library(readxl)
library(dplyr)
library(ggplot2)
library(sandwich)
library(lmtest)
library(knitr)

Data

The Qualtrics export and demographic append file are in this folder.

qualtrics_file <- "Technology and National Security_August 5, 2026_09.34.xlsx"
demographics_file <- "6a68eb13df3330a45698e7cc_appends.rds"

Read the two files and merge respondent demographics onto the Qualtrics data by vsid.

qualtrics <- read_excel(qualtrics_file)
demographics <- readRDS(demographics_file)

survey <- qualtrics %>%
  filter(vsid != "vsid") %>%
  mutate(vsid = as.character(vsid)) %>%
  left_join(
    demographics %>%
      mutate(vsid = as.character(vsid)),
    by = "vsid"
  )

names(survey) <- gsub("\u00A0", " ", names(survey))

Quick merge check.

## [1] 3081
## [1] 2915
## [1] 3080
## [1] 122

Recode Party ID

Analysis

Respondents per Condition

## # A tibble: 5 × 3
##   condition_agressor condition_autonomy     n
##   <chr>              <chr>              <int>
## 1 China              autonomous           743
## 2 China              human                742
## 3 US                 autonomous           743
## 4 US                 human                742
## 5 <NA>               <NA>                 110

Baseline Intentionality Regression

## 
## Call:
## lm(formula = perceived_intentionality ~ country_scenario * autonomous_ai, 
##     data = baseline_intentionality_regression_data)
## 
## Residuals:
##    Min     1Q Median     3Q    Max 
## -72.11 -13.83  -2.11  18.91  45.79 
## 
## Coefficients:
##                                Estimate Std. Error t value Pr(>|t|)    
## (Intercept)                      62.834      0.892  70.444  < 2e-16 ***
## country_scenario                  9.276      1.260   7.364 2.31e-13 ***
## autonomous_ai                    -8.622      1.264  -6.823 1.08e-11 ***
## country_scenario:autonomous_ai   -1.397      1.783  -0.783    0.433    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 24.15 on 2931 degrees of freedom
## Multiple R-squared:  0.06455,    Adjusted R-squared:  0.06359 
## F-statistic: 67.42 on 3 and 2931 DF,  p-value: < 2.2e-16
##                                     2.5 %    97.5 %
## (Intercept)                     61.084617 64.582504
## country_scenario                 6.806323 11.746365
## autonomous_ai                  -11.099645 -6.144400
## country_scenario:autonomous_ai  -4.893117  2.099169

Intentionality Regression

## 
## t test of coefficients:
## 
##                                 Estimate Std. Error t value  Pr(>|t|)    
## (Intercept)                    62.898304   0.918890 68.4503 < 2.2e-16 ***
## country_scenario                9.147090   1.243596  7.3554 2.475e-13 ***
## autonomous_ai                  -8.998647   1.311613 -6.8608 8.365e-12 ***
## relative_ai                    -0.040334   0.020227 -1.9941  0.046240 *  
## ai_comfort                      1.319646   0.443908  2.9728  0.002976 ** 
## country_scenario:autonomous_ai -1.359874   1.795185 -0.7575  0.448806    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##                                       2.5 %        97.5 %
## (Intercept)                     61.09654799 64.7000591584
## country_scenario                 6.70865422 11.5855258734
## autonomous_ai                  -11.57044972 -6.4268437656
## relative_ai                     -0.07999598 -0.0006730036
## ai_comfort                       0.44923401  2.1900573395
## country_scenario:autonomous_ai  -4.87986283  2.1601141461

Preregistered Analyses

The following sections implement the analysis plan in OSF_PreReg.docx. The primary analyses are intent-to-treat: manipulation-check failures are not excluded from the confirmatory models. The experimental indicators are coded so that the reference category is the human-operated U.S.-strike / Chinese-victim condition.

