MPA 2026 - Gender, Source Credibility, and the Third-Person Effect in Science Misinformation

Poster

knitr::include_graphics("sophia poster.jpg")

Poster Text

INTRODUCTION

INFORMATION DISORDER & SOURCE CREDIBILITY

• Information disorder: Involves the widespread presence of misinformation, disinformation, and malinformation that blurs the line between credible and non-credible information, specifically throughout social media (Wardle & Derakhshan, 2017).

• The perceived trustworthiness and competence of a source that could impact one’s acceptance of misinformation. When individuals misjudge source credibility, they are at greater risk of accepting misinformation, which can shape beliefs and persist even after being corrected (Lewandowsky et al., 2017).

GENDER (STEREOTYPE CONTENT MODEL)

• Proposes that individuals perceive women as warmer (e.g., trustworthy) and men as competent (Cuddy et al., 2008; Fiske et al., 2002).

• These stereotypes may structure how individuals process online information by shaping initial judgments of whether a source is trustworthy or competent.

• Hypothesis: Gender may impact source credibility evaluations.

THIRD-PERSON EFFECT

• Tendency for people to believe that others are more influenced by media messages than they are themselves (Davison, 1983).

• Individuals may underestimate their own vulnerability to misinformation, while assuming others are more easily misled (Davison, 1983). This may connect to stereotypes about caretaking and responsibility for others (Cuddy et al., 2008; Fiske et al., 2002).

• Hypothesis: Gender may impact the third-person effect.

CURRENT STUDY

Public trust in science is increasingly shaped by social media, where perceptions of credibility and influence may drive misinformation acceptance. However, little is known about how the role gender plays. This study examines whether men and women differ in their assessment of source credibility and the degree to which they demonstrate the third-person effect. By integrating these factors, this work aims to better explain how people interpret and respond to information in complex online environments and to develop strategies to improve how individuals engage with scientific information online.

METHOD

Undergraduate and community participants (N = 183) viewed a set of mock social media posts with accurate or misinformation content across 15 topics (e.g., depression, love, intelligence) and answered questions about how they were influenced by the posts and how influenced they thought others would be by the posts (i.e., the third-person effect; McLeod et al., 1997). They also completed ratings of source credibility focusing on trustworthiness and competence (McCroskey & Teven, 1999).

DISCUSSION

This study examined how gender relates to perceptions of source credibility and the third-person effect in the context of online misinformation. Drawing on the stereotype content model (Cuddy et al., 2008; Fiske et al., 2002), which proposes that individuals evaluate others along dimensions of warmth (e.g., trustworthiness) and competence, we considered how gendered expectations might shape information processing, specifically the third-person effect.

Across analyses, a clear third-person effect was observed, with participants consistently reporting that others were more influenced by the social media posts than they were themselves. This pattern aligns with prior research demonstrating that individuals tend to underestimate their own susceptibility to media influence while overestimating that of others (Davison, 1983). This effect emerged regardless of gender, reinforcing findings that the third-person effect reflects a general cognitive bias rather than gendered social perceptions. While gender did not strongly shape the third-person effect directly, perceptions of source credibility played a role in how individuals viewed the influence of misinformation.

First, the results showed a significant negative relationship between trustworthiness and the third-person effect (Figure 1). This suggests that when a source is perceived as more trustworthy, individuals are less likely to believe that others are more influenced than themselves. Contrary to expectations based on the stereotype content model, there were no significant gender differences in this relationship. This indicates that although stereotypes often associate women with warmth and trustworthiness, these perceptions may not meaningfully translate into differences in how men and women experience the third-person effect in this context.

Second, competence also showed a significant negative relationship with the third-person effect, and this relationship was slightly stronger than that of trustworthiness (Figure 2). This suggests that perceptions of competence may be a notable factor in shaping how people evaluate the influence of information on themselves versus others. Unlike trustworthiness, there was a marginal effect of gender and a marginal moderation by gender, indicating that the relationship between competence and the third-person effect may differ somewhat between men and women. Specifically, the pattern suggests that as perceptions of competence decrease, men may show a greater increase in the third-person effect compared to women. Although these effects were only marginal, they hint that competence-related stereotypes (e.g., men being viewed as more competent) may play a subtle role in how individuals process information.

