Survey analysis

Regime type

Regime types of the home countries for each organization

home.regime.type <- survey.orgs.clean %>%
  group_by(home.regime.type) %>%
  summarise(num = n()) %>%
  mutate(prop = num / sum(num))
home.regime.type

## # A tibble: 2 x 3
##   home.regime.type   num      prop
##             <fctr> <int>     <dbl>
## 1        Democracy   569 0.8876755
## 2        Autocracy    72 0.1123245

Regime types of the target countries for each country-organization

work.regime.type <- survey.countries.clean %>%
  group_by(target.regime.type) %>%
  summarise(num = n())%>%
  mutate(prop = num / sum(num))
work.regime.type

## # A tibble: 2 x 3
##   target.regime.type   num      prop
##               <fctr> <int>     <dbl>
## 1          Democracy   440 0.6676783
## 2          Autocracy   219 0.3323217

Most NGOs are based in democracies (only 11% are headquartered in autocracies), but a third of them answered questions about their work in autocracies.

Relationship with government (Q4.11)

df.employees.relationship <- survey.clean.all %>%
  select(Q3.4.num, Q4.11, potential.contentiousness) %>%
  filter(!(Q4.11 %in% c("Don't know", "Prefer not to answer"))) %>%
  filter(!is.na(Q4.11)) %>%
  mutate(Q4.11 = droplevels(Q4.11),
         Q4.11 = factor(Q4.11, levels=rev(levels(Q4.11))))

df.employees.relationship.plot.means <- df.employees.relationship %>%
  group_by(Q4.11, potential.contentiousness) %>%
  summarise(average = mean(Q3.4.num, na.rm=TRUE),
            med = median(Q3.4.num, na.rm=TRUE),
            num = n())

ggplot(df.employees.relationship, aes(y=Q4.11, x=Q3.4.num)) + 
  geom_violinh(na.rm=TRUE) +
  geom_point(alpha=0.2, size=0.5) +
  geom_point(data=df.employees.relationship.plot.means,
             aes(x=med, y=Q4.11)) +
  scale_x_continuous(trans="log1p", breaks=c(0, 10^(0:5)), labels=comma) + 
  labs(x="Number of employees", y=NULL,
       title="Relationship with government and # of employees") +
  theme_ath() + facet_wrap(~ potential.contentiousness)

## Warning: Removed 9 rows containing missing values (geom_point).

ggplot(df.employees.relationship.plot.means, 
       aes(x=med, y=Q4.11, fill=potential.contentiousness)) +
  geom_barh(stat="identity", position="dodge") + 
  scale_x_continuous(expand=c(0, 0)) +
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  labs(x="Median number of employees", y=NULL, 
       title="Median # of employees and relationship with government") +
  theme_ath()

Types of regulation (Q4.16)

df.employees.reg.types <- survey.clean.all %>%
  select(Q3.4.num, starts_with("Q4.16"), -dplyr::contains("TEXT"), 
         potential.contentiousness) %>%
  gather(regulation, response, starts_with("Q4.16")) %>%
  mutate(regulation = gsub("Q4\\.16_", "", regulation)) %>%
  filter(!(response %in% c("Don't know", "Not applicable"))) %>%
  filter(!is.na(response)) %>%
  mutate(response = factor(response, levels=great.none, ordered=TRUE))

df.employees.reg.types.plot.means <- df.employees.reg.types %>%
  group_by(regulation, response, potential.contentiousness) %>%
  summarise(average = mean(Q3.4.num, na.rm=TRUE),
            med = median(Q3.4.num, na.rm=TRUE),
            num = n())

ggplot(df.employees.reg.types, aes(y=response, x=Q3.4.num, 
                                   fill=potential.contentiousness)) + 
  geom_violinh(na.rm=TRUE) + 
  geom_point(alpha=0.2, size=0.5) +
  geom_point(data=df.employees.reg.types.plot.means,
             aes(x=med, y=response)) +
  scale_x_continuous(trans="log1p", breaks=c(0, 10^(0:5)), labels=comma) + 
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  labs(x="Number of employees", y=NULL,
       title="Types of regulation, perceptions of restriction, and # of employees (logged)") +
  theme_ath() + 
  facet_wrap(~ regulation + potential.contentiousness, nrow=2)

## Warning: Removed 46 rows containing missing values (geom_point).

ggplot(df.employees.reg.types.plot.means, 
       aes(x=med, y=response, fill=potential.contentiousness)) +
  geom_barh(stat="identity", position="dodge") + 
  scale_x_continuous(expand=c(0, 0)) +
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  labs(x="Median number of employees", y=NULL, 
       title="Median # of employees and types of regulation") +
  theme_ath() + facet_wrap(~ regulation)

Overall perception of restriction (Q4.17)

df.employees.restriction <- survey.clean.all %>%
  select(Q3.4.num, Q4.17, potential.contentiousness) %>%
  filter(Q4.17 != "Don’t know") %>%
  mutate(Q4.17 = droplevels(Q4.17),
         Q4.17 = factor(Q4.17, levels=rev(levels(Q4.17))))

df.employees.restrictions.plot.means <- df.employees.restriction %>%
  group_by(Q4.17, potential.contentiousness) %>%
  summarise(average = mean(Q3.4.num, na.rm=TRUE),
            med = median(Q3.4.num, na.rm=TRUE),
            num = n())

ggplot(df.employees.restriction, aes(y=Q4.17, x=Q3.4.num)) + 
  geom_violinh(na.rm=TRUE) +
  geom_point(alpha=0.2, size=0.5) +
  geom_point(data=df.employees.restrictions.plot.means,
             aes(x=med, y=Q4.17)) +
  scale_x_continuous(trans="log1p", breaks=c(0, 10^(0:5)), labels=comma) + 
  labs(x="Number of employees", y=NULL,
       title="Perceptions of restriction and # of employees") +
  theme_ath() + facet_wrap(~ potential.contentiousness)

## Warning: Removed 10 rows containing missing values (geom_point).

ggplot(df.employees.restrictions.plot.means, 
       aes(x=med, y=Q4.17, fill=potential.contentiousness)) +
  geom_barh(stat="identity", position="dodge") + 
  scale_x_continuous(expand=c(0, 0)) +
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  labs(x="Median number of employees", y=NULL, 
       title="Median # of employees and perceptions of restriction") +
  theme_ath()

Changes in programming (Q4.19)

df.employees.changes <- survey.clean.all %>%
  select(Q3.4.num, Q4.19, potential.contentiousness) %>%
  filter(Q4.19 != "Don't know") %>%
  mutate(Q4.19 = droplevels(Q4.19),
         Q4.19 = factor(Q4.19, levels=rev(levels(Q4.19))))

df.employees.changes.plot.means <- df.employees.changes %>%
  group_by(Q4.19, potential.contentiousness) %>%
  summarise(average = mean(Q3.4.num, na.rm=TRUE),
            med = median(Q3.4.num, na.rm=TRUE),
            num = n())

ggplot(df.employees.changes, aes(y=Q4.19, x=Q3.4.num)) + 
  geom_violinh(na.rm=TRUE) +
  geom_point(alpha=0.2, size=0.5) +
  geom_point(data=df.employees.changes.plot.means,
             aes(x=med, y=Q4.19)) +
  scale_x_continuous(trans="log1p", breaks=c(0, 10^(0:5)), labels=comma) + 
  labs(x="Number of employees", y=NULL,
       title="Changes in programming and # of employees") +
  theme_ath() + facet_wrap(~ potential.contentiousness)

## Warning: Removed 8 rows containing missing values (geom_point).

ggplot(df.employees.changes.plot.means, 
       aes(x=med, y=Q4.19, fill=potential.contentiousness)) +
  geom_barh(stat="identity", position="dodge") + 
  scale_x_continuous(expand=c(0, 0)) +
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  labs(x="Median number of employees", y=NULL, 
       title="Median # of employees and changes in programming") +
  theme_ath()

Attempts to change programming (Q4.23)

df.employees.change.attempt <- survey.clean.all %>%
  select(Q3.4.num, Q4.23, potential.contentiousness) %>%
  filter(Q4.23 != "Don't know") %>%
  mutate(Q4.23 = droplevels(Q4.23),
         Q4.23 = factor(Q4.23, levels=rev(levels(Q4.23))))

df.employees.change.attempt.plot.means <- df.employees.change.attempt %>%
  group_by(Q4.23, potential.contentiousness) %>%
  summarise(average = mean(Q3.4.num, na.rm=TRUE),
            med = median(Q3.4.num, na.rm=TRUE),
            num = n())

ggplot(df.employees.change.attempt, aes(y=Q4.23, x=Q3.4.num)) + 
  geom_violinh(na.rm=TRUE) +
  geom_point(alpha=0.2, size=0.5) +
  geom_point(data=df.employees.change.attempt.plot.means,
             aes(x=med, y=Q4.23)) +
  scale_x_continuous(trans="log1p", breaks=c(0, 10^(0:5)), labels=comma) + 
  labs(x="Number of employees", y=NULL,
       title="Changes in programming and # of employees") +
  theme_ath() + facet_wrap(~ potential.contentiousness)

## Warning: Removed 8 rows containing missing values (geom_point).

ggplot(df.employees.change.attempt.plot.means, 
       aes(x=med, y=Q4.23, fill=potential.contentiousness)) +
  geom_barh(stat="identity", position="dodge") + 
  scale_x_continuous(expand=c(0, 0)) +
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  labs(x="Median number of employees", y=NULL, 
       title="Median # of employees and changes in programming") +
  theme_ath()

Overall perception of restriction (Q4.17)

df.volunteers.restriction <- survey.clean.all %>%
  select(Q3.5.num, Q4.17, potential.contentiousness) %>%
  filter(Q4.17 != "Don’t know") %>%
  mutate(Q4.17 = droplevels(Q4.17),
         Q4.17 = factor(Q4.17, levels=rev(levels(Q4.17))))

df.volunteers.restrictions.plot.means <- df.volunteers.restriction %>%
  group_by(Q4.17, potential.contentiousness) %>%
  summarise(average = mean(Q3.5.num, na.rm=TRUE),
            med = median(Q3.5.num, na.rm=TRUE),
            num = n())

ggplot(df.volunteers.restriction, aes(y=Q4.17, x=Q3.5.num)) + 
  geom_violinh(na.rm=TRUE) +
  geom_point(alpha=0.2, size=0.5) +
  geom_point(data=df.volunteers.restrictions.plot.means,
             aes(x=med, y=Q4.17)) +
  scale_x_continuous(trans="log1p", breaks=c(0, 10^(0:5)), labels=comma) + 
  labs(x="Number of volunteers", y=NULL,
       title="Perceptions of restriction and # of volunteers") +
  theme_ath() + facet_wrap(~ potential.contentiousness)

## Warning: Removed 28 rows containing missing values (geom_point).

ggplot(df.volunteers.restrictions.plot.means, 
       aes(x=med, y=Q4.17, fill=potential.contentiousness)) +
  geom_barh(stat="identity", position="dodge") + 
  scale_x_continuous(expand=c(0, 0)) +
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  labs(x="Median number of volunteers", y=NULL, 
       title="Median # of volunteers and perceptions of restriction") +
  theme_ath()

# TODO: Staffing, collaboration, funding, etc.

# TODO: Figure out how to deal with org-level regime type analysis, since organizations work in multiple countries and answered only for one. Proportion of countries they work in that are autocracies?