Preregistered Analysis Data

Preregistered analytic sample sizes.
sample n
Assigned condition and primary outcome 2935
Covariate-complete primary model with demographics 2833

Cell Means for Primary Outcome

Mean perceived intentionality and 95% confidence intervals by experimental cell.
country_condition weapons_condition n mean se ci_low ci_high
Chinese strike / U.S. victim Autonomous AI 737 62.09 0.87 60.38 63.80
Chinese strike / U.S. victim Human-operated 737 72.11 0.85 70.45 73.77
U.S. strike / Chinese victim Autonomous AI 728 54.21 0.94 52.37 56.06
U.S. strike / Chinese victim Human-operated 733 62.83 0.91 61.06 64.61

Primary Preregistered Intentionality Model

This model regresses perceived intentionality on the country-scenario indicator, the autonomous-AI indicator, their interaction, preregistered covariates, and demographic controls. Continuous covariates are demeaned. HC2 heteroskedasticity-consistent standard errors are used for two-sided tests with alpha = .05.

Primary preregistered OLS model with HC2 standard errors.
term estimate hc2_se statistic p_value ci_low ci_high
(Intercept) 63.424 2.626 24.151 0.000 58.275 68.573
country_scenario 9.053 1.253 7.223 0.000 6.595 11.510
autonomous_ai -9.034 1.315 -6.870 0.000 -11.613 -6.456
ai_comfort_c 1.233 0.458 2.693 0.007 0.335 2.131
relative_china_us_ai_c 0.047 0.021 2.257 0.024 0.006 0.088
pidDemocrat 1.463 1.389 1.054 0.292 -1.259 4.186
pidRepublican 3.154 1.454 2.169 0.030 0.303 6.006
age_ordinal -0.629 0.503 -1.252 0.211 -1.615 0.356
genderFemale 0.083 0.930 0.089 0.929 -1.741 1.906
genderOther -5.192 7.316 -0.710 0.478 -19.536 9.153
education_ordinal -0.204 0.346 -0.591 0.554 -0.882 0.473
income_ordinal -0.073 0.326 -0.223 0.823 -0.713 0.567
race_whiteWhite -1.181 1.093 -1.081 0.280 -3.324 0.961
hispanicYes 2.772 1.376 2.015 0.044 0.074 5.470
metroMetropolitan 1.355 1.390 0.974 0.330 -1.371 4.081
country_scenario:autonomous_ai -1.249 1.806 -0.692 0.489 -4.790 2.292

Planned Contrasts for H1-H4

Planned treatment contrasts using HC2 standard errors.
contrast estimate hc2_se statistic p_value ci_low ci_high
H1: Chinese strike vs. U.S. strike among human-operated systems 9.053 1.253 7.223 0.000 6.595 11.510
Autonomous AI effect in U.S.-strike / Chinese-victim condition -9.034 1.315 -6.870 0.000 -11.613 -6.456
Autonomous AI effect in Chinese-strike / U.S.-victim condition -10.284 1.236 -8.318 0.000 -12.708 -7.859
Country-scenario by autonomous-AI interaction -1.249 1.806 -0.692 0.489 -4.790 2.292 
## Warning: `geom_errobarh()` was deprecated in ggplot2 4.0.0.
## ℹ Please use the `orientation` argument of `geom_errorbar()` instead.
## This warning is displayed once every 8 hours.
## Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
## generated.
## `height` was translated to `width`.

Equivalence-Style Tests for Negligible Effects

For the preregistered equivalence-style tests, a meaningful effect is defined as Cohen’s D = .20. The table converts that threshold into raw intentionality points using the standard deviation of the primary model sample.

Equivalence-style tests using +/- 4.97 intentionality points as the meaningful-effect threshold.
contrast estimate lower_bound upper_bound hc2_se p_lower p_upper p_tost equivalent_at_alpha_05
Autonomous AI effect in U.S.-strike / Chinese-victim condition -9.034 -4.967 4.967 1.315 0.999 0 0.999 FALSE
Autonomous AI effect in Chinese-strike / U.S.-victim condition -10.284 -4.967 4.967 1.236 1.000 0 1.000 FALSE 
## `height` was translated to `width`.

