept_12	-4.984
intercept_23	-3.945
intercept_34	-3.486
intercept_45	-2.765
intercept_56	-1.599
intercept_67	-0.289
n_factors	2
est_genderFemale	-0.464
lowerCI_genderFemale	-0.894
higherCI_genderFemale	-0.034
se_genderFemale	0.219
tval_genderFemale	-2.123
pval_genderFemale	0.034
est_genderNon-binary	-2.323
lowerCI_genderNon-binary	-3.628
higherCI_genderNon-binary	-1.018
se_genderNon-binary	0.664
tval_genderNon-binary	-3.5
pval_genderNon-binary	0.001

# lgbtqia 

m_yourself_lgbtq_null <- polr(comfort_being_yourself_rating ~ 1, data=subset(survey, !is.na(lgbtq)))

m_yourself_lgbtq <- polr(comfort_being_yourself_rating ~ lgbtq,
                   data=subset(survey, !is.na(lgbtq))) 

m_yourself_lgbtq_out <- collect_out(model = m_yourself_lgbtq, null = m_yourself_lgbtq_null, name = "comf_yourself_lgbtq", n_factors = 1, type = "likert", save = "yes",  dir = "../results/survey") 

m_yourself_lgbtq_out %>% t() %>% kbl() %>%
  kable_classic_2() # almost significant

model_name	comf_yourself_lgbtq
AIC	949.434
n_obs	361
lrt_pval	0.069
lrt_chisq	3.3
intercept_12	-4.88
intercept_23	-3.655
intercept_34	-3.166
intercept_45	-2.422
intercept_56	-1.342
intercept_67	-0.027
n_factors	1
est_lgbtqYes	-0.478
lowerCI_lgbtqYes	-0.993
higherCI_lgbtqYes	0.037
se_lgbtqYes	0.262
tval_lgbtqYes	-1.826
pval_lgbtqYes	0.069

# nationality
m_yourself_nat_null <- polr(comfort_being_yourself_rating ~ 1, data=subset(survey, !is.na(nationality_continent)))

m_yourself_nat <- polr(comfort_being_yourself_rating ~ nationality_continent,
                   data=survey) 

m_yourself_nat_out <- collect_out(model = m_yourself_nat, null = m_yourself_nat_null, name = "comf_yourself_nat", n_factors = 4, type = "likert", save = "yes", 
            dir = "../results/survey") 

m_yourself_nat_out %>% t() %>% kbl() %>%
  kable_classic_2() # not significant, but East Europe is sig with higher values

model_name	comf_yourself_nat
AIC	1001.98
n_obs	375
lrt_pval	0.735
lrt_chisq	2.006
intercept_12	-4.524
intercept_23	-3.494
intercept_34	-3.044
intercept_45	-2.337
intercept_56	-1.193
intercept_67	0.081
n_factors	4
est_nationality_continentAsia	0.071
lowerCI_nationality_continentAsia	-0.574
higherCI_nationality_continentAsia	0.715
se_nationality_continentAsia	0.328
tval_nationality_continentAsia	0.215
pval_nationality_continentAsia	0.83
est_nationality_continentNorth America	-0.209
lowerCI_nationality_continentNorth America	-0.88
higherCI_nationality_continentNorth America	0.463
se_nationality_continentNorth America	0.342
tval_nationality_continentNorth America	-0.611
pval_nationality_continentNorth America	0.541
est_nationality_continentOceania	0.352
lowerCI_nationality_continentOceania	-1.124
higherCI_nationality_continentOceania	1.828
se_nationality_continentOceania	0.751
tval_nationality_continentOceania	0.469
pval_nationality_continentOceania	0.639
est_nationality_continentSouth America	0.91
lowerCI_nationality_continentSouth America	-0.748
higherCI_nationality_continentSouth America	2.567
se_nationality_continentSouth America	0.843
tval_nationality_continentSouth America	1.079
pval_nationality_continentSouth America	0.281