Regime type

Most NGOs working in democracies have been there for 10+ years, while most NGOs working in autocracies have only been there for 1-4 years, and the differences between time worked in country across regime types are significantly different from expected values. NGOs that work in autocracies tend to have less of a legacy or history of working there, are possibly less likely to have a history of working with the government.

df.time.country.regime <- survey.countries.clean %>%
  select(Q4.2, target.regime.type) %>%
  filter(Q4.2 != "Don't know") %>%
  mutate(Q4.2 = droplevels(Q4.2),
         Q4.2 = factor(Q4.2, levels=rev(levels(Q4.2))))

plot.time.country.regime <- prodplot(df.time.country.regime,
                                     ~ target.regime.type + Q4.2, mosaic("h"), 
                                     colour=NA) + 
  aes(fill=target.regime.type, colour="white") + 
  scale_fill_manual(values=ath.palette("regime"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="Length of time in country, by regime type",
       subtitle="Q4.2: How long has your organization worked in `target_country`?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.time.country.regime

time.country.table <- survey.countries.clean %>%
  filter(Q4.2 != "Don't know") %>%
  mutate(Q4.2 = droplevels(Q4.2)) %>%
  xtabs(~ Q4.2 + target.regime.type, .)

analyze.cat.var(time.country.table)

## Table counts
## target.regime.type Democracy Autocracy
## Q4.2                                  
## Less than 1 year          34        20
## 1–4 years                108        77
## 5–9 years                 69        39
## 10 years or more         192        73
## 
## Expected values
##                   target.regime.type
## Q4.2               Democracy Autocracy
##   Less than 1 year  35.55882  18.44118
##   1–4 years        121.82190  63.17810
##   5–9 years         71.11765  36.88235
##   10 years or more 174.50163  90.49837
## 
## Row proporitions
## target.regime.type Democracy Autocracy
## Q4.2                                  
## Less than 1 year   0.6296296 0.3703704
## 1–4 years          0.5837838 0.4162162
## 5–9 years          0.6388889 0.3611111
## 10 years or more   0.7245283 0.2754717
## 
## Column proporitions
## target.regime.type  Democracy  Autocracy
## Q4.2                                    
## Less than 1 year   0.08436725 0.09569378
## 1–4 years          0.26799007 0.36842105
## 5–9 years          0.17121588 0.18660287
## 10 years or more   0.47642680 0.34928230
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test
## 
## data:  ftable(cat.table)
## X-squared = 10.115, df = 3, p-value = 0.01761
## 
## Cramer's V
## [1] 0.1285602
## 
## Pearson residuals
## 2 is used as critical value by convention
## target.regime.type  Democracy  Autocracy
## Q4.2                                    
## Less than 1 year   -0.2614106  0.3629967
## 1–4 years          -1.2522900  1.7389388
## 5–9 years          -0.2511105  0.3486938
## 10 years or more    1.3246396 -1.8394040
## 
## Components of chi-squared
## Critical value (0.05 with 3 df) is 7.81
## target.regime.type  Democracy  Autocracy
## Q4.2                                    
## Less than 1 year   0.06833552 0.13176658
## 1–4 years          1.56823035 3.02390827
## 5–9 years          0.06305649 0.12158739
## 10 years or more   1.75467018 3.38340709
## 
## p for components
## target.regime.type Democracy Autocracy
## Q4.2                                  
## Less than 1 year       0.995     0.988
## 1–4 years              0.667     0.388
## 5–9 years              0.996     0.989
## 10 years or more       0.625     0.336

Potential contentiousness

NGOs working on low contention issues tend to have worked in their respecitve target countries for a long time. High contention issue NGOs are more than expected / more likely to work in their target countries for 4 years or less. This may be because there’s a burst of more contentious INGOs being allowed, or that more contentious INGOs get kicked out more regularly and can only stay in country for so long.

df.time.country.issue <- survey.countries.clean %>%
  select(Q4.2, potential.contentiousness) %>%
  filter(Q4.2 != "Don't know") %>%
  mutate(Q4.2 = droplevels(Q4.2),
         Q4.2 = factor(Q4.2, levels=rev(levels(Q4.2))))

plot.time.country.issue <- prodplot(df.time.country.issue,
                                     ~ potential.contentiousness + Q4.2, mosaic("h"), 
                                     colour=NA) + 
  aes(fill=potential.contentiousness, colour="white") + 
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="Length of time in country, by potential contentiousness",
       subtitle="Q4.2: How long has your organization worked in `target_country`?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.time.country.issue

time.country.table.issue <- survey.countries.clean %>%
  filter(Q4.2 != "Don't know") %>%
  mutate(Q4.2 = droplevels(Q4.2)) %>%
  xtabs(~ Q4.2 + potential.contentiousness, .)

analyze.cat.var(time.country.table.issue)

## Table counts
## potential.contentiousness Low contention High contention
## Q4.2                                                    
## Less than 1 year                      35              19
## 1–4 years                            115              70
## 5–9 years                             78              30
## 10 years or more                     197              68
## 
## Expected values
##                   potential.contentiousness
## Q4.2               Low contention High contention
##   Less than 1 year        37.5000        16.50000
##   1–4 years              128.4722        56.52778
##   5–9 years               75.0000        33.00000
##   10 years or more       184.0278        80.97222
## 
## Row proporitions
## potential.contentiousness Low contention High contention
## Q4.2                                                    
## Less than 1 year               0.6481481       0.3518519
## 1–4 years                      0.6216216       0.3783784
## 5–9 years                      0.7222222       0.2777778
## 10 years or more               0.7433962       0.2566038
## 
## Column proporitions
## potential.contentiousness Low contention High contention
## Q4.2                                                    
## Less than 1 year              0.08235294      0.10160428
## 1–4 years                     0.27058824      0.37433155
## 5–9 years                     0.18352941      0.16042781
## 10 years or more              0.46352941      0.36363636
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test
## 
## data:  ftable(cat.table)
## X-squared = 8.5544, df = 3, p-value = 0.03584
## 
## Cramer's V
## [1] 0.1182277
## 
## Pearson residuals
## 2 is used as critical value by convention
## potential.contentiousness Low contention High contention
## Q4.2                                                    
## Less than 1 year              -0.4082483       0.6154575
## 1–4 years                     -1.1885970       1.7918774
## 5–9 years                      0.3464102      -0.5222330
## 10 years or more               0.9562527      -1.4416052
## 
## Components of chi-squared
## Critical value (0.05 with 3 df) is 7.81
## potential.contentiousness Low contention High contention
## Q4.2                                                    
## Less than 1 year               0.1666667       0.3787879
## 1–4 years                      1.4127628       3.2108245
## 5–9 years                      0.1200000       0.2727273
## 10 years or more               0.9144193       2.0782257
## 
## p for components
## potential.contentiousness Low contention High contention
## Q4.2                                                    
## Less than 1 year                   0.983           0.945
## 1–4 years                          0.703           0.360
## 5–9 years                          0.989           0.965
## 10 years or more                   0.822           0.556

Regime type + contentiousness

When accounting for both regime type and main issue area, an interesting story emerges. Individually, I found that INGOs working in their target countries for 5+ years were most likely to work on non-contentious issues and work in democracies. Relatively few of the long-term INGOs work in either autocracies or on more contentious issues. This remains the case when accounting for both target-country regime type and main issue area. There are fewer low contention, long-term INGOs working in autocracies than expected and more low contention, long term INGOs working in democracies.

df.time.country.issue.regime <- survey.countries.clean %>%
  select(Q4.2, potential.contentiousness, target.regime.type) %>%
  filter(Q4.2 != "Don't know") %>%
  mutate(Q4.2 = droplevels(Q4.2),
         Q4.2 = factor(Q4.2, levels=rev(levels(Q4.2))))

plot.time.country.issue.regime <- prodplot(df.time.country.issue.regime,
                                    ~ target.regime.type + potential.contentiousness +
                                      Q4.2, mosaic("v")) + 
  aes(fill=target.regime.type, linetype=potential.contentiousness) + 
  scale_fill_manual(values=ath.palette("regime")) +
  scale_linetype_manual(values=c("blank", "dashed")) +
  guides(fill=FALSE, linetype=FALSE) +
  labs(title="Length of time in country",
       subtitle="Issue area of INGO + regime type of target country") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.time.country.issue.regime

time.country.table.issue.regime <- survey.countries.clean %>%
  filter(Q4.2 != "Don't know") %>%
  mutate(Q4.2 = droplevels(Q4.2)) %>%
  xtabs(~ Q4.2 + potential.contentiousness + target.regime.type, .)

analyze.cat.var(time.country.table.issue.regime)

## Table counts
##                  target.regime.type        Democracy Autocracy
## Q4.2             potential.contentiousness                    
## Less than 1 year Low contention                   22        13
##                  High contention                  12         7
## 1–4 years        Low contention                   71        44
##                  High contention                  37        33
## 5–9 years        Low contention                   50        28
##                  High contention                  19        11
## 10 years or more Low contention                  153        44
##                  High contention                  39        29
## 
## Expected values
##           [,1]      [,2]
## [1,]  23.04739 11.952614
## [2,]  12.51144  6.488562
## [3,]  75.72712 39.272876
## [4,]  46.09477 23.905229
## [5,]  51.36275 26.637255
## [6,]  19.75490 10.245098
## [7,] 129.72386 67.276144
## [8,]  44.77778 23.222222
## 
## Row proporitions
##                  target.regime.type        Democracy Autocracy
## Q4.2             potential.contentiousness                    
## Less than 1 year Low contention            0.4074074 0.2407407
##                  High contention           0.2222222 0.1296296
## 1–4 years        Low contention            0.3837838 0.2378378
##                  High contention           0.2000000 0.1783784
## 5–9 years        Low contention            0.4629630 0.2592593
##                  High contention           0.1759259 0.1018519
## 10 years or more Low contention            0.5773585 0.1660377
##                  High contention           0.1471698 0.1094340
## 
## Column proporitions
##                  target.regime.type         Democracy  Autocracy
## Q4.2             potential.contentiousness                      
## Less than 1 year Low contention            0.05176471 0.03058824
##                  High contention           0.06417112 0.03743316
## 1–4 years        Low contention            0.16705882 0.10352941
##                  High contention           0.19786096 0.17647059
## 5–9 years        Low contention            0.11764706 0.06588235
##                  High contention           0.10160428 0.05882353
## 10 years or more Low contention            0.36000000 0.10352941
##                  High contention           0.20855615 0.15508021
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test
## 
## data:  ftable(cat.table)
## X-squared = 20.922, df = 7, p-value = 0.003887
## 
## Cramer's V
## [1] 0.1848957
## 
## Pearson residuals
## 2 is used as critical value by convention
##                  target.regime.type         Democracy  Autocracy
## Q4.2             potential.contentiousness                      
## Less than 1 year Low contention            -0.2181704  0.3029529
##                  High contention           -0.1445903  0.2007792
## 1–4 years        Low contention            -0.5432144  0.7543114
##                  High contention           -1.3395717  1.8601387
## 5–9 years        Low contention            -0.1901475  0.2640401
##                  High contention           -0.1698451  0.2358482
## 10 years or more Low contention             2.0436245 -2.8377915
##                  High contention           -0.8634348  1.1989717
## 
## Components of chi-squared
## Critical value (0.05 with 7 df) is 14.07
##                  target.regime.type         Democracy  Autocracy
## Q4.2             potential.contentiousness                      
## Less than 1 year Low contention            0.04759831 0.09178048
##                  High contention           0.02090637 0.04031228
## 1–4 years        Low contention            0.29508189 0.56898566
##                  High contention           1.79445221 3.46011598
## 5–9 years        Low contention            0.03615605 0.06971718
##                  High contention           0.02884737 0.05562436
## 10 years or more Low contention            4.17640121 8.05306070
##                  High contention           0.74551971 1.43753323
## 
## p for components
##                  target.regime.type        Democracy Autocracy
## Q4.2             potential.contentiousness                    
## Less than 1 year Low contention                1.000     1.000
##                  High contention               1.000     1.000
## 1–4 years        Low contention                1.000     0.999
##                  High contention               0.970     0.839
## 5–9 years        Low contention                1.000     1.000
##                  High contention               1.000     1.000
## 10 years or more Low contention                0.759     0.328
##                  High contention               0.998     0.984

