0.1 ## R Markdown

###ABSTRACT

This paper investigates whether external military intervention is associated with increased conflict intensity, focusing on two African cases with contrasting patterns: The Democratic Republic of Congo (DRC) and Uganda. Using the UCDP/PRIO Armed Conflict Dataset (v24.1), I analyze all conflict years from 1962 to 2023 in both countries, classifying each year based on the presence of external military actors and whether the conflict escalated to war-level intensity (defined as ≥1,000 battle-related deaths).

Descriptive analysis reveals a strong positive association between intervention and war-level conflict in the DRC: 41.2% of intervention years reached war intensity, compared to only 15.4% of non-intervention years. Uganda, by contrast, exhibits an inverted pattern, where war years were more frequent in the absence of external intervention. These descriptive trends are visualized through normalized stacked bar charts and further examined using logistic regression models, including interaction terms that allow for country-specific effects.

Regression results confirm that the effect of intervention differs by context. In the DRC, intervention is associated with a modest, statistically insignificant increase in the likelihood of war. In Uganda, however, intervention significantly reduces the probability of war-level conflict, with the interaction term (Uganda × Intervention) reaching statistical significance. These findings suggest that external intervention is not inherently escalatory or de-escalatory; rather, its effect is conditioned by local political dynamics, state capacity, and the identity and objectives of intervening actors.

1 Data Preparation

conflict <- read_csv("data/UcdpPrioConflict_v24_1.csv")

1.1 Filtering for DRC

drc <- conflict %>%
  filter(location == "DR Congo (Zaire)") %>%
  mutate(external_intervention = if_else(!is.na(side_a_2nd) | !is.na(side_b_2nd), "Yes", "No"))

1.2 Summarizing Intensity

drc_summary <- drc %>%
  group_by(year, external_intervention) %>%
  summarise(intensity = max(intensity_level, na.rm = TRUE), .groups = "drop")

summary_table <- drc_summary %>%
  mutate(intensity_label = if_else(intensity == 2, "War", "Minor")) %>%
  group_by(external_intervention, intensity_label) %>%
  summarise(n_years = n(), .groups = "drop") %>%
  pivot_wider(names_from = intensity_label, values_from = n_years, values_fill = 0) %>%
  mutate(total_years = War + Minor,
         war_rate = round((War / total_years) * 100, 1))

summary_table

#INTRODUCTION

Why do some conflicts escalate into full-scale wars while others remain relatively contained? One possible factor is the presence of external military intervention. While foreign troops are often deployed to stabilize violent situations, their involvement can also alter the dynamics of conflict in unpredictable ways—sometimes escalating violence, other times reducing it.

In this study, I explore the relationship between external military intervention and the intensity of armed conflict in two countries: the Democratic Republic of Congo (DRC) and Uganda. Both have experienced decades of political violence, but with markedly different patterns of foreign involvement and outcomes. Using data from the UCDP/PRIO Armed Conflict Dataset (v24.1), I ask whether conflict years with foreign military actors are more or less likely to escalate into full-blown war—defined here as a year with 1,000 or more battle-related deaths

I begin with the DRC, a country that has hosted repeated waves of intervention by neighboring states and international actors. I then turn to Uganda, which has experienced more centralized conflicts with relatively fewer foreign boots on the ground. By comparing the two, I aim to understand whether the effect of intervention depends on country context—and what that might mean for international responses to conflict in Africa.

#DATA, VARIABLES, AND METHODOLOGY

For this study, I use version 24.1 of the UCDP/PRIO Armed Conflict Dataset, which records instances of state-based armed conflict globally between 1946 and 2023. This dataset defines a conflict as a contested incompatibility over government or territory between a state and one or more organized groups, with a minimum threshold of 25 battle-related deaths in a calendar year. It also includes information on the presence of secondary state actors (i.e., foreign governments sending troops), allowing me to identify years with external military intervention.