These results have important implications for misinformation research and future media literacy efforts. Future research should continue to examine the conditions under which gender stereotypes become more influential. It would also be useful to explore how different types of content, like political or emotionally charged posts, influence these effects.

REFERENCES

Cuddy, A. J. C., Fiske, S. T., & Glick, P. (2008). Warmth and Competence as Universal Dimensions of Social Perception: The Stereotype Content Model and the BIAS Map. In Advances in Experimental Social Psychology (Vol. 40, pp. 61–149). Academic Press. https://doi.org/10.1016/S0065-2601(07)00002-0

Davison, W. P. (1983). The Third-Person Effect in Communication. Public Opinion Quarterly, 47(1), 1. https://doi.org/10.1086/268763

Fiske, S. T., Cuddy, A. J. C., Glick, P., & Xu, J. (2002). A model of (often mixed) stereotype content: Competence and warmth respectively follow from perceived status and competition. Journal of Personality and Social Psychology, 82(6), 878–902. https://doi.org/10.1037/0022-3514.82.6.878

Lewandowsky, S., Ecker, U. K. H., & Cook, J. (2017). Beyond misinformation: Understanding and coping with the “post-truth” era. Journal of Applied Research in Memory and Cognition, 6(4), 353–369. https://doi.org/10.1016/j.jarmac.2017.07.008

McCroskey, J. C., & Teven, J. J. (1999). Goodwill: A reexamination of the construct and its measurement. Communication Monographs, 66(1), 90–103. https://doi.org/10.1080/03637759909376464

McLeod, D. M., Eveland, W. P., & Nathanson, A. I. (1997). Support for Censorship of Violent and Misogynic Rap Lyrics: An Analysis of the Third-Person Effect. Communication Research, 24(2), 153–174. https://doi.org/10.1177/009365097024002003

Wardle, C., & Derakhshan, H. (2017). Information disorder: Toward an interdisciplinary framework for research and policymaking (Vol. 27). Council of Europe Strasbourg.

Code

Load Libraries

library(psych)
library(dplyr)
library(expss)
library(sjPlot)
library(ggplot2)

Load Data

df <- read.csv(file="data_study2_clean.csv", header=T)

df$tpe_self <- (df$Q28_1 + df$Q28_2)/2
df$tpe_othr <- (df$Q28_3 + df$Q28_5)/2
df$source_comp <- (df$Q27_1 + df$Q27_2 + df$Q27_3 + df$Q27_4 + df$Q27_5 + df$Q27_6)/6
df$source_trust <- (df$Q94_1 + df$Q94_2 + df$Q94_3)/3

df <- df %>%
  mutate(Q28_6 = case_when(
    Q28_1 == Q28_3 ~ 0,
    Q28_1 <  Q28_3 ~ 1,
    Q28_1 >  Q28_3 ~ -1
  )) %>%
  mutate(Q28_7 = case_when(
    Q28_2 == Q28_5 ~ 0,
    Q28_2 <  Q28_5 ~ 1,
    Q28_2 >  Q28_5 ~ -1
  ))
df$tpe_diff <- (df$Q28_6 + df$Q28_7)

df$gen[is.na(df$Q4)] <- NA
df$gen[df$Q4 == 1] <- "W"
df$gen[df$Q4 == 2] <- "M"
df$gen[df$Q4 == 3] <- "N"
df$gen[df$Q4 == 4] <- "N"
df$gen[df$Q4 == 5] <- "A"
df$gen[df$Q4 == 6] <- "T"
df$gen[df$Q4 == 7] <- "A"
df$gen[df$Q4 == "1,5"] <- "W"
df$gen[df$Q4 == "1,6"] <- "TW"
df$gen[df$Q4 == "2,4,6"] <- "N"
df$gen[df$Q4 == "2,5"] <- "M"

df$gen2 <- NA
df$gen2[df$gen == "M"] <- 0
df$gen2[df$gen == "W"] <- 1
df$gen2[df$gen == "TW"] <- 1
df$gen2 <- as.factor(df$gen2)

df <- subset(df, !is.na(gen2))

reg2 <- lm(tpe_diff ~ source_trust*gen2, data = df)
summary(reg2)