Regime type

NGOs in autocracies definitely pursue different operational strategies. The most common strategy for these NGOs is to provide funding to domestic NGOs, while the least common is to maintain an office staffed by foreigners (difference is statistically significant). International NGOs seem to be more likely to take a hands off approach to advocacy in target countries that are autocracies.

what.do <- c("Maintain a physical office staffed primarily by foreigners",
             "Maintain a physical office staffed primarily by people from target_country",
             "Provide funding to domestic NGOs", "Partner with domestic NGOs")
what.do.short <- c("Maintain a physical office staffed\nprimarily by foreigners",
                   "Maintain a physical office staffed\nprimarily by people from target_country",
                   "Provide funding to domestic NGOs", "Partner with domestic NGOs")

df.operations.regime <- survey.countries.clean %>%
  unnest(Q4.3_value) %>%
  select(Q4.3_value, target.regime.type) %>%
  filter(!is.na(Q4.3_value), Q4.3_value != "Don't know") %>%
  mutate(Q4.3 = factor(Q4.3_value, levels=what.do, 
                       labels=what.do.short, ordered=TRUE))

plot.operations.regime <- prodplot(df.operations.regime,
                                     ~ target.regime.type + Q4.3, mosaic("h"), 
                                     colour=NA) + 
  aes(fill=target.regime.type, colour="white") + 
  scale_fill_manual(values=ath.palette("regime"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="How INGOs work in the target country, by regime type",
       subtitle="Q4.3: What does your organization do in `target_country`?\n(multiple answers allowed)") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.operations.regime

operations.table <- survey.countries.clean %>%
  unnest(Q4.3_value) %>%
  filter(Q4.3_value != "Don't know") %>%
  xtabs(~ Q4.3_value + target.regime.type, .)

analyze.cat.var(operations.table)

## Table counts
## target.regime.type                                                         Democracy Autocracy
## Q4.3_value                                                                                    
## Maintain a physical office staffed primarily by foreigners                        54        15
## Maintain a physical office staffed primarily by people from target_country       158        93
## Partner with domestic NGOs                                                       235       139
## Provide funding to domestic NGOs                                                  35        33
## 
## Expected values
##                                                                             target.regime.type
## Q4.3_value                                                                   Democracy Autocracy
##   Maintain a physical office staffed primarily by foreigners                  43.64567  25.35433
##   Maintain a physical office staffed primarily by people from target_country 158.76903  92.23097
##   Partner with domestic NGOs                                                 236.57218 137.42782
##   Provide funding to domestic NGOs                                            43.01312  24.98688
## 
## Row proporitions
## target.regime.type                                                         Democracy Autocracy
## Q4.3_value                                                                                    
## Maintain a physical office staffed primarily by foreigners                 0.7826087 0.2173913
## Maintain a physical office staffed primarily by people from target_country 0.6294821 0.3705179
## Partner with domestic NGOs                                                 0.6283422 0.3716578
## Provide funding to domestic NGOs                                           0.5147059 0.4852941
## 
## Column proporitions
## target.regime.type                                                          Democracy  Autocracy
## Q4.3_value                                                                                      
## Maintain a physical office staffed primarily by foreigners                 0.11203320 0.05357143
## Maintain a physical office staffed primarily by people from target_country 0.32780083 0.33214286
## Partner with domestic NGOs                                                 0.48755187 0.49642857
## Provide funding to domestic NGOs                                           0.07261411 0.11785714
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test
## 
## data:  ftable(cat.table)
## X-squared = 10.786, df = 3, p-value = 0.01294
## 
## Cramer's V
## [1] 0.1189748
## 
## Pearson residuals
## 2 is used as critical value by convention
## target.regime.type                                                           Democracy   Autocracy
## Q4.3_value                                                                                        
## Maintain a physical office staffed primarily by foreigners                  1.56729754 -2.05634489
## Maintain a physical office staffed primarily by people from target_country -0.06103230  0.08007635
## Partner with domestic NGOs                                                 -0.10221627  0.13411104
## Provide funding to domestic NGOs                                           -1.22180343  1.60304547
## 
## Components of chi-squared
## Critical value (0.05 with 3 df) is 7.81
## target.regime.type                                                           Democracy   Autocracy
## Q4.3_value                                                                                        
## Maintain a physical office staffed primarily by foreigners                 2.456421592 4.228554311
## Maintain a physical office staffed primarily by people from target_country 0.003724942 0.006412221
## Partner with domestic NGOs                                                 0.010448165 0.017985770
## Provide funding to domestic NGOs                                           1.492803613 2.569754792
## 
## p for components
## target.regime.type                                                         Democracy Autocracy
## Q4.3_value                                                                                    
## Maintain a physical office staffed primarily by foreigners                     0.483     0.238
## Maintain a physical office staffed primarily by people from target_country     1.000     1.000
## Partner with domestic NGOs                                                     1.000     0.999
## Provide funding to domestic NGOs                                               0.684     0.463

Potential contentiousness

There’s an overall significant difference in frequencies, driven primarily by more high contention INGOs working with foreigners. The individual cell effects, however, aren’t particularly significant. In general, more INGOs than expected use foreigners in their target countries).

df.operations.issue <- survey.countries.clean %>%
  unnest(Q4.3_value) %>%
  select(Q4.3_value, potential.contentiousness) %>%
  filter(!is.na(Q4.3_value), Q4.3_value != "Don't know") %>%
  mutate(Q4.3 = factor(Q4.3_value, levels=what.do, 
                       labels=what.do.short, ordered=TRUE))

plot.operations.issue <- prodplot(df.operations.issue,
                                  ~ potential.contentiousness + Q4.3, mosaic("h"), 
                                  colour=NA) + 
  aes(fill=potential.contentiousness, colour="white") + 
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="How INGOs work in the target country, by issue",
       subtitle="Q4.3: What does your organization do in `target_country`?\n(multiple answers allowed)") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.operations.issue

operations.table.issue <- survey.countries.clean %>%
  unnest(Q4.3_value) %>%
  filter(Q4.3_value != "Don't know") %>%
  xtabs(~ Q4.3_value + potential.contentiousness, .)

analyze.cat.var(operations.table.issue)

## Table counts
## potential.contentiousness                                                  Low contention High contention
## Q4.3_value                                                                                               
## Maintain a physical office staffed primarily by foreigners                             43              26
## Maintain a physical office staffed primarily by people from target_country            176              75
## Partner with domestic NGOs                                                            262             112
## Provide funding to domestic NGOs                                                       45              23
## 
## Expected values
##                                                                             potential.contentiousness
## Q4.3_value                                                                   Low contention High contention
##   Maintain a physical office staffed primarily by foreigners                       47.62992        21.37008
##   Maintain a physical office staffed primarily by people from target_country      173.26247        77.73753
##   Partner with domestic NGOs                                                      258.16798       115.83202
##   Provide funding to domestic NGOs                                                 46.93963        21.06037
## 
## Row proporitions
## potential.contentiousness                                                  Low contention High contention
## Q4.3_value                                                                                               
## Maintain a physical office staffed primarily by foreigners                      0.6231884       0.3768116
## Maintain a physical office staffed primarily by people from target_country      0.7011952       0.2988048
## Partner with domestic NGOs                                                      0.7005348       0.2994652
## Provide funding to domestic NGOs                                                0.6617647       0.3382353
## 
## Column proporitions
## potential.contentiousness                                                  Low contention High contention
## Q4.3_value                                                                                               
## Maintain a physical office staffed primarily by foreigners                     0.08174905      0.11016949
## Maintain a physical office staffed primarily by people from target_country     0.33460076      0.31779661
## Partner with domestic NGOs                                                     0.49809886      0.47457627
## Provide funding to domestic NGOs                                               0.08555133      0.09745763
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test
## 
## data:  ftable(cat.table)
## X-squared = 2.0352, df = 3, p-value = 0.5651
## 
## Cramer's V
## [1] 0.05168098
## 
## Pearson residuals
## 2 is used as critical value by convention
## potential.contentiousness                                                  Low contention High contention
## Q4.3_value                                                                                               
## Maintain a physical office staffed primarily by foreigners                     -0.6708627       1.0015452
## Maintain a physical office staffed primarily by people from target_country      0.2079731      -0.3104874
## Partner with domestic NGOs                                                      0.2384936      -0.3560521
## Provide funding to domestic NGOs                                               -0.2831064       0.4226555
## 
## Components of chi-squared
## Critical value (0.05 with 3 df) is 7.81
## potential.contentiousness                                                  Low contention High contention
## Q4.3_value                                                                                               
## Maintain a physical office staffed primarily by foreigners                     0.45005682      1.00309274
## Maintain a physical office staffed primarily by people from target_country     0.04325279      0.09640241
## Partner with domestic NGOs                                                     0.05687919      0.12677310
## Provide funding to domestic NGOs                                               0.08014921      0.17863764
## 
## p for components
## potential.contentiousness                                                  Low contention High contention
## Q4.3_value                                                                                               
## Maintain a physical office staffed primarily by foreigners                          0.930           0.801
## Maintain a physical office staffed primarily by people from target_country          0.998           0.992
## Partner with domestic NGOs                                                          0.996           0.988
## Provide funding to domestic NGOs                                                    0.994           0.981

Regime type

There’s a slight difference in how NGOs register across regimes, with more NGOs registering in autocracies than in democracies, perhaps because they are more likely to be required to register in autocracies. The difference is not significant, though.

df.registered.regime <- survey.countries.clean %>%
  select(Q4.4, target.regime.type) %>%
  filter(Q4.4 != "Don't know") %>%
  mutate(Q4.4 = droplevels(Q4.4),
         Q4.4 = factor(Q4.4, levels=rev(levels(Q4.4))))

plot.registered.regime <- prodplot(df.registered.regime,
                                   ~ target.regime.type + Q4.4, mosaic("h"), 
                                   colour=NA) + 
  aes(fill=target.regime.type, colour="white") + 
  scale_fill_manual(values=ath.palette("regime"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="Registration status, by regime type",
       subtitle="Q4.4: Is your organization registered with the national government in `target_country`?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.registered.regime

registered.table <- survey.countries.clean %>%
  filter(Q4.4 != "Don't know") %>%
  mutate(Q4.4 = droplevels(Q4.4)) %>%
  xtabs(~ Q4.4 + target.regime.type, .)

analyze.cat.var(registered.table)

## Table counts
## target.regime.type Democracy Autocracy
## Q4.4                                  
## Yes                      180       105
## No                       198        98
## 
## Expected values
##      target.regime.type
## Q4.4  Democracy Autocracy
##   Yes  185.4217  99.57831
##   No   192.5783 103.42169
## 
## Row proporitions
## target.regime.type Democracy Autocracy
## Q4.4                                  
## Yes                0.6315789 0.3684211
## No                 0.6689189 0.3310811
## 
## Column proporitions
## target.regime.type Democracy Autocracy
## Q4.4                                  
## Yes                0.4761905 0.5172414
## No                 0.5238095 0.4827586
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)
## X-squared = 0.73389, df = 1, p-value = 0.3916
## 
## Cramer's V
## [1] 0.03915149
## 
## Pearson residuals
## 2 is used as critical value by convention
## target.regime.type  Democracy  Autocracy
## Q4.4                                    
## Yes                -0.3981568  0.5433154
## No                  0.3906886 -0.5331245
## 
## Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84
## target.regime.type Democracy Autocracy
## Q4.4                                  
## Yes                0.1585289 0.2951917
## No                 0.1526376 0.2842217
## 
## p for components
## target.regime.type Democracy Autocracy
## Q4.4                                  
## Yes                    0.691     0.587
## No                     0.696     0.594