# affiliation
m_yourself_aff_null <- polr(comfort_being_yourself_rating ~ 1, data=subset(survey, !is.na(affiliation_continent)))

m_yourself_aff <- polr(comfort_being_yourself_rating ~ affiliation_continent,
                   data=subset(survey, !is.na(affiliation_continent))) 

m_yourself_aff_out <- collect_out(model = m_yourself_aff, null = m_yourself_aff_null, name = "comf_yourself_aff", n_factors = 5, type = "likert", save = "yes", dir = "../results/survey") 

m_yourself_aff_out %>% t() %>% kbl() %>%
  kable_classic_2() # not significant

model_name	comf_yourself_aff
AIC	999.597
n_obs	378
lrt_pval	0.29
lrt_chisq	6.171
intercept_12	-4.515
intercept_23	-3.583
intercept_34	-3.095
intercept_45	-2.355
intercept_56	-1.194
intercept_67	0.102
n_factors	5
est_affiliation_continentAsia	0.588
lowerCI_affiliation_continentAsia	-0.43
higherCI_affiliation_continentAsia	1.607
se_affiliation_continentAsia	0.518
tval_affiliation_continentAsia	1.136
pval_affiliation_continentAsia	0.257
est_affiliation_continentAfrica	15.292
lowerCI_affiliation_continentAfrica	15.292
higherCI_affiliation_continentAfrica	15.292
se_affiliation_continentAfrica	0
tval_affiliation_continentAfrica	5516237801
pval_affiliation_continentAfrica	0
est_affiliation_continentNorth America	0.131
lowerCI_affiliation_continentNorth America	-0.753
higherCI_affiliation_continentNorth America	1.015
se_affiliation_continentNorth America	0.45
tval_affiliation_continentNorth America	0.291
pval_affiliation_continentNorth America	0.771
est_affiliation_continentOceania	-0.349
lowerCI_affiliation_continentOceania	-1.555
higherCI_affiliation_continentOceania	0.858
se_affiliation_continentOceania	0.614
tval_affiliation_continentOceania	-0.569
pval_affiliation_continentOceania	0.57
est_affiliation_continentSouth America	15.292
lowerCI_affiliation_continentSouth America	15.292
higherCI_affiliation_continentSouth America	15.292
se_affiliation_continentSouth America	0
tval_affiliation_continentSouth America	2758136541
pval_affiliation_continentSouth America	0

# expat

m_yourself_expat_null <- polr(comfort_being_yourself_rating ~ 1, data=subset(survey, !is.na(expat)))

m_yourself_expat <- polr(comfort_being_yourself_rating ~ expat,
                   data=subset(survey, !is.na(expat))) 

m_yourself_expat_out <- collect_out(model = m_yourself_expat, null = m_yourself_expat_null, name = "comf_yourself_expat", n_factors = 1, type = "likert", save = "yes",  dir = "../results/survey") 

m_yourself_expat_out %>% t() %>% kbl() %>%
  kable_classic_2() # not significant

model_name	comf_yourself_expat
AIC	986.957
n_obs	372
lrt_pval	0.945
lrt_chisq	0.005
intercept_12	-4.53
intercept_23	-3.595
intercept_34	-3.109
intercept_45	-2.37
intercept_56	-1.208
intercept_67	0.069
n_factors	1
est_expatExpat	-0.013
lowerCI_expatExpat	-0.394
higherCI_expatExpat	0.368
se_expatExpat	0.194
tval_expatExpat	-0.069
pval_expatExpat	0.945

# english level 
survey$english_comfort_rating <- as.numeric(as.character(survey$english_comfort_rating))

m_yourself_english_null <- polr(comfort_being_yourself_rating ~ 1, data=subset(survey, !is.na(english_comfort_rating)))

m_yourself_english <- polr(comfort_being_yourself_rating ~ english_comfort_rating,
                   data=subset(survey, !is.na(english_comfort_rating))) 

m_yourself_english_out <- collect_out(model = m_yourself_english, null = m_yourself_english_null, name = "comf_yourself_english", n_factors = 1, type = "likert", save = "yes",  dir = "../results/survey") 

m_yourself_english_out %>% t() %>% kbl() %>%
  kable_classic_2() # significant

model_name	comf_yourself_english
AIC	1008.542
n_obs	385
lrt_pval	0.001
lrt_chisq	10.598
intercept_12	-2.451
intercept_23	-1.413
intercept_34	-0.95
intercept_45	-0.227
intercept_56	0.944
intercept_67	2.253
n_factors	1
est_english_comfort_rating	0.339
lowerCI_english_comfort_rating	0.135
higherCI_english_comfort_rating	0.542
se_english_comfort_rating	0.104
tval_english_comfort_rating	3.269
pval_english_comfort_rating	0.001