Call:
lm(formula = tpe_diff ~ source_trust * gen2, data = df)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.1040 -0.8130  0.1038  1.1038  1.8536 

Coefficients:
                   Estimate Std. Error t value Pr(>|t|)    
(Intercept)         2.10612    0.45177   4.662 6.15e-06 ***
source_trust       -0.30943    0.09269  -3.338  0.00103 ** 
gen21              -0.79420    0.60684  -1.309  0.19232    
source_trust:gen21  0.18470    0.12690   1.455  0.14732    
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 1.126 on 177 degrees of freedom
Multiple R-squared:  0.07114,   Adjusted R-squared:  0.05539 
F-statistic: 4.518 on 3 and 177 DF,  p-value: 0.00443

# Extract the data from the interaction plot
p <- plot_model(reg2, type = "int")
# p$layers <- p$layers[sapply(p$layers, function(x) !inherits(x$geom, "GeomPoint"))]

# Recreate with your custom styling
p <- p +
  geom_jitter(data = df, aes(x = source_comp, y = tpe_diff, color = gen2),
             alpha = 0.6, inherit.aes = FALSE, show.legend = FALSE, width=.5, height=.5) +
  scale_color_manual(
    name   = "Gender",
    values = c("black", "#A90533"),
    labels = c("Men", "Women")
  ) +
  scale_fill_manual(
    name   = "Gender",
    values = c("black", "#A90533"),
    labels = c("Men", "Women")
  ) +
  guides(fill = "none") +
  labs(
    title=NULL,
    x = "Perceived Source Trustworthiness",
    y = "Third Person Effect"
  ) +
  theme_bw() +
  theme(
    text = element_text(size = 12),
    legend.position = c(0.95, 0.95),
    legend.justification = c(1, 1),
    legend.key.spacing.y = unit(0.25, "cm"),
    legend.background = element_rect(color = "black", linewidth = 0.5),
    plot.background  = element_rect(fill = "transparent", color = NA)
  )

p

# 
# ggsave("interaction_plot1.svg", plot = p, width = 10.4, height = 8.4, bg = "transparent")

reg4 <- lm(tpe_diff ~ source_comp*gen2, data = df)
summary(reg4)

Call:
lm(formula = tpe_diff ~ source_comp * gen2, data = df)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.0930 -0.7045  0.0863  1.0163  1.9005 

Coefficients:
                  Estimate Std. Error t value Pr(>|t|)    
(Intercept)         2.6785     0.4854   5.518 1.20e-07 ***
source_comp        -0.4421     0.1027  -4.303 2.78e-05 ***
gen21              -1.1074     0.6686  -1.656   0.0995 .  
source_comp:gen21   0.2628     0.1418   1.854   0.0655 .  
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

Residual standard error: 1.101 on 177 degrees of freedom
Multiple R-squared:  0.1116,    Adjusted R-squared:  0.09656 
F-statistic: 7.413 on 3 and 177 DF,  p-value: 0.000105

# Extract the data from the interaction plot
p <- plot_model(reg4, type = "int")

# Recreate with your custom styling
p <- p +
  geom_jitter(data = df, aes(x = source_comp, y = tpe_diff, color = gen2),
             alpha = 0.6, inherit.aes = FALSE, show.legend = FALSE, width=.5, height=.5) +
  scale_color_manual(
    name   = "Gender",
    values = c("black", "#A90533"),
    labels = c("Men", "Women")
  ) +
  scale_fill_manual(
    name   = "Gender",
    values = c("black", "#A90533"),
    labels = c("Men", "Women")
  ) +
  guides(fill = "none") +
  labs(
    title=NULL,
    x = "Perceived Source Competence",
    y = "Third Person Effect"
  ) +
  theme_bw() +
  theme(
    text = element_text(size = 12),
    legend.position = c(0.95, 0.95),
    legend.justification = c(1, 1),
    legend.key.spacing.y = unit(0.25, "cm"),
    legend.background = element_rect(color = "black", linewidth = 0.5),
    plot.background  = element_rect(fill = "transparent", color = NA)
  )

p

# 
# # Save as SVG
# ggsave("interaction_plot2.svg", plot = p, width = 10.4, height = 8.4, bg = "transparent")