Potential contentiousness

There is no significant difference in the registration status of low and high contentious INGOs.

df.registered.issue <- survey.countries.clean %>%
  select(Q4.4, potential.contentiousness) %>%
  filter(Q4.4 != "Don't know") %>%
  mutate(Q4.4 = droplevels(Q4.4),
         Q4.4 = factor(Q4.4, levels=rev(levels(Q4.4))))

plot.registered.issue <- prodplot(df.registered.issue,
                                  ~ potential.contentiousness + Q4.4, mosaic("h"), 
                                  colour=NA) + 
  aes(fill=potential.contentiousness, colour="white") + 
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="Registration status, by issue",
       subtitle="Q4.4: Is your organization registered with the national government in `target_country`?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.registered.issue

registered.table.issue <- survey.countries.clean %>%
  filter(Q4.4 != "Don't know") %>%
  mutate(Q4.4 = droplevels(Q4.4)) %>%
  xtabs(~ Q4.4 + potential.contentiousness, .)

analyze.cat.var(registered.table.issue)

## Table counts
## potential.contentiousness Low contention High contention
## Q4.4                                                    
## Yes                                  202              83
## No                                   199              97
## 
## Expected values
##      potential.contentiousness
## Q4.4  Low contention High contention
##   Yes        196.704        88.29604
##   No         204.296        91.70396
## 
## Row proporitions
## potential.contentiousness Low contention High contention
## Q4.4                                                    
## Yes                            0.7087719       0.2912281
## No                             0.6722973       0.3277027
## 
## Column proporitions
## potential.contentiousness Low contention High contention
## Q4.4                                                    
## Yes                            0.5037406       0.4611111
## No                             0.4962594       0.5388889
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)
## X-squared = 0.74087, df = 1, p-value = 0.3894
## 
## Cramer's V
## [1] 0.03943218
## 
## Pearson residuals
## 2 is used as critical value by convention
## potential.contentiousness Low contention High contention
## Q4.4                                                    
## Yes                            0.3776112      -0.5636127
## No                            -0.3705283       0.5530410
## 
## Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84
## potential.contentiousness Low contention High contention
## Q4.4                                                    
## Yes                            0.1425902       0.3176592
## No                             0.1372912       0.3058543
## 
## p for components
## potential.contentiousness Low contention High contention
## Q4.4                                                    
## Yes                                0.706           0.573
## No                                 0.711           0.580

Regime type

Most NGOs don’t know, and I didn’t include an “other” category here. The univariate distribution has a clear trend, with most reporting “Rarely” (and only a few “Never”; “Never” and “Once a month” are collapsed because of low expected values). There’s also a trend by regime type—more INGOs working in autocracies see annual changes in regulations, possibly reflecting a more volatile regulatory environment.

plot.reg.change.regime <- prodplot(df.freq.change,
                                        ~ target.regime.type + Q4.14.collapsed, mosaic("h"),
                                        colour=NA) +
  aes(fill=target.regime.type, colour="white") +
  scale_fill_manual(values=ath.palette("regime"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="Frequency of changes, by regime type",
       subtitle="Q4.14: How often do regulations for international NGOs in `target_country` change?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.reg.change.regime

reg.change.regime.table <- df.freq.change %>%
  xtabs(~ Q4.14.collapsed + target.regime.type, .)

analyze.cat.var(reg.change.regime.table)

## Table counts
## target.regime.type   Democracy Autocracy
## Q4.14.collapsed                         
## Don't know                 203        83
## Rarely or never             89        45
## Once every few years        65        44
## Once a year+                23        31
## 
## Expected values
##                       target.regime.type
## Q4.14.collapsed        Democracy Autocracy
##   Don't know           186.41509  99.58491
##   Rarely or never       87.34134  46.65866
##   Once every few years  71.04631  37.95369
##   Once a year+          35.19726  18.80274
## 
## Row proporitions
## target.regime.type   Democracy Autocracy
## Q4.14.collapsed                         
## Don't know           0.7097902 0.2902098
## Rarely or never      0.6641791 0.3358209
## Once every few years 0.5963303 0.4036697
## Once a year+         0.4259259 0.5740741
## 
## Column proporitions
## target.regime.type    Democracy  Autocracy
## Q4.14.collapsed                           
## Don't know           0.53421053 0.40886700
## Rarely or never      0.23421053 0.22167488
## Once every few years 0.17105263 0.21674877
## Once a year+         0.06052632 0.15270936
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test
## 
## data:  ftable(cat.table)
## X-squared = 17.945, df = 3, p-value = 0.0004515
## 
## Cramer's V
## [1] 0.1754434
## 
## Pearson residuals
## 2 is used as critical value by convention
## target.regime.type    Democracy  Autocracy
## Q4.14.collapsed                           
## Don't know            1.2147096 -1.6619435
## Rarely or never       0.1774794 -0.2428241
## Once every few years -0.7173313  0.9814396
## Once a year+         -2.0559272  2.8128819
## 
## Components of chi-squared
## Critical value (0.05 with 3 df) is 7.81
## target.regime.type    Democracy  Autocracy
## Q4.14.collapsed                           
## Don't know           1.47551944 2.76205610
## Rarely or never      0.03149894 0.05896354
## Once every few years 0.51456423 0.96322368
## Once a year+         4.22683647 7.91230472
## 
## p for components
## target.regime.type   Democracy Autocracy
## Q4.14.collapsed                         
## Don't know               0.688     0.430
## Rarely or never          0.999     0.996
## Once every few years     0.916     0.810
## Once a year+             0.238     0.048

Potential contentiousness

There’s no difference in the frequency of changes across issue areas, which is to be expected. Actual legal restrictions are a blunt instrument and don’t really target specific sectors of INGOs.

plot.reg.change.issue <- prodplot(df.freq.change,
                                  ~ potential.contentiousness + Q4.14.collapsed, mosaic("h"),
                                  colour=NA) +
  aes(fill=potential.contentiousness, colour="white") +
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="Frequency of changes, by issue",
       subtitle="Q4.14: How often do regulations for international NGOs in `target_country` change?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.reg.change.issue

reg.change.issue.table <- df.freq.change %>%
  xtabs(~ Q4.14.collapsed + potential.contentiousness, .)

analyze.cat.var(reg.change.issue.table)

## Table counts
## potential.contentiousness Low contention High contention
## Q4.14.collapsed                                         
## Don't know                           195              91
## Rarely or never                       88              46
## Once every few years                  83              26
## Once a year+                          35              19
## 
## Expected values
##                       potential.contentiousness
## Q4.14.collapsed        Low contention High contention
##   Don't know                196.71698        89.28302
##   Rarely or never            92.16810        41.83190
##   Once every few years       74.97256        34.02744
##   Once a year+               37.14237        16.85763
## 
## Row proporitions
## potential.contentiousness Low contention High contention
## Q4.14.collapsed                                         
## Don't know                     0.6818182       0.3181818
## Rarely or never                0.6567164       0.3432836
## Once every few years           0.7614679       0.2385321
## Once a year+                   0.6481481       0.3518519
## 
## Column proporitions
## potential.contentiousness Low contention High contention
## Q4.14.collapsed                                         
## Don't know                     0.4862843       0.5000000
## Rarely or never                0.2194514       0.2527473
## Once every few years           0.2069825       0.1428571
## Once a year+                   0.0872818       0.1043956
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test
## 
## data:  ftable(cat.table)
## X-squared = 3.8009, df = 3, p-value = 0.2838
## 
## Cramer's V
## [1] 0.0807439
## 
## Pearson residuals
## 2 is used as critical value by convention
## potential.contentiousness Low contention High contention
## Q4.14.collapsed                                         
## Don't know                    -0.1224178       0.1817109
## Rarely or never               -0.4341576       0.6444421
## Once every few years           0.9270991      -1.3761400
## Once a year+                  -0.3515273       0.5217898
## 
## Components of chi-squared
## Critical value (0.05 with 3 df) is 7.81
## potential.contentiousness Low contention High contention
## Q4.14.collapsed                                         
## Don't know                    0.01498612      0.03301887
## Rarely or never               0.18849282      0.41530562
## Once every few years          0.85951267      1.89376142
## Once a year+                  0.12357146      0.27226460
## 
## p for components
## potential.contentiousness Low contention High contention
## Q4.14.collapsed                                         
## Don't know                         1.000           0.998
## Rarely or never                    0.979           0.937
## Once every few years               0.835           0.595
## Once a year+                       0.989           0.965

Regime type

!!! It works !!!

INGOs that are restricted tend to work in autocracies and there’s almost a perfect trend.

df.reg.effect.general.regime <- survey.countries.clean %>%
  select(Q4.17, target.regime.type) %>%
  filter(Q4.17 != "Don’t know") %>%
  mutate(Q4.17 = droplevels(Q4.17),
         Q4.17 = factor(Q4.17, levels=rev(levels(Q4.17))))

plot.reg.effect.general.regime <- prodplot(df.reg.effect.general.regime,
                                   ~ target.regime.type + Q4.17, mosaic("h"),
                                   colour=NA) +
  aes(fill=target.regime.type, colour="white") +
  scale_fill_manual(values=ath.palette("regime"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="General restrictions, by regime type",
       subtitle="Q4.17: Overall, how is your organization's work affected by government regulations in `target_country`?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.reg.effect.general.regime

reg.effect.general.regime.table <- survey.countries.clean %>%
  xtabs(~ Q4.17 + target.regime.type, .)

analyze.cat.var(reg.effect.general.regime.table)

## Table counts
## target.regime.type    Democracy Autocracy
## Q4.17                                    
## Not restricted at all       167        54
## Slightly restricted          78        55
## Moderately restricted        41        39
## Very restricted              12        16
## Extremely restricted          5        12
## Don’t know                   56        17
## 
## Expected values
##                        target.regime.type
## Q4.17                   Democracy Autocracy
##   Not restricted at all 143.73007 77.269928
##   Slightly restricted    86.49819 46.501812
##   Moderately restricted  52.02899 27.971014
##   Very restricted        18.21014  9.789855
##   Extremely restricted   11.05616  5.943841
##   Don’t know             47.47645 25.523551
## 
## Row proporitions
## target.regime.type    Democracy Autocracy
## Q4.17                                    
## Not restricted at all 0.7556561 0.2443439
## Slightly restricted   0.5864662 0.4135338
## Moderately restricted 0.5125000 0.4875000
## Very restricted       0.4285714 0.5714286
## Extremely restricted  0.2941176 0.7058824
## Don’t know            0.7671233 0.2328767
## 
## Column proporitions
## target.regime.type     Democracy  Autocracy
## Q4.17                                      
## Not restricted at all 0.46518106 0.27979275
## Slightly restricted   0.21727019 0.28497409
## Moderately restricted 0.11420613 0.20207254
## Very restricted       0.03342618 0.08290155
## Extremely restricted  0.01392758 0.06217617
## Don’t know            0.15598886 0.08808290
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test
## 
## data:  ftable(cat.table)
## X-squared = 39.772, df = 5, p-value = 1.66e-07
## 
## Cramer's V
## [1] 0.2684213
## 
## Pearson residuals
## 2 is used as critical value by convention
## target.regime.type     Democracy  Autocracy
## Q4.17                                      
## Not restricted at all  1.9409807 -2.6472184
## Slightly restricted   -0.9137404  1.2462105
## Moderately restricted -1.5290190  2.0853620
## Very restricted       -1.4552749  1.9847856
## Extremely restricted  -1.8213573  2.4840694
## Don’t know             1.2370334 -1.6871357
## 
## Components of chi-squared
## Critical value (0.05 with 5 df) is 11.07
## target.regime.type    Democracy Autocracy
## Q4.17                                    
## Not restricted at all 3.7674059 7.0077654
## Slightly restricted   0.8349216 1.5530407
## Moderately restricted 2.3378992 4.3487347
## Very restricted       2.1178250 3.9393739
## Extremely restricted  3.3173424 6.1706007
## Don’t know            1.5302517 2.8464267
## 
## p for components
## target.regime.type    Democracy Autocracy
## Q4.17                                    
## Not restricted at all     0.583     0.220
## Slightly restricted       0.975     0.907
## Moderately restricted     0.801     0.500
## Very restricted           0.833     0.558
## Extremely restricted      0.651     0.290
## Don’t know                0.910     0.724

