10What other gender disparities do we observe in oral sessions?
Author
Rebecca S. Chen & Petroula Botsidou
Note: this analysis is only part of the supplementary information of the manuscript
What other gender disparities can we observe in oral sessions?
More specifically, we also collected data on whether positive appraisal was given by a questioner, whether a person asked a question without raising their hand or being chosen by the session host to do so (a jumper), whether a speaker talked longer than their allocated time, and if a question was criticizing the speaker. Depending on the question, we investigated the effect of the question asker’s gender, speaker’s gender, host’s gender while correcting for appropriate confounders such as age or talk number.
For all analysis, we only excluded questions that were follow-up questions, and those asked by the host. This means that these data include manipulated and unmanipulated sessions, as we do not expect our manipulation to interfere with what a questioner says exactly.
10.1 Positive appraisal
Positive appraisal was defined as any positive words towards the speaker, but excluding when simply “Thank you for your talk” was said. However, if this comment included a compliment such as “Thank you for your nice talk”, this was counted as positive appraisal.
We used binomial GLMMs to address the effect of the gender and age of the questioner and gender and career stage of the speaker, and corrected for the question number within that Q&A session. We assessed the fit of each variable using a likelihood ratio test and included only the variables that explained significant variation in the final model.
10.1.1 Receiving
# explore the data table(data_analysis$compliment, data_analysis$questioner_gender) %>%kbl() %>%kable_classic_2()
## speaker career stagemodel_compliment_s_career <-glmer(compliment ~ speaker_career_short + (1|session_id/talk_id), data =subset(data_analysis, !is.na(speaker_career_short)), family ="binomial") model_compliment_s_career_null <-glmer(compliment ~1+ (1|session_id/talk_id), data =subset(data_analysis, !is.na(speaker_career_short)), family ="binomial") anova(model_compliment_s_career, model_compliment_s_career_null) # almost significant
# only question number and age questioner significant (LRT p < 0.05)# build final modelmodel_compliment_null <-glmer(compliment ~1+ (1|session_id/talk_id), data =subset(data_analysis, !is.na(question_nr) &!is.na(questioner_age)), family ="binomial") model_compliment <-glmer(compliment ~ question_nr + questioner_age + (1|session_id/talk_id), data =subset(data_analysis, !is.na(question_nr) &!is.na(questioner_age)), family ="binomial") # use helper function to collect outputmodel_compliment_out <-collect_out(model = model_compliment, null = model_compliment_null, n_factors =3, name ="m_compliment", type ="qa", save ="yes", dir ="../results/question-asking/")model_compliment_out %>%t() %>%kbl() %>%kable_classic_2()
model_name
m_compliment
AIC
1032.187
n_obs
822
lrt_pval
0
lrt_chisq
59.56
intercept_estimate
0.921
intercept_estimate_prop
0.715
intercept_pval
0
intercept_ci_lower
0.497
intercept_ci_higher
1.344
n_factors
3
est_question_nr
-0.365
lowerCI_question_nr
-0.473
higherCI_question_nr
-0.257
pval_question_nr
0
zval_question_nr
-6.607
est_questioner_age2
-0.607
lowerCI_questioner_age2
-0.976
higherCI_questioner_age2
-0.238
pval_questioner_age2
0.001
zval_questioner_age2
-3.222
est_questioner_age3
-0.493
lowerCI_questioner_age3
-0.99
higherCI_questioner_age3
0.004
pval_questioner_age3
0.052
zval_questioner_age3
-1.945
These results show is that the likelihood of a questioner giving a words of positive appraisal is higher for questions asked earlier in the Q&A, and that especially younger audience members (age < 35 years) are more likely to give words of positive appraisal compared to older age categories. Gender of either the question asker or speaker were not significant.
10.2 Jumpers
Next, we asked whether men or women are more likely to ask a question without being allocated to do so (i.e. chosen by the session host to ask your question). Since this might have to do with the perceived ‘authority’ of the session host, we controlled for the age and gender of the session host as well.
We used binomial GLMMs to address the effect of the gender and age of the questioner, and corrected for question number and age and gender of the session host. We assessed the fit of each variable using a likelihood ratio test and included only the variables that explained significant variation in the final model.
# recode NA to 'no jumper' = 0jumperdata <- data_analysis %>%mutate(jumper =as.factor(case_when(is.na(jumper) ~"0", jumper =="1"~"1" )))# explore the data table(jumperdata$jumper, jumperdata$questioner_gender) %>%kbl() %>%kable_classic_2()
# nothing is significant, trend for gender questioner
These results show is that the likelihood of a person jumping a question not significantly affected by any of the variables. There was however a tendency for men to be more likely to jump a question compared to women. Note that the inference of these models is however limited, since jumpers were rare (N = 18).