I restrict my analysis to two countries: The Democratic Republic of Congo (DRC) and Uganda, covering the period 1962 to 2023. Each observation in the dataset represents a country-year in which conflict occurred. My unit of analysis is therefore the conflict-year, and I analyze only those years in which the dataset records state-based conflict activity.

##KEY VARIABLES

  1. Conflict Intensity (Dependent Variable)

I use the intensity_level variable to create a binary measure of war-level conflict. In line with UCDP definitions: • war_binary = 1 for years with ≥ 1,000 battle-related deaths (intensity_level = 2), and • war_binary = 0 for years with 25–999 deaths (intensity_level = 1). This binary outcome captures whether a conflict-year reached “war-level” severity.

  1. External Military Intervention (Main Independent Variable)

I code external intervention using the presence of any secondary state actor listed in the side_a_2nd or side_b_2nd fields. In the UCDP/PRIO dataset, side_a_2nd and side_b_2nd refer to secondary state actors—foreign governments that send military forces to actively support one of the primary conflict parties. side_a_2nd records all foreign states that intervene militarily in support of Side A, which is typically the recognized government in intrastate conflicts. side_b_2nd captures foreign states that intervene on behalf of Side B, which usually consists of one or more rebel or opposition groups. These variables indicate the presence of external military involvement in a given conflict-year. They are not limited to peacekeeping forces; they include any troop deployments that align with one side of the incompatibility—regardless of the form or intent of the intervention. In this study, I use the presence of any entry in either side_a_2nd or side_b_2nd as the operational definition of external military intervention.

If any foreign military actor is recorded, I assign a value of 1 to a binary variable called intervention_binary; otherwise, it is coded 0. This approach captures formal military presence by foreign states, which may include peacekeeping, joint operations, or unilateral intervention

  1. Country Identifier

To allow for cross-country comparison and interaction effects, I include a dummy variable for Uganda, using the DRC as the reference category in regression models.

##METHODOLOGICAL APPROACH

I begin with descriptive analysis—comparing the proportion of war and minor conflict years across intervention and non-intervention periods in each country. These are visualized through normalized stacked bar charts, which make it easy to assess how the distribution of conflict intensity varies with the presence of foreign military actors. Next, I estimate logistic regression models to assess whether intervention significantly predicts the likelihood of war-level conflict. I start with a baseline model pooling both countries, testing the general effect of external intervention. I then introduce a country dummy and, finally, an interaction term between country and intervention. This allows me to test whether the effect of intervention differs between Uganda and the DRC. All models use the binary war outcome (war_binary) as the dependent variable and are estimated using a binomial logit link function. I interpret both statistical significance and direction of effects, and discuss potential alternative explanations including reverse causality and strategic differences in intervention types.

##ANALYSIS

An analysis of the UCDP/PRIO Armed Conflict Dataset (v24.1) for the Democratic Republic of Congo (DRC) reveals a noteworthy descriptive pattern in the relationship between external military intervention and conflict intensity. Between 1962 and 2023, the dataset records 30 conflict years in the DRC, of which 17 featured external military intervention—defined by the presence of foreign troops coded in the side_a_2nd or side_b_2nd variables. Among these 17 intervention years, 41.2% escalated into war-level conflict, defined by the UCDP as ≥1,000 battle-related deaths within a calendar year. In contrast, among the 13 conflict years without foreign military involvement, only 15.4% reached this threshold. These figures translate into a striking pattern: conflict years with external military intervention were nearly three times more likely to escalate into full-scale war than those without such involvement.

To visualize this relationship, I used a normalized stacked bar chart, with each bar representing 100% of the conflict years within each intervention category. The “war” segment (≥1,000 deaths) appears in red, while “minor” conflicts (25–999 deaths) are shown in gray. In non-intervention years, only 2 of 13 years (15.4%) reached war intensity, whereas 7 of 17 intervention years (41.2%) did. This provides an intuitive illustration of how conflict severity appears distributed across the two groups.