Potential contentiousness

Differences in general restrictions by issue are less pronounced. There is a significant overall difference, driven primarily by high contention INGOs that report feeling extremely restricted. All levels of restriction are relatively even, following a 65-35 split between low and high contention INGOs.

df.reg.effect.general.issue <- survey.countries.clean %>%
  select(Q4.17, potential.contentiousness) %>%
  filter(Q4.17 != "Don’t know") %>%
  mutate(Q4.17 = droplevels(Q4.17),
         Q4.17 = factor(Q4.17, levels=rev(levels(Q4.17))))

plot.reg.effect.general.issue <- prodplot(df.reg.effect.general.issue,
                                          ~ potential.contentiousness + Q4.17, mosaic("h"),
                                          colour=NA) +
  aes(fill=potential.contentiousness, colour="white") +
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="General restrictions, by issue",
       subtitle="Q4.17: Overall, how is your organization's work affected by government regulations in `target_country`?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.reg.effect.general.issue

reg.effect.general.issue.table <- survey.countries.clean %>%
  xtabs(~ Q4.17 + potential.contentiousness, .)

analyze.cat.var(reg.effect.general.issue.table)

## Table counts
## potential.contentiousness Low contention High contention
## Q4.17                                                   
## Not restricted at all                162              59
## Slightly restricted                   91              42
## Moderately restricted                 59              21
## Very restricted                       18              10
## Extremely restricted                   6              11
## Don’t know                            47              26
## 
## Expected values
##                        potential.contentiousness
## Q4.17                   Low contention High contention
##   Not restricted at all      153.33877       67.661232
##   Slightly restricted         92.28080       40.719203
##   Moderately restricted       55.50725       24.492754
##   Very restricted             19.42754        8.572464
##   Extremely restricted        11.79529        5.204710
##   Don’t know                  50.65036       22.349638
## 
## Row proporitions
## potential.contentiousness Low contention High contention
## Q4.17                                                   
## Not restricted at all          0.7330317       0.2669683
## Slightly restricted            0.6842105       0.3157895
## Moderately restricted          0.7375000       0.2625000
## Very restricted                0.6428571       0.3571429
## Extremely restricted           0.3529412       0.6470588
## Don’t know                     0.6438356       0.3561644
## 
## Column proporitions
## potential.contentiousness Low contention High contention
## Q4.17                                                   
## Not restricted at all         0.42297650      0.34911243
## Slightly restricted           0.23759791      0.24852071
## Moderately restricted         0.15404700      0.12426036
## Very restricted               0.04699739      0.05917160
## Extremely restricted          0.01566580      0.06508876
## Don’t know                    0.12271540      0.15384615
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test
## 
## data:  ftable(cat.table)
## X-squared = 12.876, df = 5, p-value = 0.02457
## 
## Cramer's V
## [1] 0.1527288
## 
## Pearson residuals
## 2 is used as critical value by convention
## potential.contentiousness Low contention High contention
## Q4.17                                                   
## Not restricted at all          0.6994452      -1.0529548
## Slightly restricted           -0.1333290       0.2007154
## Moderately restricted          0.4688059      -0.7057472
## Very restricted               -0.3238756       0.4875670
## Extremely restricted          -1.6874109       2.5402525
## Don’t know                    -0.5129142       0.7721483
## 
## Components of chi-squared
## Critical value (0.05 with 5 df) is 11.07
## potential.contentiousness Low contention High contention
## Q4.17                                                   
## Not restricted at all         0.48922356      1.10871374
## Slightly restricted           0.01777663      0.04028667
## Moderately restricted         0.21977901      0.49807907
## Very restricted               0.10489543      0.23772159
## Extremely restricted          2.84735559      6.45288279
## Don’t know                    0.26308094      0.59621302
## 
## p for components
## potential.contentiousness Low contention High contention
## Q4.17                                                   
## Not restricted at all              0.993           0.953
## Slightly restricted                1.000           1.000
## Moderately restricted              0.999           0.992
## Very restricted                    1.000           0.999
## Extremely restricted               0.724           0.265
## Don’t know                         0.998           0.988

Regime type + contentiousness

There are more unrestricted low contentious NGOs working in democracies than expected and far fewer unrestricted low contentious NGOs in autocracies. Additionally, there are fewer highly restricted INGOs in democracies (especially high contentious ones), and far more highly restricted, highly contentious INGOs working in autocracies than expected.

# Collapse very and extremely restriced because of low expected frequencies
restricted <- c("Not restricted at all", "Slightly restricted",
                "Moderately restricted", "Very or extremely restricted")

df.reg.effect.general.issue.regime <- survey.countries.clean %>%
  select(Q4.17, potential.contentiousness, target.regime.type) %>%
  filter(Q4.17 != "Don’t know") %>%
  mutate(Q4.17 = recode_factor(Q4.17, `Very restricted` = restricted[4],
                        `Extremely restricted` = restricted[4]),
         Q4.17 = factor(Q4.17, levels=restricted, ordered=TRUE))

plot.reg.effect.general.issue.regime <- prodplot(df.reg.effect.general.issue.regime,
                                          ~ target.regime.type + potential.contentiousness +
                                            Q4.17, mosaic("v")) +
  aes(fill=target.regime.type, linetype=potential.contentiousness) + 
  scale_fill_manual(values=ath.palette("regime")) +
  scale_linetype_manual(values=c("blank", "dashed")) +
  guides(fill=FALSE, linetype=FALSE) +
  labs(title="General restrictions, by issue and regime type",
       subtitle="Q4.17: Overall, how is your organization's work affected by government regulations in `target_country`?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.reg.effect.general.issue.regime

reg.effect.general.regime.issue.table <- df.reg.effect.general.issue.regime %>%
  xtabs(~ Q4.17 + potential.contentiousness + target.regime.type, .)

analyze.cat.var(reg.effect.general.regime.issue.table)

## Table counts
##                              target.regime.type        Democracy Autocracy
## Q4.17                        potential.contentiousness                    
## Not restricted at all        Low contention                  123        39
##                              High contention                  44        15
## Slightly restricted          Low contention                   57        34
##                              High contention                  21        21
## Moderately restricted        Low contention                   32        27
##                              High contention                   9        12
## Very or extremely restricted Low contention                   12        12
##                              High contention                   5        16
## 
## Expected values
##           [,1]      [,2]
## [1,] 102.47599 59.524008
## [2,]  37.32150 21.678497
## [3,]  57.56367 33.436326
## [4,]  26.56785 15.432150
## [5,]  37.32150 21.678497
## [6,]  13.28392  7.716075
## [7,]  15.18163  8.818372
## [8,]  13.28392  7.716075
## 
## Row proporitions
##                              target.regime.type        Democracy Autocracy
## Q4.17                        potential.contentiousness                    
## Not restricted at all        Low contention            0.5565611 0.1764706
##                              High contention           0.1990950 0.0678733
## Slightly restricted          Low contention            0.4285714 0.2556391
##                              High contention           0.1578947 0.1578947
## Moderately restricted        Low contention            0.4000000 0.3375000
##                              High contention           0.1125000 0.1500000
## Very or extremely restricted Low contention            0.2666667 0.2666667
##                              High contention           0.1111111 0.3555556
## 
## Column proporitions
##                              target.regime.type         Democracy  Autocracy
## Q4.17                        potential.contentiousness                      
## Not restricted at all        Low contention            0.36607143 0.11607143
##                              High contention           0.30769231 0.10489510
## Slightly restricted          Low contention            0.16964286 0.10119048
##                              High contention           0.14685315 0.14685315
## Moderately restricted        Low contention            0.09523810 0.08035714
##                              High contention           0.06293706 0.08391608
## Very or extremely restricted Low contention            0.03571429 0.03571429
##                              High contention           0.03496503 0.11188811
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test
## 
## data:  ftable(cat.table)
## X-squared = 39.33, df = 7, p-value = 1.691e-06
## 
## Cramer's V
## [1] 0.2865449
## 
## Pearson residuals
## 2 is used as critical value by convention
##                              target.regime.type          Democracy   Autocracy
## Q4.17                        potential.contentiousness                        
## Not restricted at all        Low contention             2.02745451 -2.66021108
##                              High contention            1.09319884 -1.43437974
## Slightly restricted          Low contention            -0.07429404  0.09748078
##                              High contention           -1.08021275  1.41734077
## Moderately restricted        Low contention            -0.87107341  1.14293028
##                              High contention           -1.17538092  1.54221037
## Very or extremely restricted Low contention            -0.81656409  1.07140891
##                              High contention           -2.27286136  2.98220796
## 
## Components of chi-squared
## Critical value (0.05 with 7 df) is 14.07
##                              target.regime.type          Democracy   Autocracy
## Q4.17                        potential.contentiousness                        
## Not restricted at all        Low contention            4.110571774 7.076722997
##                              High contention           1.195083710 2.057445251
## Slightly restricted          Low contention            0.005519605 0.009502502
##                              High contention           1.166859588 2.008854859
## Moderately restricted        Low contention            0.758768892 1.306289627
##                              High contention           1.381520317 2.378412819
## Very or extremely restricted Low contention            0.666776907 1.147917062
##                              High contention           5.165898755 8.893564334
## 
## p for components
##                              target.regime.type        Democracy Autocracy
## Q4.17                        potential.contentiousness                    
## Not restricted at all        Low contention                0.767     0.421
##                              High contention               0.991     0.957
## Slightly restricted          Low contention                1.000     1.000
##                              High contention               0.992     0.959
## Moderately restricted        Low contention                0.998     0.988
##                              High contention               0.986     0.936
## Very or extremely restricted Low contention                0.999     0.992
##                              High contention               0.640     0.260

Regime type

Slightly more NGOs that work in autocracies have changed their programming in the past decade, but not in a statistically significant way. In general, most INGOs have not changed much, and the split is proportional across regime type.

df.change.programming.regime <- survey.countries.clean %>%
  filter(!is.na(Q4.19)) %>%
  filter(Q4.19 != "Don't know") %>%
  mutate(Q4.19 = droplevels(Q4.19),
         Q4.19 = factor(Q4.19, levels=rev(levels(Q4.19)))) %>%
  select(Q4.19, target.regime.type)

plot.change.programming.regime <- prodplot(df.change.programming.regime,
                                    ~ target.regime.type + Q4.19, mosaic("h"), 
                                    colour=NA) + 
  aes(fill=target.regime.type, colour="white") + 
  scale_fill_manual(values=ath.palette("regime"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="Change in programming, by regime type",
       subtitle="Q4.19: Over the last 10 years, has your organization changed its mix of programming in `target_country`?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.change.programming.regime

change.programming.table <- df.change.programming.regime %>%
  xtabs(~ Q4.19 + target.regime.type, .)

analyze.cat.var(change.programming.table)