Manipulation Check Descriptives

The current Qualtrics export contains manipulation checks for the country whose weapons system launched the missiles and whether the system was autonomous or human-operated. It does not contain a separate victim-country manipulation check.

Overall manipulation-check pass rates.
n country_check_response_n country_check_correct autonomy_check_response_n autonomy_check_correct both_checks_correct
2935 2905 0.738 2935 0.808 0.66
Manipulation-check pass rates by condition.
country_condition weapons_condition n country_check_correct autonomy_check_correct both_checks_correct
Chinese strike / U.S. victim Autonomous AI 737 0.800 0.821 0.722
Chinese strike / U.S. victim Human-operated 737 0.749 0.775 0.644
U.S. strike / Chinese victim Autonomous AI 728 0.693 0.834 0.638
U.S. strike / Chinese victim Human-operated 733 0.709 0.802 0.634

Exploratory Model Among Manipulation-Check Passers

Exploratory primary model among manipulation-check passers only (N = 1880).
term estimate hc2_se statistic p_value ci_low ci_high
(Intercept) 70.867 3.436 20.622 0.000 64.127 77.607
country_scenario 11.398 1.557 7.319 0.000 8.343 14.452
autonomous_ai -12.485 1.725 -7.236 0.000 -15.869 -9.101
ai_comfort_c 1.274 0.577 2.206 0.027 0.141 2.406
relative_china_us_ai_c 0.044 0.029 1.525 0.127 -0.013 0.100
pidDemocrat 0.012 1.869 0.007 0.995 -3.653 3.677
pidRepublican 1.701 1.948 0.873 0.383 -2.120 5.521
age_ordinal -1.100 0.621 -1.772 0.077 -2.317 0.118
genderFemale -0.496 1.141 -0.435 0.664 -2.733 1.742
genderOther -7.844 7.582 -1.035 0.301 -22.714 7.026
education_ordinal -0.770 0.436 -1.763 0.078 -1.626 0.086
income_ordinal -0.250 0.395 -0.633 0.527 -1.025 0.525
race_whiteWhite -0.780 1.418 -0.550 0.582 -3.560 2.000
hispanicYes 3.549 1.800 1.972 0.049 0.019 7.079
metroMetropolitan 0.813 1.777 0.457 0.647 -2.672 4.298
country_scenario:autonomous_ai -0.849 2.264 -0.375 0.708 -5.289 3.592

Secondary Outcome Models

The preregistration lists escalation likelihood, escalation justifiability, and responsibility attributions as exploratory dependent variables. The following models use the same treatment specification and preregistered covariates as the primary model.

Likelihood of escalation

Likelihood of escalation model with HC2 standard errors.
term estimate hc2_se statistic p_value ci_low ci_high
(Intercept) 64.936 2.385 27.229 0.000 60.260 69.612
country_scenario 1.094 1.137 0.962 0.336 -1.136 3.325
autonomous_ai 0.734 1.100 0.667 0.505 -1.422 2.890
ai_comfort_c 0.536 0.414 1.295 0.196 -0.276 1.348
relative_china_us_ai_c 0.007 0.020 0.337 0.736 -0.032 0.046
pidDemocrat 1.646 1.298 1.268 0.205 -0.899 4.191
pidRepublican 0.703 1.336 0.526 0.599 -1.917 3.323
age_ordinal -0.732 0.449 -1.628 0.104 -1.613 0.150
genderFemale -0.977 0.853 -1.146 0.252 -2.649 0.695
genderOther -5.873 5.762 -1.019 0.308 -17.170 5.425
education_ordinal -0.024 0.323 -0.073 0.942 -0.658 0.610
income_ordinal 0.149 0.282 0.528 0.598 -0.404 0.701
race_whiteWhite -1.001 1.027 -0.974 0.330 -3.015 1.014
hispanicYes -2.216 1.383 -1.602 0.109 -4.927 0.496
metroMetropolitan 0.470 1.279 0.367 0.714 -2.039 2.979
country_scenario:autonomous_ai -1.662 1.635 -1.017 0.309 -4.868 1.544