2 DRC Conflict Intensity by Intervention

drc_plot_data <- drc_summary %>%
  mutate(intensity_label = if_else(intensity == 2, "War", "Minor")) %>%
  group_by(external_intervention, intensity_label) %>%
  summarise(n = n(), .groups = "drop") %>%
  group_by(external_intervention) %>%
  mutate(percent = n / sum(n) * 100)

# Plot

ggplot(drc_plot_data, aes(x = external_intervention, y = percent, fill = intensity_label)) +
  geom_col(width = 0.6) +
  scale_fill_manual(values = c("Minor" = "gray", "War" = "darkred")) +
  labs(title = "Conflict Intensity by Intervention Status",
       subtitle = "Proportion of Minor vs War Years in DRC (UCDP/PRIO v24.1)",
       x = "External Intervention", y = "Percentage of Conflict Years", fill = "Intensity") +
  theme_minimal()

While the data clearly supports the hypothesis that intervention years tend to be deadlier, it is crucial to approach these results with caution. The observed association between intervention and higher conflict intensity does not, by itself, establish causality. In fact, a plausible alternative explanation is reverse causality: foreign military involvement may often be a reaction to already escalating violence rather than the cause of that escalation. Thus, while the stacked bar chart reflects a normalized distribution of conflict intensity conditional on intervention status, it is best understood as descriptive rather than explanatory

From a scholarly perspective, these findings suggest that the presence of external military actors coincides with more intense conflict periods in the DRC. This insight is valuable for both empirical and policy analysis, as it raises further questions about the mechanisms by which interventions may interact with domestic conflict dynamics. While this pattern highlights a strong association between external intervention and war-level intensity, it is important to emphasize that correlation does not imply causation, and the observed relationship may be influenced by underlying or reverse causal dynamics. A deeper causal inquiry requires statistical modeling, which I attempt in the next section.

3 DRC and Uganda Comparative Data

df <- conflict %>%
  filter(location %in% c("DR Congo (Zaire)", "Uganda"), year >= 1962) %>%
  mutate(
    intervention_binary = if_else(!is.na(side_a_2nd) | !is.na(side_b_2nd), 1, 0),
    war_binary = if_else(intensity_level == 2, 1, 0),
    country = if_else(location == "DR Congo (Zaire)", "DRC", "Uganda")
  )

4 Summary Table: War Rates by Country and Intervention

df %>%
  group_by(country, intervention_binary) %>%
  summarise(
    war_years = sum(war_binary),
    total_years = n(),
    war_rate = round(100 * war_years / total_years, 1),
    .groups = "drop"
  )

###EXTENDING THE INQUIRY: INTRODUCING UGANDA FOR COMPARATIVE ANALYSIS

While the Democratic Republic of Congo presents a compelling case in which external intervention appears to coincide with elevated conflict intensity, these descriptive patterns alone raise as many questions as they answer. Chief among them is whether this observed relationship is unique to the DRC or indicative of a broader trend. To investigate this further, I introduce Uganda as a comparative case. Like the DRC, Uganda has experienced multiple episodes of state-based armed conflict over the past six decades, including prolonged insurgencies and foreign military involvement. However, Uganda’s conflict dynamics, governance structure, and intervention contexts differ in meaningful ways.

By placing Uganda alongside the DRC, I aim to assess whether the association between external intervention and war-level conflict holds consistently across cases, or whether it is shaped by country-specific factors. This comparative step enables us to explore whether intervention is inherently escalatory or if its impact is conditional on domestic political, institutional, or strategic variables. In the next section, I present descriptive patterns from Uganda’s conflict history, followed by a unified regression model that tests for differential effects across the two countries.

##DESCRIPTIVE PATTERNS IN UGANDA’S CONFLICT YEARS

Uganda offers an instructive comparative case to the DRC, having also experienced state-based armed conflict across multiple decades. These include major episodes such as the war against the Lord’s Resistance Army (LRA), insurgencies in the north, and regional tensions involving actors from Sudan and Rwanda. However, unlike the DRC, Uganda’s state apparatus has demonstrated relatively higher levels of cohesion and territorial control, and foreign military engagement has generally been more targeted or aligned with state objectives.