## Table counts
## target.regime.type Democracy Autocracy
## Q4.19                                 
## No                       222       115
## Yes                      101        67
## 
## Expected values
##      target.regime.type
## Q4.19 Democracy Autocracy
##   No   215.5465 121.45347
##   Yes  107.4535  60.54653
## 
## Row proporitions
## target.regime.type Democracy Autocracy
## Q4.19                                 
## No                 0.6587537 0.3412463
## Yes                0.6011905 0.3988095
## 
## Column proporitions
## target.regime.type Democracy Autocracy
## Q4.19                                 
## No                 0.6873065 0.6318681
## Yes                0.3126935 0.3681319
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)
## X-squared = 1.3715, df = 1, p-value = 0.2416
## 
## Cramer's V
## [1] 0.05649081
## 
## Pearson residuals
## 2 is used as critical value by convention
## target.regime.type  Democracy  Autocracy
## Q4.19                                   
## No                  0.4395643 -0.5855824
## Yes                -0.6225622  0.8293700
## 
## Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84
## target.regime.type Democracy Autocracy
## Q4.19                                 
## No                 0.1932168 0.3429068
## Yes                0.3875837 0.6878546
## 
## p for components
## target.regime.type Democracy Autocracy
## Q4.19                                 
## No                     0.660     0.558
## Yes                    0.534     0.407

Potential contentiousness

Slightly fewer INGOs working on highly contentious issues have changed their programming over the past decade, but again, the difference between expected and actual is not significant.

df.change.programming.issue <- survey.countries.clean %>%
  filter(!is.na(Q4.19)) %>%
  filter(Q4.19 != "Don't know") %>%
  mutate(Q4.19 = droplevels(Q4.19),
         Q4.19 = factor(Q4.19, levels=rev(levels(Q4.19)))) %>%
  select(Q4.19, potential.contentiousness)

plot.change.programming.issue <- prodplot(df.change.programming.issue,
                                   ~ potential.contentiousness + Q4.19, mosaic("h"), 
                                   colour=NA) + 
  aes(fill=potential.contentiousness, colour="white") + 
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="Change in programming, by issue",
       subtitle="Q4.19: Over the last 10 years, has your organization changed its mix of programming in `target_country`?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.change.programming.issue

change.programming.issue.table <- df.change.programming.issue %>%
  xtabs(~ Q4.19 + potential.contentiousness, .)

analyze.cat.var(change.programming.issue.table)

## Table counts
## potential.contentiousness Low contention High contention
## Q4.19                                                   
## No                                   231             106
## Yes                                  123              45
## 
## Expected values
##      potential.contentiousness
## Q4.19 Low contention High contention
##   No        236.2337       100.76634
##   Yes       117.7663        50.23366
## 
## Row proporitions
## potential.contentiousness Low contention High contention
## Q4.19                                                   
## No                             0.6854599       0.3145401
## Yes                            0.7321429       0.2678571
## 
## Column proporitions
## potential.contentiousness Low contention High contention
## Q4.19                                                   
## No                             0.6525424       0.7019868
## Yes                            0.3474576       0.2980132
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)
## X-squared = 0.95356, df = 1, p-value = 0.3288
## 
## Cramer's V
## [1] 0.04804381
## 
## Pearson residuals
## 2 is used as critical value by convention
## potential.contentiousness Low contention High contention
## Q4.19                                                   
## No                            -0.3405139       0.5213724
## Yes                            0.4822755      -0.7384283
## 
## Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84
## potential.contentiousness Low contention High contention
## Q4.19                                                   
## No                             0.1159497       0.2718292
## Yes                            0.2325897       0.5452764
## 
## p for components
## potential.contentiousness Low contention High contention
## Q4.19                                                   
## No                                 0.733           0.602
## Yes                                0.630           0.460

Regime type

labels.changes <- data_frame(levels=c("funding", "issues", "comm_govt", 
                                      "comm_donors", "locations", "country_office",
                                      "local_staff", "foreign_staff"),
                             labels=c("Changed sources of funding",
                                      "Changed issues worked on",
                                      "Changed communication with the government",
                                      "Changed communication with donors",
                                      "Changed locations worked in",
                                      "Changed location country office",
                                      "Used more local staff and/or volunteers",
                                      "Used more foreign staff and/or volunteers"))

df.changes.response <- survey.countries.clean %>%
  select(dplyr::contains("Q4.21"), -dplyr::contains("TEXT"),
         target.regime.type, potential.contentiousness) %>%
  gather(question, response, -c(target.regime.type, potential.contentiousness)) %>%
  mutate(question = str_replace(question, "Q4\\.21_", ""),
         question = factor(question, levels=labels.changes$levels,
                           labels=labels.changes$labels, ordered=TRUE)) %>%
  filter(!(response %in% c("Don't know", "Not applicable")))


show.output <- function(chunk) {
  current.question <- as.character(chunk$question[1])
  cat("\n")
  cat(paste0(rep("-", nchar(current.question) + 2), collapse=""))
  cat(paste0("\n", current.question, "\n"))
  cat(paste0(rep("-", nchar(current.question) + 2), collapse=""))
  cat("\n")
  
  regime.table <- chunk %>%
    xtabs(~ response + target.regime.type, .)
  
  issue.table <- chunk %>%
    xtabs(~ response + potential.contentiousness, .)
  
  # Use diff for the two column effect!
  # http://stackoverflow.com/a/9214177/120898
  tmp.regime <- tempfile()
  tmp.issue <- tempfile()
  
  cat(capture.output( analyze.cat.var(regime.table) ), sep="\n", file=tmp.regime)
  cat(capture.output( analyze.cat.var(issue.table) ), sep="\n", file=tmp.issue)
 
  # system does weeeeird stuff with knitr (see https://github.com/yihui/knitr/issues/1203),
  # so system(..., intern=TRUE) + cat(paste(..., collapse="\n)) does the trick
  system.output <- system(sprintf("diff -y -W 140 %s %s", tmp.regime, tmp.issue), intern=TRUE)
  
  cat(paste(system.output, collapse="\n"))
}

show.plots <- function(chunk) {
  current.question <- as.character(chunk$question[1])
  
  plot.regime <- prodplot(chunk,
                          ~ target.regime.type + response, mosaic("h"),
                          colour=NA) + 
    aes(fill=target.regime.type, colour="white") + 
    scale_fill_manual(values=ath.palette("regime"), name=NULL) +
    guides(fill=FALSE) +
    labs(title=current.question) +
    theme_ath() + theme(axis.title=element_blank(),
                        panel.grid=element_blank())
  
  plot.issue <- prodplot(chunk,
                         ~ potential.contentiousness + response, mosaic("h"),
                         colour=NA) + 
    aes(fill=potential.contentiousness, colour="white") + 
    scale_fill_manual(values=ath.palette("contention"), name=NULL) +
    guides(fill=FALSE) +
    labs(title=" ") +
    theme_ath() + theme(axis.title=element_blank(),
                        panel.grid=element_blank())

  # Can't do this because knitr chokes...
  # plot.both <- arrangeGrob(plot.regime, plot.issue, nrow=1)
  # grid::grid.draw(plot.both)
  
  grid.arrange(plot.regime, plot.issue, nrow=1)
}

# show.output(chunk)
# 


# This could be done with purrr, but I can't get it to work right (the output
# and plots go in two huge chunks instead of alternating between text and
# plots; if I put show.plots() in show.output(), only one plot shows up). So
# instead of using fancy R vectorization, yay loops. \(•◡•)/
#
# Something like this with purrr almost works:
# suppressWarnings(response.to.regulations <- df.changes.response %>%
#   split(.$question) %>%
#   map(~ show.output(.))) %>%
#   map(~ show.plots(.))

for (Q in unique(df.changes.response$question)) {
  chunk <- filter(df.changes.response, question == Q)
  
  show.plots(chunk)
  suppressWarnings(show.output(chunk))
  cat("\n")
}

## 
## ----------------------------
## Changed sources of funding
## ----------------------------
## Table counts                             Table counts
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                       177        95                    |  No                                   191              81
## Yes                       32        36                    |  Yes                                   44              24
## 
## Expected values                              Expected values
##         target.regime.type                        |          potential.contentiousness
## response Democracy Autocracy                      |  response Low contention High contention
##      No      167.2     104.8                      |       No             188              84
##      Yes      41.8      26.2                      |       Yes             47              21
## 
## Row proporitions                         Row proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.6507353 0.3492647                    |  No                             0.7022059       0.2977941
## Yes                0.4705882 0.5294118                    |  Yes                            0.6470588       0.3529412
## 
## Column proporitions                          Column proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.8468900 0.7251908                    |  No                             0.8127660       0.7714286
## Yes                0.1531100 0.2748092                    |  Yes                            0.1872340       0.2285714
## 
## Chi-squared test for table                       Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction        Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)                     data:  ftable(cat.table)
## X-squared = 6.7129, df = 1, p-value = 0.009572                |  X-squared = 0.53825, df = 1, p-value = 0.4632
## 
## Cramer's V                               Cramer's V
## [1] 0.1480667                                 |  [1] 0.04774553
## 
## Pearson residuals                            Pearson residuals
## 2 is used as critical value by convention                2 is used as critical value by convention
## target.regime.type  Democracy  Autocracy              |  potential.contentiousness Low contention High contention
## response                                              |  response                                                
## No                  0.7578931 -0.9572942              |  No                             0.2187975      -0.3273268
## Yes                -1.5157862  1.9145884              |  Yes                           -0.4375950       0.6546537
## 
## Components of chi-squared                        Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84                  Critical value (0.05 with 1 df) is 3.84
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.5744019 0.9164122                    |  No                            0.04787234      0.10714286
## Yes                2.2976077 3.6656489                    |  Yes                           0.19148936      0.42857143
## 
## p for components                         p for components
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                     0.449     0.338                    |  No                                 0.827           0.743
## Yes                    0.130     0.056                    |  Yes                                0.662           0.513

## 
## --------------------------
## Changed issues worked on
## --------------------------
## Table counts                             Table counts
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                       181       104                    |  No                                   200              85
## Yes                       50        39                    |  Yes                                   58              31
## 
## Expected values                              Expected values
##         target.regime.type                        |          potential.contentiousness
## response Democracy Autocracy                      |  response Low contention High contention
##      No  176.02941 108.97059                      |       No       196.60428        88.39572
##      Yes  54.97059  34.02941                      |       Yes       61.39572        27.60428
## 
## Row proporitions                         Row proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.6350877 0.3649123                    |  No                             0.7017544       0.2982456
## Yes                0.5617978 0.4382022                    |  Yes                            0.6516854       0.3483146
## 
## Column proporitions                          Column proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.7835498 0.7272727                    |  No                             0.7751938       0.7327586
## Yes                0.2164502 0.2727273                    |  Yes                            0.2248062       0.2672414
## 
## Chi-squared test for table                       Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction        Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)                     data:  ftable(cat.table)
## X-squared = 1.2478, df = 1, p-value = 0.264               |  X-squared = 0.57785, df = 1, p-value = 0.4472
## 
## Cramer's V                               Cramer's V
## [1] 0.06422261                                |  [1] 0.04609432
## 
## Pearson residuals                            Pearson residuals
## 2 is used as critical value by convention                2 is used as critical value by convention
## target.regime.type  Democracy  Autocracy              |  potential.contentiousness Low contention High contention
## response                                              |  response                                                
## No                  0.3746409 -0.4761603              |  No                             0.2421785      -0.3611740
## Yes                -0.6704133  0.8520804              |  Yes                           -0.4333742       0.6463145
## 
## Components of chi-squared                        Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84                  Critical value (0.05 with 1 df) is 3.84
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.1403558 0.2267286                    |  No                            0.05865044      0.13044667
## Yes                0.4494539 0.7260410                    |  Yes                           0.18781321      0.41772248
## 
## p for components                         p for components
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                     0.708     0.634                    |  No                                 0.809           0.718
## Yes                    0.503     0.394                    |  Yes                                0.665           0.518