Justifiability of escalation

Justifiability of escalation model with HC2 standard errors.
term estimate hc2_se statistic p_value ci_low ci_high
(Intercept) 60.095 2.709 22.183 0.000 54.783 65.407
country_scenario 2.339 1.324 1.767 0.077 -0.257 4.935
autonomous_ai -2.397 1.316 -1.821 0.069 -4.978 0.184
ai_comfort_c 1.697 0.492 3.447 0.001 0.732 2.663
relative_china_us_ai_c 0.012 0.023 0.512 0.609 -0.033 0.057
pidDemocrat -0.244 1.414 -0.173 0.863 -3.017 2.528
pidRepublican 4.289 1.503 2.853 0.004 1.341 7.237
age_ordinal 0.250 0.532 0.470 0.639 -0.793 1.293
genderFemale -2.916 0.988 -2.952 0.003 -4.853 -0.979
genderOther 2.089 6.595 0.317 0.751 -10.842 15.021
education_ordinal 0.175 0.370 0.472 0.637 -0.550 0.899
income_ordinal -0.039 0.330 -0.119 0.905 -0.686 0.607
race_whiteWhite -2.583 1.123 -2.299 0.022 -4.786 -0.380
hispanicYes 1.205 1.461 0.825 0.410 -1.660 4.071
metroMetropolitan 1.986 1.516 1.310 0.190 -0.986 4.959
country_scenario:autonomous_ai 1.372 1.899 0.723 0.470 -2.352 5.096 
## `height` was translated to `width`.

Responsibility: Senior political leaders

Treatment coefficients for responsibility attributed to Senior political leaders.
term estimate hc2_se statistic p_value ci_low ci_high
country_scenario 5.117 1.303 3.927 0.000 2.562 7.671
autonomous_ai 1.545 1.347 1.147 0.251 -1.096 4.187
country_scenario:autonomous_ai -1.569 1.854 -0.846 0.397 -5.204 2.066

Responsibility: Military officials

Treatment coefficients for responsibility attributed to Military officials.
term estimate hc2_se statistic p_value ci_low ci_high
country_scenario 0.877 1.208 0.726 0.468 -1.492 3.246
autonomous_ai 0.474 1.219 0.389 0.697 -1.917 2.865
country_scenario:autonomous_ai 0.462 1.722 0.269 0.788 -2.914 3.839

Responsibility: Human operator

Treatment coefficients for responsibility attributed to Human operator.
term estimate hc2_se statistic p_value ci_low ci_high
country_scenario 1.767 1.413 1.251 0.211 -1.005 4.538

Responsibility: AI weapons system

Treatment coefficients for responsibility attributed to AI weapons system.
term estimate hc2_se statistic p_value ci_low ci_high
country_scenario -1.341 1.657 -0.809 0.419 -4.591 1.91 
## `height` was translated to `width`.

Exploratory Moderation Models

These models extend the primary specification by interacting the country-scenario indicator, autonomous-AI indicator, and moderator. Continuous moderators are demeaned. Models include Verasight demographic controls; moderation models not focused on party identification also control for pid.