An analysis of the UCDP/PRIO Armed Conflict Dataset (v24.1) from 1962 to 2023 reveals 21 conflict years for Uganda. Of these, 9 years involved external military intervention as indicated by the presence of secondary state actors (side_a_2nd or side_b_2nd), while 12 conflict years occurred without such involvement. Interestingly, only 1 of the 9 intervention years (11.1%) escalated into war-level intensity (≥1,000 deaths), while 5 of the 12 non-intervention years (41.7%) reached this threshold. In contrast to the DRC, where intervention years were more likely to be war years, Uganda exhibits the opposite pattern: conflict years without foreign intervention were more likely to escalate into war-level violence

This difference is visually apparent in the corresponding stacked bar chart. The majority (88.9%) of intervention years in Uganda were minor conflicts, while only 58.3% of non-intervention years fell into that category. These descriptive statistics suggest a preliminary association between foreign military involvement and lower conflict intensity in Uganda. However, as in the case of the DRC, I exercise caution in interpreting these patterns. The lower frequency of war-level conflict during intervention years does not in itself establish a causal link. It is possible that foreign involvement occurred precisely in periods when conflict was already being contained or when state forces had the upper hand.

Nevertheless, this reversal in trend compared to the DRC raises an important comparative question: Does the impact of external intervention vary systematically by context? In the following section, I bring these two cases into a single analytic framework to formally test whether the effect of intervention differs between Uganda and the DRC.

5 Visualization: War Rate by Country & Intervention

df %>%
  mutate(intervention = if_else(intervention_binary == 1, "Yes", "No")) %>%
  group_by(country, intervention) %>%
  summarise(war_rate = mean(war_binary) * 100, .groups = "drop") %>%
  ggplot(aes(x = country, y = war_rate, fill = intervention)) +
  geom_col(position = "dodge", width = 0.6) +
  labs(title = "War-Level Conflict Rates in DRC and Uganda (1962–2023)",
       subtitle = "By External Military Intervention Presence",
       x = "Country", y = "War Years (%)", fill = "External Intervention") +
  scale_fill_manual(values = c("No" = "gray", "Yes" = "darkred")) +
  theme_minimal()

This figure presents the proportion of conflict years classified as “war” in the Democratic Republic of Congo (DRC) and Uganda between 1962 and 2023, based on the UCDP/PRIO Armed Conflict Dataset (v24.1). In this dataset, a “war” year is defined as any year with at least 1,000 battle-related deaths, corresponding to an intensity_level of 2. Years with 25–999 deaths are classified as “minor” (intensity_level 1). The graph shows that in Uganda, nearly 40% of conflict years without external military intervention were war years, compared to less than 10% of years with external intervention. This indicates that most of Uganda’s war years occurred in the absence of foreign troop support. In contrast, the DRC displays the opposite pattern: a higher proportion of war years occurred during years of external intervention than during years without it. These visual patterns suggest that in the DRC, intervention may be associated with conflict escalation, while in Uganda, war intensity appears to have occurred largely without foreign involvement. It is important to note that the graph shows descriptive proportions and does not establish causation. The operational threshold for war (≥1,000 deaths/year) is derived directly from the UCDP codebook but is not labeled on the graph itself; this definition should be kept in mind when interpreting the results.