## 
## -------------------------------------------
## Changed communication with the government
## -------------------------------------------
## Table counts                             Table counts
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                       167        84                    |  No                                   174              77
## Yes                       50        44                    |  Yes                                   65              29
## 
## Expected values                              Expected values
##         target.regime.type                        |          potential.contentiousness
## response Democracy Autocracy                      |  response Low contention High contention
##      No  157.87536  93.12464                      |       No       173.88116        77.11884
##      Yes  59.12464  34.87536                      |       Yes       65.11884        28.88116
## 
## Row proporitions                         Row proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.6653386 0.3346614                    |  No                             0.6932271       0.3067729
## Yes                0.5319149 0.4680851                    |  Yes                            0.6914894       0.3085106
## 
## Column proporitions                          Column proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.7695853 0.6562500                    |  No                             0.7280335       0.7264151
## Yes                0.2304147 0.3437500                    |  Yes                            0.2719665       0.2735849
## 
## Chi-squared test for table                       Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction        Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)                     data:  ftable(cat.table)
## X-squared = 4.6609, df = 1, p-value = 0.03086                 |  X-squared = 1.7481e-30, df = 1, p-value = 1
## 
## Cramer's V                               Cramer's V
## [1] 0.12297                               |  [1] 0.001676993
## 
## Pearson residuals                            Pearson residuals
## 2 is used as critical value by convention                2 is used as critical value by convention
## target.regime.type  Democracy  Autocracy              |  potential.contentiousness Low contention High contention
## response                                              |  response                                                
## No                  0.7262037 -0.9455475              |  No                           0.009012359    -0.013532702
## Yes                -1.1866739  1.5450988              |  Yes                         -0.014726903     0.022113498
## 
## Components of chi-squared                        Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84                  Critical value (0.05 with 1 df) is 3.84
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.5273718 0.8940600                    |  No                          8.122262e-05    1.831340e-04
## Yes                1.4081949 2.3873304                    |  Yes                         2.168817e-04    4.890068e-04
## 
## p for components                         p for components
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                     0.468     0.344                    |  No                                 0.993           0.989
## Yes                    0.235     0.122                    |  Yes                                0.988           0.982

## 
## -----------------------------------
## Changed communication with donors
## -----------------------------------
## Table counts                             Table counts
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                       165        99                    |  No                                   184              80
## Yes                       38        39                    |  Yes                                   51              26
## 
## Expected values                              Expected values
##         target.regime.type                        |          potential.contentiousness
## response Democracy Autocracy                      |  response Low contention High contention
##      No  157.16129 106.83871                      |       No       181.93548        82.06452
##      Yes  45.83871  31.16129                      |       Yes       53.06452        23.93548
## 
## Row proporitions                         Row proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.6250000 0.3750000                    |  No                             0.6969697       0.3030303
## Yes                0.4935065 0.5064935                    |  Yes                            0.6623377       0.3376623
## 
## Column proporitions                          Column proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.8128079 0.7173913                    |  No                             0.7829787       0.7547170
## Yes                0.1871921 0.2826087                    |  Yes                            0.2170213       0.2452830
## 
## Chi-squared test for table                       Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction        Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)                     data:  ftable(cat.table)
## X-squared = 3.75, df = 1, p-value = 0.05281               |  X-squared = 0.19167, df = 1, p-value = 0.6615
## 
## Cramer's V                               Cramer's V
## [1] 0.1120119                                 |  [1] 0.03128517
## 
## Pearson residuals                            Pearson residuals
## 2 is used as critical value by convention                2 is used as critical value by convention
## target.regime.type  Democracy  Autocracy              |  potential.contentiousness Low contention High contention
## response                                              |  response                                                
## No                  0.6252760 -0.7583685              |  No                             0.1530592      -0.2278980
## Yes                -1.1577863  1.4042255              |  Yes                           -0.2834107       0.4219852
## 
## Components of chi-squared                        Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84                  Critical value (0.05 with 1 df) is 3.84
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.3909701 0.5751227                    |  No                            0.02342713      0.05193751
## Yes                1.3404690 1.9718493                    |  Yes                           0.08032160      0.17807147
## 
## p for components                         p for components
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                     0.532     0.448                    |  No                                 0.878           0.820
## Yes                    0.247     0.160                    |  Yes                                0.777           0.673

## 
## -----------------------------
## Changed locations worked in
## -----------------------------
## Table counts                             Table counts
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                       178        87                    |  No                                   185              80
## Yes                       42        44                    |  Yes                                   58              28
## 
## Expected values                              Expected values
##         target.regime.type                        |          potential.contentiousness
## response Democracy Autocracy                      |  response Low contention High contention
##      No  166.09687  98.90313                      |       No       183.46154        81.53846
##      Yes  53.90313  32.09687                      |       Yes       59.53846        26.46154
## 
## Row proporitions                         Row proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.6716981 0.3283019                    |  No                             0.6981132       0.3018868
## Yes                0.4883721 0.5116279                    |  Yes                            0.6744186       0.3255814
## 
## Column proporitions                          Column proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.8090909 0.6641221                    |  No                             0.7613169       0.7407407
## Yes                0.1909091 0.3358779                    |  Yes                            0.2386831       0.2592593
## 
## Chi-squared test for table                       Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction        Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)                     data:  ftable(cat.table)
## X-squared = 8.5611, df = 1, p-value = 0.003434                |  X-squared = 0.07797, df = 1, p-value = 0.7801
## 
## Cramer's V                               Cramer's V
## [1] 0.1630232                                 |  [1] 0.02208038
## 
## Pearson residuals                            Pearson residuals
## 2 is used as critical value by convention                2 is used as critical value by convention
## target.regime.type  Democracy  Autocracy              |  potential.contentiousness Low contention High contention
## response                                              |  response                                                
## No                  0.9235929 -1.1968957              |  No                             0.1135832      -0.1703748
## Yes                -1.6212661  2.1010191              |  Yes                           -0.1993829       0.2990743
## 
## Components of chi-squared                        Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84                  Critical value (0.05 with 1 df) is 3.84
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.8530239 1.4325592                    |  No                            0.01290114      0.02902758
## Yes                2.6285039 4.4142813                    |  Yes                           0.03975353      0.08944544
## 
## p for components                         p for components
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                     0.356     0.231                    |  No                                 0.910           0.865
## Yes                    0.105     0.036                    |  Yes                                0.842           0.765

## 
## ---------------------------------
## Changed location country office
## ---------------------------------
## Table counts                             Table counts
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                       183       110                    |  No                                   206              87
## Yes                       21        17                    |  Yes                                   23              15
## 
## Expected values                              Expected values
##         target.regime.type                        |          potential.contentiousness
## response Democracy Autocracy                      |  response Low contention High contention
##      No  180.58006 112.41994                      |       No       202.70997        90.29003
##      Yes  23.41994  14.58006                      |       Yes       26.29003        11.70997
## 
## Row proporitions                         Row proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.6245734 0.3754266                    |  No                             0.7030717       0.2969283
## Yes                0.5526316 0.4473684                    |  Yes                            0.6052632       0.3947368
## 
## Column proporitions                          Column proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.8970588 0.8661417                    |  No                             0.8995633       0.8529412
## Yes                0.1029412 0.1338583                    |  Yes                            0.1004367       0.1470588
## 
## Chi-squared test for table                       Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction        Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)                     data:  ftable(cat.table)
## X-squared = 0.46342, df = 1, p-value = 0.496              |  X-squared = 1.0855, df = 1, p-value = 0.2975
## 
## Cramer's V                               Cramer's V
## [1] 0.04716176                                |  [1] 0.06752809
## 
## Pearson residuals                            Pearson residuals
## 2 is used as critical value by convention                2 is used as critical value by convention
## target.regime.type  Democracy  Autocracy              |  potential.contentiousness Low contention High contention
## response                                              |  response                                                
## No                  0.1800817 -0.2282353              |  No                             0.2310800      -0.3462422
## Yes                -0.5000479  0.6337601              |  Yes                           -0.6416591       0.9614396
## 
## Components of chi-squared                        Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84                  Critical value (0.05 with 1 df) is 3.84
## target.regime.type  Democracy  Autocracy              |  potential.contentiousness Low contention High contention
## response                                              |  response                                                
## No                 0.03242943 0.05209136              |  No                            0.05339796      0.11988365
## Yes                0.25004794 0.40165180              |  Yes                           0.41172637      0.92436607
## 
## p for components                         p for components
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                     0.857     0.819                    |  No                                 0.817           0.729
## Yes                    0.617     0.526                    |  Yes                                0.521           0.336

## 
## -----------------------------------------
## Used more local staff and/or volunteers
## -----------------------------------------
## Table counts                             Table counts
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                       142        67                    |  No                                   146              63
## Yes                       70        65                    |  Yes                                   94              41
## 
## Expected values                              Expected values
##         target.regime.type                        |          potential.contentiousness
## response Democracy Autocracy                      |  response Low contention High contention
##      No  128.80233  80.19767                      |       No       145.81395        63.18605
##      Yes  83.19767  51.80233                      |       Yes       94.18605        40.81395
## 
## Row proporitions                         Row proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.6794258 0.3205742                    |  No                             0.6985646       0.3014354
## Yes                0.5185185 0.4814815                    |  Yes                            0.6962963       0.3037037
## 
## Column proporitions                          Column proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.6698113 0.5075758                    |  No                             0.6083333       0.6057692
## Yes                0.3301887 0.4924242                    |  Yes                            0.3916667       0.3942308
## 
## Chi-squared test for table                       Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction        Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)                     data:  ftable(cat.table)
## X-squared = 8.3125, df = 1, p-value = 0.003937                |  X-squared = 1.0288e-29, df = 1, p-value = 1
## 
## Cramer's V                               Cramer's V
## [1] 0.1615701                                 |  [1] 0.002411669
## 
## Pearson residuals                            Pearson residuals
## 2 is used as critical value by convention                2 is used as critical value by convention
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                  1.162882 -1.473725                    |  No                            0.01540714     -0.02340512
## Yes                -1.446911  1.833677                    |  Yes                          -0.01917027      0.02912173
## 
## Components of chi-squared                        Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84                  Critical value (0.05 with 1 df) is 3.84
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                  1.352294  2.171866                    |  No                          0.0002373799    0.0005477998
## Yes                 2.093551  3.362370                    |  Yes                         0.0003674993    0.0008480753
## 
## p for components                         p for components
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                     0.245     0.141                    |  No                                 0.988           0.981
## Yes                    0.148     0.067                    |  Yes                                0.985           0.977