# expert rating
survey$expertise_rating <- as.numeric(as.character(survey$expertise_rating))

m_yourself_expert_null <- polr(comfort_being_yourself_rating ~ 1, data=subset(survey, !is.na(expertise_rating)))

m_yourself_expert <- polr(comfort_being_yourself_rating ~ expertise_rating,
                   data=subset(survey, !is.na(expertise_rating))) 

m_yourself_expert_out <- collect_out(model = m_yourself_expert, null = m_yourself_expert_null, name = "comf_yourself_expert", n_factors = 1, type = "likert", save = "yes",  dir = "../results/survey") 

m_yourself_expert_out %>% t() %>% kbl() %>%
  kable_classic_2() # significant

model_name	comf_yourself_expert
AIC	1001.373
n_obs	385
lrt_pval	0
lrt_chisq	17.767
intercept_12	-3.436
intercept_23	-2.399
intercept_34	-1.942
intercept_45	-1.228
intercept_56	-0.061
intercept_67	1.268
n_factors	1
est_expertise_rating	0.248
lowerCI_expertise_rating	0.131
higherCI_expertise_rating	0.365
se_expertise_rating	0.059
tval_expertise_rating	4.169
pval_expertise_rating	0

### build final model only with significant variables

m_yourself_null <- polr(comfort_being_yourself_rating ~ 1, data=subset(survey, !is.na(gender) & !is.na(expertise_rating) & !is.na(english_comfort_rating)))

m_yourself <- polr(comfort_being_yourself_rating ~ gender + english_comfort_rating + expertise_rating, data=subset(survey, !is.na(gender) & !is.na(expertise_rating) & !is.na(english_comfort_rating)))

drop1(m_yourself, test = "Chisq")

Single term deletions

Model:
comfort_being_yourself_rating ~ gender + english_comfort_rating + 
    expertise_rating
                       Df    AIC     LRT  Pr(>Chi)    
<none>                    962.68                      
gender                  2 971.35 12.6677 0.0017752 ** 
english_comfort_rating  1 967.20  6.5161 0.0106902 *  
expertise_rating        1 973.32 12.6387 0.0003778 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

m_yourself_out <- collect_out(model = m_yourself, null = m_yourself_null, name = "comf_yourself_final", n_factors = 4, type = "likert", save = "yes",  dir = "../results/survey") 

m_yourself_out %>% kbl() %>%
  kable_classic_2() 

model_name	AIC	n_obs	lrt_pval	lrt_chisq	intercept_12	intercept_23	intercept_34	intercept_45	intercept_56	intercept_67	n_factors	est_genderFemale	lowerCI_genderFemale	higherCI_genderFemale	se_genderFemale	tval_genderFemale	pval_genderFemale	est_genderNon-binary	lowerCI_genderNon-binary	higherCI_genderNon-binary	se_genderNon-binary	tval_genderNon-binary	pval_genderNon-binary	est_english_comfort_rating	lowerCI_english_comfort_rating	higherCI_english_comfort_rating	se_english_comfort_rating	tval_english_comfort_rating	pval_english_comfort_rating	est_expertise_rating	lowerCI_expertise_rating	higherCI_expertise_rating	se_expertise_rating	tval_expertise_rating	pval_expertise_rating
comf_yourself_final	962.684	374	0	37.41	-2.277	-1.218	-0.74	0.002	1.2	2.572	4	-0.475	-0.911	-0.039	0.222	-2.141	0.033	-2.261	-3.588	-0.934	0.675	-3.351	0.001	0.281	0.067	0.496	0.109	2.579	0.01	0.218	0.097	0.339	0.062	3.534	0

The results indicate that lgbtq and nationality did not affect whether a person felt comfortable being themselves during the congress (but lgbtq came close to significance). However, men feel more comfortable than women and non-binary people feel less comfortable being themselves compared to women. People who are more comfortable speaking English felt more comfortable being themselves, and people that rated their expertise in the field higher also felt more comfortable being themselves.

