Wave 7 of the World Values Survey was conducted in 2017-2021. Among
many measures, the survey included two measures of interest to this
project:
1. Confidence in institutions. The asked about trust in
a wide variety of institutions. Here, we focus on trust in the three
main branches of governments: The executive branch (“The government”),
the judicial branch (“The courts”), and the legislative branch
(“Parliament”). We take the mean score of the three confidence ratings
(alpha = 0.82)
2. Support for radical change. Participants were asked
to select the statement that best describes their opinion: (1) The
entire way our society is organized must be radically changed by
revolutionary action; (2) Our society must be gradually
improved by reforms; or (3) Our present society must be
valiantly defended against all subversive forces. We dummy-coded
this variable so that each response gets a 0 or 1, depending on
response. We will look at the likelihood of selecting option 1 as the
main DV, but we will also verify this with separate models predicting
the other two options.
3. Control variables. We have individual-level controls
and country-level controls. On the individual level, we control for
participants’ ideology, urban/rural, sex, age, education, and income (on
a within-country income scale). On the country-level, we control for GDP
per capita.
missing values for our two variables of interest: Confidence in institutions and support for radical change.
country | N_eligible | N_missing |
---|---|---|
AND | 959 | 45 |
ARG | 995 | 8 |
ARM | 1175 | 48 |
AUL | 1755 | 58 |
BNG | 1178 | 22 |
BOL | 2004 | 63 |
BRA | 1631 | 131 |
CAN | 4018 | 0 |
CHL | 950 | 50 |
CHN | 3003 | 33 |
COL | 1520 | 0 |
CYP | 925 | 75 |
CZR | 1164 | 36 |
DRV | 1188 | 12 |
ECU | 1177 | 23 |
EGY | 959 | 241 |
ETH | 1198 | 32 |
GMY | 1472 | 56 |
GRC | 1146 | 54 |
GUA | 1229 | 0 |
HKG | 2036 | 39 |
INS | 3170 | 30 |
IRN | 1476 | 23 |
IRQ | 1192 | 8 |
JOR | 1180 | 23 |
JPN | 1066 | 287 |
KEN | 1242 | 24 |
KYR | 1154 | 46 |
KZK | 1091 | 185 |
LEB | 1191 | 9 |
LIB | 1154 | 42 |
MAD | 1030 | 9 |
MAL | 1312 | 1 |
MAU | 1012 | 11 |
MEX | 1708 | 33 |
MNG | 1604 | 34 |
MOR | 1200 | 0 |
MYA | 1200 | 0 |
NEW | 977 | 80 |
NIC | 1141 | 59 |
NIG | 1214 | 23 |
NIRL | 433 | 14 |
NTH | 1650 | 495 |
PAK | 1945 | 50 |
PER | 1366 | 34 |
PHI | 1196 | 4 |
PRI | 1068 | 59 |
ROK | 1245 | 0 |
ROM | 1165 | 92 |
RUS | 1654 | 156 |
SIN | 1801 | 211 |
SLO | 1158 | 42 |
SRB | 946 | 100 |
TAJ | 1188 | 12 |
TAW | 1212 | 11 |
THI | 1343 | 157 |
TUN | 1183 | 25 |
TUR | 2336 | 79 |
UKG | 2490 | 119 |
UKR | 1177 | 112 |
URU | 969 | 31 |
USA | 2552 | 44 |
VEN | 1190 | 0 |
ZIM | 1207 | 8 |
Great, that leaves us with a total of 90470 eligible participants over 64 countries.
I’ll look only at eligible participants here (I’ll check later if the missing data is missing at random or if it impacts our analyses)
sex | N | perc |
---|---|---|
female | 47413 | 52.41 |
male | 42976 | 47.50 |
no answer | 81 | 0.09 |
age_mean | age_sd |
---|---|
43.28238 | 16.52272 |
Education descriptions vary by country, so to standardize, they went
with the ISCED (https://datatopics.worldbank.org/education/wRsc/classification)
definitions from the world bank:
0. Early childhood education; 1. Primary education; 2. Lower secondary
education; 3. Upper secondary education; 4. Post-secondary non-tertiary
education; 5. Short-cycle tertiary education; 6. Bachelor or equivalent;
7. Master or equivalent; 8. Doctoral or equivalent.
Respondents’ income was just measured as a 1-10 scale (each point is an “income group in your country”).
region | N | perc |
---|---|---|
rural | 28396 | 31.39 |
urban | 62047 | 68.58 |
no data | 27 | 0.03 |
Respondents were asked to indicate their confidence in several
organizations on a 1 (A great deal to 4 (None at all) scale. I later
reverse-scored for ease of interpretability. The three organizations we
are looking at represent the three branches of government: Justice
System/Courts (the judicial branch), the Government (the executive
branch), and Parliament (the legislative branch).