## 
## -------------------------------------------
## Used more foreign staff and/or volunteers
## -------------------------------------------
## Table counts                             Table counts
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                       171       107                    |  No                                   193              85
## Yes                       34        19                    |  Yes                                   37              16
## 
## Expected values                              Expected values
##         target.regime.type                        |          potential.contentiousness
## response Democracy Autocracy                      |  response Low contention High contention
##      No  172.17523 105.82477                      |       No       193.17221        84.82779
##      Yes  32.82477  20.17523                      |       Yes       36.82779        16.17221
## 
## Row proporitions                         Row proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.6151079 0.3848921                    |  No                             0.6942446       0.3057554
## Yes                0.6415094 0.3584906                    |  Yes                            0.6981132       0.3018868
## 
## Column proporitions                          Column proporitions
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.8341463 0.8492063                    |  No                             0.8391304       0.8415842
## Yes                0.1658537 0.1507937                    |  Yes                            0.1608696       0.1584158
## 
## Chi-squared test for table                       Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction        Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)                     data:  ftable(cat.table)
## X-squared = 0.043445, df = 1, p-value = 0.8349                |  X-squared = 1.637e-30, df = 1, p-value = 1
## 
## Cramer's V                               Cramer's V
## [1] 0.01994009                                |  [1] 0.003080987
## 
## Pearson residuals                            Pearson residuals
## 2 is used as critical value by convention                2 is used as critical value by convention
## target.regime.type   Democracy   Autocracy                |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 -0.08956458  0.11424256                |  No                           -0.01239009      0.01869725
## Yes                 0.20512601 -0.26164497                |  Yes                           0.02837651     -0.04282154
## 
## Components of chi-squared                        Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84                  Critical value (0.05 with 1 df) is 3.84
## target.regime.type   Democracy   Autocracy                |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                 0.008021813 0.013051363                |  No                          0.0001535144    0.0003495872
## Yes                0.042076681 0.068458093                |  Yes                         0.0008052264    0.0018336839
## 
## p for components                         p for components
## target.regime.type Democracy Autocracy                    |  potential.contentiousness Low contention High contention
## response                                                  |  response                                                
## No                     0.929     0.909                    |  No                                 0.990           0.985
## Yes                    0.837     0.794                    |  Yes                                0.977           0.966

Regime type

INGOs working in autocracies are far more likely to have discussed NGO regulations with their host governments—the majority actually have (54% of those responding “yes” work in autocracies; 32% of those working in autocracies responded “yes”), and the difference is statistically significant. This is likely because INGOs working in autocracies have more of a reason to discuss the regulations, as they see more regulations in general.

df.discuss.regime <- survey.countries.clean %>%
  filter(!is.na(Q4.22)) %>%
  filter(Q4.22 != "Don't know") %>%
  mutate(Q4.22 = droplevels(Q4.22),
         Q4.22 = factor(Q4.22, levels=rev(levels(Q4.22)))) %>%
  select(Q4.22, target.regime.type)

plot.discuss.regime <- prodplot(df.discuss.regime,
                                    ~ target.regime.type + Q4.22, mosaic("h"), 
                                    colour=NA) + 
  aes(fill=target.regime.type, colour="white") + 
  scale_fill_manual(values=ath.palette("regime"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="Discussions about regulations, by regime type",
       subtitle="Q4.22: Has your organization discussed NGO regulations with government officials in `target_country`?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.discuss.regime

discuss.table <- df.discuss.regime %>%
  xtabs(~ Q4.22 + target.regime.type, .)

analyze.cat.var(discuss.table)

## Table counts
## target.regime.type Democracy Autocracy
## Q4.22                                 
## No                       257       107
## Yes                       50        61
## 
## Expected values
##      target.regime.type
## Q4.22 Democracy Autocracy
##   No  235.25895 128.74105
##   Yes  71.74105  39.25895
## 
## Row proporitions
## target.regime.type Democracy Autocracy
## Q4.22                                 
## No                 0.7060440 0.2939560
## Yes                0.4504505 0.5495495
## 
## Column proporitions
## target.regime.type Democracy Autocracy
## Q4.22                                 
## No                 0.8371336 0.6369048
## Yes                0.1628664 0.3630952
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)
## X-squared = 23.204, df = 1, p-value = 1.457e-06
## 
## Cramer's V
## [1] 0.2262237
## 
## Pearson residuals
## 2 is used as critical value by convention
## target.regime.type Democracy Autocracy
## Q4.22                                 
## No                  1.417449 -1.916117
## Yes                -2.566828  3.469854
## 
## Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84
## target.regime.type Democracy Autocracy
## Q4.22                                 
## No                  2.009162  3.671505
## Yes                 6.588604 12.039889
## 
## p for components
## target.regime.type Democracy Autocracy
## Q4.22                                 
## No                     0.156     0.055
## Yes                    0.010     0.001

Potential contentiousness

There’s no difference across issue areas in whether or not an INGO discusses regulations with the government—the same proportion (66% haven’t discussed; 20% have; 10% don’t know) holds across both levels of potential contentiousness.

df.discuss.issue <- survey.countries.clean %>%
  filter(!is.na(Q4.22)) %>%
  filter(Q4.22 != "Don't know") %>%
  mutate(Q4.22 = droplevels(Q4.22),
         Q4.22 = factor(Q4.22, levels=rev(levels(Q4.22)))) %>%
  select(Q4.22, potential.contentiousness)

plot.discuss.issue <- prodplot(df.discuss.issue,
                               ~ potential.contentiousness + Q4.22, mosaic("h"), 
                               colour=NA) + 
  aes(fill=potential.contentiousness, colour="white") + 
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="Discussions about regulations, by issue",
       subtitle="Q4.22: Has your organization discussed NGO regulations with government officials in `target_country`?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.discuss.issue

discuss.issue.table <- df.discuss.issue %>%
  xtabs(~ Q4.22 + potential.contentiousness, .)

analyze.cat.var(discuss.issue.table)

## Table counts
## potential.contentiousness Low contention High contention
## Q4.22                                                   
## No                                   255             109
## Yes                                   76              35
## 
## Expected values
##      potential.contentiousness
## Q4.22 Low contention High contention
##   No       253.65053       110.34947
##   Yes       77.34947        33.65053
## 
## Row proporitions
## potential.contentiousness Low contention High contention
## Q4.22                                                   
## No                             0.7005495       0.2994505
## Yes                            0.6846847       0.3153153
## 
## Column proporitions
## potential.contentiousness Low contention High contention
## Q4.22                                                   
## No                             0.7703927       0.7569444
## Yes                            0.2296073       0.2430556
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)
## X-squared = 0.040157, df = 1, p-value = 0.8412
## 
## Cramer's V
## [1] 0.01460665
## 
## Pearson residuals
## 2 is used as critical value by convention
## potential.contentiousness Low contention High contention
## Q4.22                                                   
## No                            0.08473182     -0.12846336
## Yes                          -0.15343899      0.23263148
## 
## Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84
## potential.contentiousness Low contention High contention
## Q4.22                                                   
## No                           0.007179481     0.016502836
## Yes                          0.023543524     0.054117407
## 
## p for components
## potential.contentiousness Low contention High contention
## Q4.22                                                   
## No                                 0.932           0.898
## Yes                                0.878           0.816

Regime type

Most INGOs don’t attempt to change regulations, but those that do are most likely to work in autocracies (and the difference is significant). Additionally, only 8% of INGOs working in democracies have attempted to change regulations, while 16% of those working in autocracies have, despite the fact that the legislative process is much more difficult to navigate in autocracies. INGOs appear to be more likely to engage in the political process to adjust regulations despite these challenges.

df.change.regs.regime <- survey.countries.clean %>%
  filter(!is.na(Q4.23)) %>%
  filter(Q4.23 != "Don't know") %>%
  mutate(Q4.23 = droplevels(Q4.23),
         Q4.23 = factor(Q4.23, levels=rev(levels(Q4.23)))) %>%
  select(Q4.23, target.regime.type)

plot.change.regs.regime <- prodplot(df.change.regs.regime,
                                ~ target.regime.type + Q4.23, mosaic("h"), 
                                colour=NA) + 
  aes(fill=target.regime.type, colour="white") + 
  scale_fill_manual(values=ath.palette("regime"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="Attempts to change regulations, by regime type",
       subtitle="Q4.23: Has your organization tried to change NGO regulations in `target_country`?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.change.regs.regime

change.regs.table <- df.change.regs.regime %>%
  xtabs(~ Q4.23 + target.regime.type, .)

analyze.cat.var(change.regs.table)

## Table counts
## target.regime.type Democracy Autocracy
## Q4.23                                 
## No                       295       143
## Yes                       22        28
## 
## Expected values
##      target.regime.type
## Q4.23 Democracy Autocracy
##   No  284.52049 153.47951
##   Yes  32.47951  17.52049
## 
## Row proporitions
## target.regime.type Democracy Autocracy
## Q4.23                                 
## No                  0.673516  0.326484
## Yes                 0.440000  0.560000
## 
## Column proporitions
## target.regime.type  Democracy  Autocracy
## Q4.23                                   
## No                 0.93059937 0.83625731
## Yes                0.06940063 0.16374269
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)
## X-squared = 9.7494, df = 1, p-value = 0.001794
## 
## Cramer's V
## [1] 0.1484263
## 
## Pearson residuals
## 2 is used as critical value by convention
## target.regime.type  Democracy  Autocracy
## Q4.23                                   
## No                  0.6212754 -0.8458935
## Yes                -1.8388071  2.5036162
## 
## Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84
## target.regime.type Democracy Autocracy
## Q4.23                                 
## No                 0.3859831 0.7155359
## Yes                3.3812117 6.2680941
## 
## p for components
## target.regime.type Democracy Autocracy
## Q4.23                                 
## No                     0.534     0.398
## Yes                    0.066     0.012

Potential contentiousness

There’s also a difference across issue area. INGOs working on highly contentious issues are more likely to try to change regulations, similar to regime type. The difference is statistically significant overall, but is not driven by a single significant cell (though yes + high comes close).

df.change.regs.issue <- survey.countries.clean %>%
  filter(!is.na(Q4.23)) %>%
  filter(Q4.23 != "Don't know") %>%
  mutate(Q4.23 = droplevels(Q4.23),
         Q4.23 = factor(Q4.23, levels=rev(levels(Q4.23)))) %>%
  select(Q4.23, potential.contentiousness)

plot.change.regs.issue <- prodplot(df.change.regs.issue,
                               ~ potential.contentiousness + Q4.23, mosaic("h"), 
                               colour=NA) + 
  aes(fill=potential.contentiousness, colour="white") + 
  scale_fill_manual(values=ath.palette("contention"), name=NULL) +
  guides(fill=FALSE) +
  labs(title="Attempts to change regulations, by issue",
       subtitle="Q4.23: Has your organization tried to change NGO regulations in `target_country`?") +
  theme_ath() + theme(axis.title=element_blank(),
                      panel.grid=element_blank())

plot.change.regs.issue

change.regs.issue.table <- df.change.regs.issue %>%
  xtabs(~ Q4.23 + potential.contentiousness, .)

analyze.cat.var(change.regs.issue.table)

## Table counts
## potential.contentiousness Low contention High contention
## Q4.23                                                   
## No                                   310             128
## Yes                                   30              20
## 
## Expected values
##      potential.contentiousness
## Q4.23 Low contention High contention
##   No       305.16393       132.83607
##   Yes       34.83607        15.16393
## 
## Row proporitions
## potential.contentiousness Low contention High contention
## Q4.23                                                   
## No                             0.7077626       0.2922374
## Yes                            0.6000000       0.4000000
## 
## Column proporitions
## potential.contentiousness Low contention High contention
## Q4.23                                                   
## No                            0.91176471      0.86486486
## Yes                           0.08823529      0.13513514
## 
## Chi-squared test for table
## 
##  Pearson's Chi-squared test with Yates' continuity correction
## 
## data:  ftable(cat.table)
## X-squared = 1.9827, df = 1, p-value = 0.1591
## 
## Cramer's V
## [1] 0.07109182
## 
## Pearson residuals
## 2 is used as critical value by convention
## potential.contentiousness Low contention High contention
## Q4.23                                                   
## No                             0.2768379      -0.4195987
## Yes                           -0.8193654       1.2418989
## 
## Components of chi-squared
## Critical value (0.05 with 1 df) is 3.84
## potential.contentiousness Low contention High contention
## Q4.23                                                   
## No                            0.07663923      0.17606311
## Yes                           0.67135969      1.54231280
## 
## p for components
## potential.contentiousness Low contention High contention
## Q4.23                                                   
## No                                 0.782           0.675
## Yes                                0.413           0.214