7.2.3 Sense of belonging

# first, we explore the distribution of answers

ggplot(survey, aes(sense_of_belonging_rating)) + 
  geom_histogram(stat="count", aes(y=stat(count/sum(count)*100)), fill = clrs[11]) + 
  geom_text(aes(label = paste0("N = ", ..count..), y = stat(count/sum(count)*100)), 
            stat="count", vjust=-1) +
  ylim(0, 100) + 
  labs(x = "Attending the Behaviour 2023 congress helped me 
feel like I belong in my research field", y = "% of responses") 

# then we test the effect of each of the social identity variables

# gender
m_sob_gender_null <- polr(sense_of_belonging_rating ~ 1, data=subset(survey, !is.na(gender)))

m_sob_gender <- polr(sense_of_belonging_rating ~ gender,
                   data=survey) 

m_sob_gender_out <- collect_out(model = m_sob_gender, null = m_sob_gender_null, name = "belonging_gender", n_factors = 2, type = "likert", save = "yes",  dir = "../results/survey") 

m_sob_gender_out %>% t() %>% kbl() %>%
  kable_classic_2() # not significant

model_name	belonging_gender
AIC	1137.815
n_obs	375
lrt_pval	0.107
lrt_chisq	4.475
intercept_12	-4.216
intercept_23	-3.662
intercept_34	-2.89
intercept_45	-1.684
intercept_56	-0.84
intercept_67	0.072
n_factors	2
est_genderFemale	-0.316
lowerCI_genderFemale	-0.724
higherCI_genderFemale	0.092
se_genderFemale	0.207
tval_genderFemale	-1.525
pval_genderFemale	0.128
est_genderNon-binary	-1.181
lowerCI_genderNon-binary	-2.505
higherCI_genderNon-binary	0.144
se_genderNon-binary	0.674
tval_genderNon-binary	-1.753
pval_genderNon-binary	0.081

# lgbtqia 

m_sob_lgbtq_null <- polr(sense_of_belonging_rating ~ 1, data=subset(survey, !is.na(lgbtq)))

m_sob_lgbtq <- polr(sense_of_belonging_rating ~ lgbtq,
                   data=subset(survey, !is.na(lgbtq))) 

m_sob_lgbtq_out <- collect_out(model = m_sob_lgbtq, null = m_sob_lgbtq_null, name = "belonging_lgbtq", n_factors = 1, type = "likert", save = "yes",  dir = "../results/survey") 

m_sob_lgbtq_out %>% t() %>% kbl() %>%
  kable_classic_2() # not significant

model_name	belonging_lgbtq
AIC	1100.466
n_obs	362
lrt_pval	0.365
lrt_chisq	0.82
intercept_12	-3.969
intercept_23	-3.505
intercept_34	-2.687
intercept_45	-1.469
intercept_56	-0.618
intercept_67	0.274
n_factors	1
est_lgbtqYes	-0.23
lowerCI_lgbtqYes	-0.729
higherCI_lgbtqYes	0.269
se_lgbtqYes	0.254
tval_lgbtqYes	-0.908
pval_lgbtqYes	0.365