Respondents were asked to choose which of the following best
describes their opinion:
1. The entire way our society is organized must be radically changed by
revolutionary action.
2. Our society must be gradually improved by reforms.
3. Our present society must be valiantly defended against all subversive
forces.
To analyze this, I will follow the script used by Wiwad et al. (2020;
https://www.nature.com/articles/s41562-020-0835-8; https://osf.io/s8f7r/).
My primary outcome variable is the dummy-coded support for radical
change (0 and 1). My primary predictor is the composite score of
confidence in institutions.
First step in running an MLM is determining whether or not it’s actually necessary. Here is the null model:
Term | \(\hat{\beta}\) | 95% CI | \(t\) | \(\mathit{df}\) | \(p\) |
---|---|---|---|---|---|
Intercept | 0.19 | [0.17, 0.21] | 17.80 | 91,086 | < .001 |
Calculating the ICC based on the output of the null model:
ICC = 0.046
While the effect of the clustering is small, the data set is large so
this is enough to bias the model output. So, moving forward with the
MLM. First the predictor only model.
Term | \(\hat{\beta}\) | 95% CI | \(t\) | \(\mathit{df}\) | \(p\) |
---|---|---|---|---|---|
Intercept | 0.27 | [0.24, 0.29] | 23.86 | 90,405 | < .001 |
Conf inst | -0.03 | [-0.04, -0.03] | -18.60 | 90,405 | < .001 |
And now the full model with all Level 1 covariates.
Term | \(\hat{\beta}\) | 95% CI | \(t\) | \(\mathit{df}\) | \(p\) |
---|---|---|---|---|---|
Intercept | 0.40 | [0.37, 0.43] | 28.81 | 64,162 | < .001 |
Conf inst | -0.03 | [-0.04, -0.03] | -15.41 | 64,162 | < .001 |
Ideo | -0.01 | [-0.01, 0.00] | -8.35 | 64,162 | < .001 |
Sexmale | 0.02 | [0.01, 0.02] | 5.14 | 64,162 | < .001 |
Age | 0.00 | [0.00, 0.00] | -19.95 | 64,162 | < .001 |
Edu | -0.01 | [-0.01, 0.00] | -6.53 | 64,162 | < .001 |
Income | 0.00 | [0.00, 0.00] | -1.68 | 64,162 | .092 |
Urban | 0.00 | [-0.01, 0.00] | -0.68 | 64,162 | .496 |
Note that this cut our sample by ~30,000 people due to all of the
missing values. I’ll run robustness checks with imputation of missing
values to see if these and the following effects hold.
And now the full model with all Level 1 and level 2 covariates.
Term | \(\hat{\beta}\) | 95% CI | \(t\) | \(\mathit{df}\) | \(p\) |
---|---|---|---|---|---|
Intercept | 0.45 | [0.41, 0.48] | 25.57 | 62,651 | < .001 |
Conf inst | -0.03 | [-0.04, -0.03] | -14.97 | 62,651 | < .001 |
Ideo | -0.01 | [-0.01, 0.00] | -8.21 | 62,651 | < .001 |
Sexmale | 0.02 | [0.01, 0.02] | 5.00 | 62,651 | < .001 |
Age | 0.00 | [0.00, 0.00] | -19.67 | 62,651 | < .001 |
Edu | -0.01 | [-0.01, 0.00] | -6.13 | 62,651 | < .001 |
Income | 0.00 | [0.00, 0.00] | -1.70 | 62,651 | .089 |
Urban | 0.00 | [-0.01, 0.00] | -0.69 | 62,651 | .492 |
Gdppc | 0.00 | [0.00, 0.00] | -3.64 | 49 | .001 |
Predictor | \(b\) | 95% CI | \(t\) | \(\mathit{df}\) | \(p\) |
---|---|---|---|---|---|
Intercept | 0.66 | [0.57, 0.74] | 15.34 | 2462 | < .001 |
Conf inst | -0.10 | [-0.12, -0.07] | -7.93 | 2462 | < .001 |
Ideo | 0.00 | [-0.01, 0.01] | 0.16 | 2462 | .871 |
Sexmale | 0.00 | [-0.03, 0.03] | 0.19 | 2462 | .847 |
Age | 0.00 | [0.00, 0.00] | -7.85 | 2462 | < .001 |
Edu | -0.02 | [-0.03, -0.01] | -4.44 | 2462 | < .001 |
Income | -0.01 | [-0.02, 0.00] | -2.79 | 2462 | .005 |
Urban | 0.00 | [-0.04, 0.04] | -0.10 | 2462 | .924 |
The red line represents the effect of confidence in institutions on support for radical change, adjusting for all the covariates (level 1 and level 2) in the model.