6 Logistic Regression Models

model1 <- glm(war_binary ~ intervention_binary, family = binomial(), data = df)
model2 <- glm(war_binary ~ intervention_binary + country, family = binomial(), data = df)
model_interact <- glm(war_binary ~ country * intervention_binary, family = binomial(), data = df)

6.1 Stargazer Output

print(
  stargazer(model1, model2, model_interact,
            type = "html",
            title = "Logistic Regression Results: Predicting War-Level Conflict",
            dep.var.labels = "War Year (Binary)",
            covariate.labels = c(
              "External Intervention",
              "Uganda (vs. DRC)",
              "Uganda × Intervention"
            ),
            omit.stat = c("ll", "aic"),
            star.cutoffs = c(0.1, 0.05, 0.01),
            no.space = TRUE)
)
Logistic Regression Results: Predicting War-Level Conflict
Dependent variable:
War Year (Binary)
(1) (2) (3)
External Intervention -0.645 -0.556 0.647
(0.516) (0.549) (0.878)
Uganda (vs. DRC) -2.696*
(1.416)
Uganda × Intervention 0.254 1.356
(0.544) (0.856)
Constant -0.802** -0.974* -1.792**
(0.334) (0.502) (0.764)
Observations 84 84 84
Note: p<0.1; p<0.05; p<0.01
[1] “”
[2] “ ”
[3] “ ”
[4] “ ”
[5] “ ”
[6] “ ”
[7] “ ”
[8] “ ”
[9] “ ”
[10] “ ”
[11] “ ”
[12] “ ”
[13] “ ”
[14] “ ”
[15] “ ”
[16] “ ” [17] “
Logistic Regression Results: Predicting War-Level Conflict
Dependent variable:
War Year (Binary)
(1) (2) (3)
External Intervention -0.645 -0.556 0.647
(0.516) (0.549) (0.878)
Uganda (vs. DRC) -2.696*
(1.416)
Uganda × Intervention 0.254 1.356
(0.544) (0.856)
Constant -0.802** -0.974* -1.792**
(0.334) (0.502) (0.764)
Observations 84 84 84
Note: p<0.1; p<0.05; p<0.01

”

##INTERPRETATION

The results are revealing. The coefficient for external intervention in the DRC is positive (0.647) but not statistically significant, suggesting a possible but inconclusive escalation effect. The Uganda dummy variable is also positive (1.356), indicating that, in the absence of intervention, Uganda has a slightly higher baseline probability of war compared to the DRC. However, the critical result lies in the interaction term: Uganda × Intervention is negative (−2.696) and statistically significant at the 0.1 level. This means that, in Uganda, external military intervention is associated with a significantly lower likelihood of war-level conflict, in contrast to the DRC, where no such reduction is observed.

In practical terms, this suggests that external military intervention operates differently across these two contexts. In the DRC, intervention neither significantly increases nor decreases the odds of war, whereas in Uganda, the presence of foreign forces is associated with a meaningful de-escalation of conflict intensity. This divergence may reflect underlying differences in political institutions, military capacity, external actors’ objectives, or the sequencing and timing of intervention. For instance, in the DRC, external intervention has often been complex, fragmented, and driven by competing geopolitical interests. The country has hosted numerous multinational forces, including the United Nations Organization Stabilization Mission in the DRC (MONUSCO)—one of the largest and longest-running UN peacekeeping operations—as well as regional military actors such as Rwanda, Uganda, Angola, and Zimbabwe, who have intervened either to support proxy groups or assert regional influence during the Congo Wars. These overlapping and sometimes adversarial interventions have arguably contributed to political fragmentation and prolonged instability.

In contrast, Uganda’s experience with external military presence has been more limited and strategically coordinated. Interventions have generally involved support from international partners such as the United States through counter-LRA operations, or African Union forces cooperating with the Ugandan government in stabilization efforts. Crucially, these deployments were often aligned with the Ugandan state’s counterinsurgency agenda and did not involve large-scale foreign rivalries on Ugandan soil.

Thus, beyond the statistical patterns, the differing effects of intervention in the DRC and Uganda reflect deeper structural and geopolitical contrasts. The DRC’s fragmented state authority, history of regional proxy wars, and overlapping interventions stand in sharp contrast to Uganda’s more centralized governance and coordinated military partnerships. This reinforces the argument that external intervention cannot be treated as a uniform phenomenon—its impact is shaped by the identity, alignment, and goals of the interveners, as well as the internal political order of the host state.