Focal three-way interaction terms from exploratory moderation models.
moderator term estimate hc2_se statistic p_value ci_low ci_high
AI comfort country_scenario:autonomous_ai:ai_comfort_c 4.408 1.774 2.485 0.013 0.930 7.887
U.S. AI effectiveness country_scenario:autonomous_ai:us_ai_effectiveness_c 0.087 0.081 1.076 0.282 -0.072 0.246
Chinese AI effectiveness country_scenario:autonomous_ai:china_ai_effectiveness_c 0.233 0.077 3.017 0.003 0.082 0.384
South Korean AI effectiveness country_scenario:autonomous_ai:southkorea_ai_effectiveness_c 0.225 0.083 2.700 0.007 0.062 0.389
Chinese minus U.S. AI effectiveness country_scenario:autonomous_ai:relative_china_us_ai_c 0.166 0.080 2.080 0.038 0.010 0.323
Party identification country_scenario:autonomous_ai:pidDemocrat -11.599 5.424 -2.139 0.033 -22.234 -0.965
Party identification country_scenario:autonomous_ai:pidRepublican -10.248 5.628 -1.821 0.069 -21.284 0.788 
## `height` was translated to `width`.

Exploratory Downstream Association Models

The preregistration describes mediation-style analyses as exploratory and associational. The following models test whether perceived intentionality is associated with downstream escalation and responsibility judgments, controlling for treatment assignment, AI comfort, relative China-U.S. AI effectiveness, party identification, and Verasight demographic controls.

Association between perceived intentionality and downstream judgments using HC2 standard errors.
outcome term estimate hc2_se statistic p_value ci_low ci_high
Likelihood of escalation perceived_intentionality_c 0.322 0.019 17.348 0.000 0.286 0.359
Justifiability of escalation perceived_intentionality_c 0.406 0.022 18.874 0.000 0.364 0.448
Responsibility: Senior political leaders perceived_intentionality_c 0.383 0.021 18.286 0.000 0.342 0.424
Responsibility: Military officials perceived_intentionality_c 0.272 0.020 13.275 0.000 0.232 0.312
Responsibility: Human operator perceived_intentionality_c 0.141 0.033 4.272 0.000 0.076 0.206
Responsibility: AI weapons system perceived_intentionality_c -0.015 0.037 -0.390 0.696 -0.088 0.059 
## `height` was translated to `width`.

Main Outcome Plots

Intentionality

Escalation

Justification

Responsibility

## `geom_line()`: Each group consists of only one observation.
## ℹ Do you need to adjust the group aesthetic?
## `geom_line()`: Each group consists of only one observation.
## ℹ Do you need to adjust the group aesthetic?
## `geom_line()`: Each group consists of only one observation.
## ℹ Do you need to adjust the group aesthetic?
## `geom_line()`: Each group consists of only one observation.
## ℹ Do you need to adjust the group aesthetic?
## `geom_line()`: Each group consists of only one observation.
## ℹ Do you need to adjust the group aesthetic?
## `geom_line()`: Each group consists of only one observation.
## ℹ Do you need to adjust the group aesthetic?

Covariate and Moderator Plots

Covariate Distributions

AI Comfort and Intentionality

## `geom_smooth()` using formula = 'y ~ x'
## Warning: Removed 353 rows containing missing values or values outside the scale range
## (`geom_point()`).

Relative AI Effectiveness and Intentionality

## `geom_smooth()` using formula = 'y ~ x'
## Warning: Removed 172 rows containing missing values or values outside the scale range
## (`geom_point()`).

Party Identification and Intentionality

## Warning: Removed 146 rows containing missing values or values outside the scale range
## (`geom_point()`).

U.S. AI Effectiveness and Intentionality

## `geom_smooth()` using formula = 'y ~ x'
## Warning: Removed 264 rows containing missing values or values outside the scale range
## (`geom_point()`).

Chinese AI Effectiveness and Intentionality

## `geom_smooth()` using formula = 'y ~ x'
## Warning: Removed 228 rows containing missing values or values outside the scale range
## (`geom_point()`).

South Korean AI Effectiveness and Intentionality

## `geom_smooth()` using formula = 'y ~ x'
## Warning: Removed 219 rows containing missing values or values outside the scale range
## (`geom_point()`).