# nationality
m_sob_nat_null <- polr(sense_of_belonging_rating ~ 1, data=subset(survey, !is.na(nationality_continent)))

m_sob_nat <- polr(sense_of_belonging_rating ~ nationality_continent,
                   data=survey) 

m_sob_nat_out <- collect_out(model = m_sob_nat, null = m_sob_nat_null, name = "belonging_nat", n_factors = 4, type = "likert", save = "yes", 
            dir = "../results/survey") 

m_sob_nat_out %>% t() %>% kbl() %>%
  kable_classic_2() # not significant, but North America is borderline sig with higher values

model_name	belonging_nat
AIC	1137.541
n_obs	376
lrt_pval	0.112
lrt_chisq	7.504
intercept_12	-3.895
intercept_23	-3.343
intercept_34	-2.532
intercept_45	-1.366
intercept_56	-0.567
intercept_67	0.377
n_factors	4
est_nationality_continentAsia	0.314
lowerCI_nationality_continentAsia	-0.357
higherCI_nationality_continentAsia	0.985
se_nationality_continentAsia	0.341
tval_nationality_continentAsia	0.919
pval_nationality_continentAsia	0.358
est_nationality_continentNorth America	0.647
lowerCI_nationality_continentNorth America	-0.012
higherCI_nationality_continentNorth America	1.306
se_nationality_continentNorth America	0.335
tval_nationality_continentNorth America	1.93
pval_nationality_continentNorth America	0.054
est_nationality_continentOceania	-0.645
lowerCI_nationality_continentOceania	-1.905
higherCI_nationality_continentOceania	0.616
se_nationality_continentOceania	0.641
tval_nationality_continentOceania	-1.006
pval_nationality_continentOceania	0.315
est_nationality_continentSouth America	1.159
lowerCI_nationality_continentSouth America	-0.517
higherCI_nationality_continentSouth America	2.834
se_nationality_continentSouth America	0.852
tval_nationality_continentSouth America	1.36
pval_nationality_continentSouth America	0.175

# affiliation
m_sob_aff_null <- polr(sense_of_belonging_rating ~ 1, data=subset(survey, !is.na(affiliation_continent)))

m_sob_aff <- polr(sense_of_belonging_rating ~ affiliation_continent,
                   data=subset(survey, !is.na(affiliation_continent))) 

m_sob_aff_out <- collect_out(model = m_sob_aff, null = m_sob_aff_null, name = "belonging_aff", n_factors = 5, type = "likert", save = "yes", 
             dir = "../results/survey") 

m_sob_aff_out %>% t() %>% kbl() %>%
  kable_classic_2() # significant

model_name	belonging_aff
AIC	1141.766
n_obs	379
lrt_pval	0.013
lrt_chisq	14.457
intercept_12	-3.899
intercept_23	-3.346
intercept_34	-2.533
intercept_45	-1.381
intercept_56	-0.551
intercept_67	0.415
n_factors	5
est_affiliation_continentAsia	0.734
lowerCI_affiliation_continentAsia	-0.227
higherCI_affiliation_continentAsia	1.694
se_affiliation_continentAsia	0.488
tval_affiliation_continentAsia	1.502
pval_affiliation_continentAsia	0.134
est_affiliation_continentAfrica	-0.063
lowerCI_affiliation_continentAfrica	-2.932
higherCI_affiliation_continentAfrica	2.806
se_affiliation_continentAfrica	1.459
tval_affiliation_continentAfrica	-0.043
pval_affiliation_continentAfrica	0.966
est_affiliation_continentNorth America	1.439
lowerCI_affiliation_continentNorth America	0.409
higherCI_affiliation_continentNorth America	2.469
se_affiliation_continentNorth America	0.524
tval_affiliation_continentNorth America	2.747
pval_affiliation_continentNorth America	0.006
est_affiliation_continentOceania	0.214
lowerCI_affiliation_continentOceania	-0.827
higherCI_affiliation_continentOceania	1.255
se_affiliation_continentOceania	0.529
tval_affiliation_continentOceania	0.405
pval_affiliation_continentOceania	0.686
est_affiliation_continentSouth America	16.065
lowerCI_affiliation_continentSouth America	16.065
higherCI_affiliation_continentSouth America	16.065
se_affiliation_continentSouth America	0
tval_affiliation_continentSouth America	2685232830
pval_affiliation_continentSouth America	0