#ALTERNATIVE EXPLANATIONS

While the data suggest that years of external military intervention in the Democratic Republic of Congo (DRC) are more likely to coincide with war-level conflict (defined by the UCDP/PRIO dataset as ≥1,000 battle-related deaths per year), the same is not true for Uganda. In Uganda, war years appear more frequent in the absence of external intervention. These contrasting patterns call for a careful examination of possible alternative explanations beyond the assumption that intervention directly causes or reduces conflict intensity. Contextual, structural, and strategic factors likely shape the relationship between intervention and war in each country differently.

In the DRC, one of the most cited explanations relates to the country’s abundance of natural resources and the strategic interests they attract. The DRC has long been a site of competition over valuable minerals such as gold, coltan, cobalt, and diamonds. Scholars have shown how such resource wealth often fuels armed conflict and incentivizes the involvement of foreign actors, sometimes under the guise of peace enforcement or humanitarian intervention (Le Billon, 2001; Reno, 1999). In this light, interventions may not always aim to de-escalate violence but to protect or access strategic interests. Consequently, conflict intensity may be exacerbated rather than reduced during periods of external involvement.

Additionally, timing and causality are critical. External interventions are often reactive, deployed after violence has escalated. In these cases, war may precede intervention, not result from it. This complicates any assumption of causation. As Fortna (2008) emphasizes, peacekeeping or military intervention often occurs in response to a deteriorating situation, and unless timing is explicitly controlled for, we risk confusing correlation for causation. The DRC’s interventions often occurred during or after spikes in violence, particularly during the First and Second Congo Wars and their aftermath, suggesting that the interventions were symptoms, not causes, of war-level conflict.

A third factor is the structural nature of conflict in the two countries. The DRC has experienced multiple internationalized intrastate conflicts (UCDP Conflict Type 4), which by definition involve both internal and external parties. These conflicts tend to be more complex and intense due to the number of actors involved and the regional dynamics they introduce (Gleditsch et al., 2002; Pettersson & Öberg, 2020). Uganda, by contrast, has mostly experienced internal intrastate conflicts (Type 3), such as the wars against the National Resistance Army in the early 1980s and the Lord’s Resistance Army (LRA) from the late 1980s to the 2000s. These were brutal and long-lasting but were largely contained within national borders, and external intervention was either minimal or limited to late-stage international involvement.

Uganda’s pattern also reflects differences in state capacity and central control. Although Uganda has faced significant political violence, especially in its early post-independence years, its government has generally maintained greater territorial control and centralized authority than the DRC. The Ugandan state has historically managed to suppress rebellion internally without relying heavily on foreign troop presence. In fact, Uganda has often acted as an external intervener in neighboring countries such as South Sudan and the DRC itself. Thus, the lack of foreign intervention within Uganda does not imply peace, but rather that intense conflict occurred in a domestic context, with the state as the primary actor in both violence and conflict resolution (Tripp, 2010).

Finally, the objectives and nature of the interventions themselves differ between the two countries. In the DRC, many interventions have involved complex multinational arrangements—such as MONUC and later MONUSCO—with ambiguous mandates and limited effectiveness. Some foreign actors have even been accused of fueling conflict indirectly by backing rebel factions or prioritizing strategic corridors over civilian protection (Autesserre, 2010). In Uganda, where intervention did occur, it was often limited or late-stage, such as international pressure on the government during the LRA insurgency. These interventions tended to support already ongoing state efforts, rather than reshape conflict dynamics dramatically.

In sum, the differing patterns of war and intervention in the DRC and Uganda likely reflect a combination of strategic interests, timing, state capacity, and the structural characteristics of the conflicts themselves. The relationship between intervention and violence is far from linear, and simple correlations between troop presence and intensity can mask the deeper political and historical dynamics at play in each context.