10.3 Speaker over time
Next, we investigated the probability that a speaker talks for longer than their allocated speaking time is affected by speaker gender or career stage. We did not expect that any other confounding variables would explain variation in speaking overtime, since this is something that was prepared by only the speaker.
# have to summarize by talk not per questiondata_overtime <- data_analysis %>%select(talk_id, session_id, overtime, speaker_gender, speaker_career_short) %>%unique()# N = nrow(data_overtime)
Generalized linear mixed model fit by maximum likelihood (Laplace
Approximation) [glmerMod]
Family: binomial ( logit )
Formula: overtime ~ speaker_career_short + (1 | session_id/talk_id)
Data: subset(data_overtime, !is.na(speaker_career_short))
AIC BIC logLik deviance df.resid
148.8 167.3 -69.4 138.8 294
Scaled residuals:
Min 1Q Median 3Q Max
-0.005965 -0.002171 -0.001696 -0.001311 0.078898
Random effects:
Groups Name Variance Std.Dev.
talk_id:session_id (Intercept) 2882.12 53.685
session_id (Intercept) 83.74 9.151
Number of obs: 299, groups: talk_id:session_id, 299; session_id, 63
Fixed effects:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -12.700754 0.001747 -7271.0 <2e-16 ***
speaker_career_shortLate career 1.103260 0.001747 631.6 <2e-16 ***
speaker_career_shortMid career -0.566997 0.001747 -324.6 <2e-16 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) sp__Lc
spkr_crr_Lc 0.000
spkr_crr_Mc 0.000 0.000
optimizer (Nelder_Mead) convergence code: 0 (OK)
Model failed to converge with max|grad| = 0.0404888 (tol = 0.002, component 1)
# use helper function to collect outputmodel_overtime_out <-collect_out(model = model_overtime_s_career, null = model_overtime_s_career_null, n_factors =2, name ="m_overtime", type ="qa", save ="yes", dir ="../results/question-asking/")model_overtime_out %>%t() %>%kbl() %>%kable_classic_2()
model_name
m_overtime
AIC
148.762
n_obs
299
lrt_pval
0
lrt_chisq
108.48
intercept_estimate
-12.701
intercept_estimate_prop
0
intercept_pval
0
intercept_ci_lower
-12.704
intercept_ci_higher
-12.697
n_factors
2
est_speaker_career_shortLate career
1.103
lowerCI_speaker_career_shortLate career
1.1
higherCI_speaker_career_shortLate career
1.107
pval_speaker_career_shortLate career
0
zval_speaker_career_shortLate career
631.607
est_speaker_career_shortMid career
-0.567
lowerCI_speaker_career_shortMid career
-0.57
higherCI_speaker_career_shortMid career
-0.564
pval_speaker_career_shortMid career
0
zval_speaker_career_shortMid career
-324.618
The model output shows that late career speakers were more likely to speak overtime and mid-career speakers were the least likely to speak overtime.
10.4 Critical question
Next, we assessed whether the likelihood of receiving a ‘critical’ question is affected by the gender and age of the questioner or the gender and career stage of the speaker.
## speaker career stagemodel_critical_s_career <-glmer(question_type_e ~ speaker_career_short + (1|session_id/talk_id), data =subset(data_analysis, !is.na(speaker_career_short)), family ="binomial") model_critical_s_career_null <-glmer(question_type_e ~1+ (1|session_id/talk_id), data =subset(data_analysis, !is.na(speaker_career_short)), family ="binomial") anova(model_critical_s_career, model_critical_s_career_null) # not significant
model_critical_q_age_out <-collect_out(model = model_critical_q_age, null = model_critical_q_age_null, n_factors =2, name ="m_critical_q_age", type ="qa", save ="yes", dir ="../results/question-asking/")model_critical_q_age_out %>%t() %>%kbl() %>%kable_classic_2()
model_name
m_critical_q_age
AIC
197.072
n_obs
805
lrt_pval
0.006
lrt_chisq
10.212
intercept_estimate
-10.888
intercept_estimate_prop
0
intercept_pval
0
intercept_ci_lower
-10.893
intercept_ci_higher
-10.883
n_factors
2
est_questioner_age2
2.188
lowerCI_questioner_age2
2.183
higherCI_questioner_age2
2.193
pval_questioner_age2
0
zval_questioner_age2
826.121
est_questioner_age3
3.068
lowerCI_questioner_age3
3.063
higherCI_questioner_age3
3.073
pval_questioner_age3
0
zval_questioner_age3
1158.491
# only age questioner significant
These results show is that the likelihood of a person asking or receiving a critical question was affected by the age category of the question asker. Note that the inference of these models is however limited, since critical questions were rare and subject to observer bias (N = 43).