# expat

m_sob_expat_null <- polr(sense_of_belonging_rating ~ 1, data=subset(survey, !is.na(expat)))

m_sob_expat <- polr(sense_of_belonging_rating ~ expat,
                   data=subset(survey, !is.na(expat))) 

m_sob_expat_out <- collect_out(model = m_sob_expat, null = m_sob_expat_null, name = "belonging_expat", n_factors = 1, type = "likert", save = "yes",  dir = "../results/survey") 

m_sob_expat_out %>% t() %>% kbl() %>%
  kable_classic_2() # not significant

model_name	belonging_expat
AIC	1130.934
n_obs	373
lrt_pval	0.471
lrt_chisq	0.519
intercept_12	-4.023
intercept_23	-3.47
intercept_34	-2.657
intercept_45	-1.514
intercept_56	-0.714
intercept_67	0.223
n_factors	1
est_expatExpat	-0.136
lowerCI_expatExpat	-0.508
higherCI_expatExpat	0.236
se_expatExpat	0.189
tval_expatExpat	-0.72
pval_expatExpat	0.472

# english level 
survey$english_comfort_rating <- as.numeric(as.character(survey$english_comfort_rating))

m_sob_english_null <- polr(sense_of_belonging_rating ~ 1, data=subset(survey, !is.na(english_comfort_rating)))

m_sob_english <- polr(sense_of_belonging_rating ~ english_comfort_rating,
                   data=subset(survey, !is.na(english_comfort_rating))) 

m_sob_english_out <- collect_out(model = m_sob_english, null = m_sob_english_null, name = "belonging_english", n_factors = 1, type = "likert", save = "yes",  dir = "../results/survey") 

m_sob_english_out %>% t() %>% kbl() %>%
  kable_classic_2() # significant

model_name	belonging_english
AIC	1151.747
n_obs	386
lrt_pval	0
lrt_chisq	19.426
intercept_12	-1.316
intercept_23	-0.759
intercept_34	0.063
intercept_45	1.27
intercept_56	2.144
intercept_67	3.105
n_factors	1
est_english_comfort_rating	0.433
lowerCI_english_comfort_rating	0.241
higherCI_english_comfort_rating	0.626
se_english_comfort_rating	0.098
tval_english_comfort_rating	4.421
pval_english_comfort_rating	0

# expert rating
survey$expertise_rating <- as.numeric(as.character(survey$expertise_rating))

m_sob_expert_null <- polr(sense_of_belonging_rating ~ 1, data=subset(survey, !is.na(expertise_rating)))

m_sob_expert <- polr(sense_of_belonging_rating ~ expertise_rating,
                   data=subset(survey, !is.na(expertise_rating))) 

m_sob_expert_out <- collect_out(model = m_sob_expert, null = m_sob_expert_null, name = "belonging_expert", n_factors = 1, type = "likert", save = "yes",  dir = "../results/survey") 

m_sob_expert_out %>% t() %>% kbl() %>%
  kable_classic_2() # significant

model_name	belonging_expert
AIC	1124.873
n_obs	386
lrt_pval	0
lrt_chisq	46.3
intercept_12	-2.269
intercept_23	-1.71
intercept_34	-0.886
intercept_45	0.339
intercept_56	1.257
intercept_67	2.281
n_factors	1
est_expertise_rating	0.4
lowerCI_expertise_rating	0.282
higherCI_expertise_rating	0.518
se_expertise_rating	0.06
tval_expertise_rating	6.661
pval_expertise_rating	0

### build final model only with significant variables

m_sob_null <- polr(sense_of_belonging_rating ~ 1, data=subset(survey, !is.na(expertise_rating) & !is.na(english_comfort_rating) & !is.na(affiliation_continent)))

m_sob <- polr(sense_of_belonging_rating ~ affiliation_continent + english_comfort_rating + expertise_rating, data=subset(survey, !is.na(expertise_rating) & !is.na(english_comfort_rating)& !is.na(affiliation_continent)))

drop1(m_sob, test = "Chisq")

Single term deletions

Model:
sense_of_belonging_rating ~ affiliation_continent + english_comfort_rating + 
    expertise_rating
                       Df    AIC    LRT  Pr(>Chi)    
<none>                    1097.0                     
affiliation_continent   5 1098.7 11.733  0.038629 *  
english_comfort_rating  1 1104.0  9.022  0.002667 ** 
expertise_rating        1 1126.7 31.693 1.806e-08 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

coeftest(m_sob)

t test of coefficients:

                                      Estimate  Std. Error     t value
affiliation_continentAsia           8.8114e-01  5.1664e-01  1.7055e+00
affiliation_continentAfrica        -1.0164e+00  1.4763e+00 -6.8850e-01
affiliation_continentNorth America  1.1562e+00  5.2882e-01  2.1864e+00
affiliation_continentOceania        6.4092e-02  5.2423e-01  1.2230e-01
affiliation_continentSouth America  1.5758e+01  2.7271e-08  5.7783e+08
english_comfort_rating              3.1388e-01  1.0360e-01  3.0298e+00
expertise_rating                    3.5048e-01  6.3142e-02  5.5507e+00
                                    Pr(>|t|)    
affiliation_continentAsia           0.088949 .  
affiliation_continentAfrica         0.491597    
affiliation_continentNorth America  0.029421 *  
affiliation_continentOceania        0.902761    
affiliation_continentSouth America < 2.2e-16 ***
english_comfort_rating              0.002621 ** 
expertise_rating                   5.472e-08 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

m_sob_out <- collect_out(model = m_sob, null = m_sob_null, name = "belonging_final", n_factors = 7, type = "likert", save = "yes",  dir = "../results/survey") 

m_sob_out %>% t() %>% kbl() %>%
  kable_classic_2() 

model_name	belonging_final
AIC	1096.961
n_obs	379
lrt_pval	0
lrt_chisq	63.263
intercept_12	-0.446
intercept_23	0.118
intercept_34	0.95
intercept_45	2.172
intercept_56	3.096
intercept_67	4.171
n_factors	7
est_affiliation_continentAsia	0.881
lowerCI_affiliation_continentAsia	-0.135
higherCI_affiliation_continentAsia	1.897
se_affiliation_continentAsia	0.517
tval_affiliation_continentAsia	1.706
pval_affiliation_continentAsia	0.089
est_affiliation_continentAfrica	-1.016
lowerCI_affiliation_continentAfrica	-3.919
higherCI_affiliation_continentAfrica	1.887
se_affiliation_continentAfrica	1.476
tval_affiliation_continentAfrica	-0.688
pval_affiliation_continentAfrica	0.492
est_affiliation_continentNorth America	1.156
lowerCI_affiliation_continentNorth America	0.116
higherCI_affiliation_continentNorth America	2.196
se_affiliation_continentNorth America	0.529
tval_affiliation_continentNorth America	2.186
pval_affiliation_continentNorth America	0.029
est_affiliation_continentOceania	0.064
lowerCI_affiliation_continentOceania	-0.967
higherCI_affiliation_continentOceania	1.095
se_affiliation_continentOceania	0.524
tval_affiliation_continentOceania	0.122
pval_affiliation_continentOceania	0.903
est_affiliation_continentSouth America	15.758
lowerCI_affiliation_continentSouth America	15.758
higherCI_affiliation_continentSouth America	15.758
se_affiliation_continentSouth America	0
tval_affiliation_continentSouth America	577826929
pval_affiliation_continentSouth America	0
est_english_comfort_rating	0.314
lowerCI_english_comfort_rating	0.11
higherCI_english_comfort_rating	0.518
se_english_comfort_rating	0.104
tval_english_comfort_rating	3.03
pval_english_comfort_rating	0.003
est_expertise_rating	0.35
lowerCI_expertise_rating	0.226
higherCI_expertise_rating	0.475
se_expertise_rating	0.063
tval_expertise_rating	5.551
pval_expertise_rating	0

The results are very similar to the “Feeling heard” part: none of the social identity variables (gender, lgbtq, nationality) affected whether a person felt that attending the congress increased their Sense of Belonging. However, people who are more comfortable speaking English felt more like it did, and people that rated their expertise in the field higher also felt like attending the congress increased their feeling like they belong in the field.