Source: "Facebook Rudeness: What Is an Individual's Reaction to Negative Feedback?" – Florida International University
Difficulty: Intermediate | Prerequisites: Parts 1 and 2 of these study notes; understanding of factorial ANOVA (main effects and interaction effects).
Study Two builds directly on Study One by asking: does the commenter's perceived age change how rude observers think they are? The researchers expected younger users to be judged more harshly, based on stereotypes about adolescent immaturity. This study introduces a 2×2 factorial design (rudeness × age), which is a step up in complexity from Study One's single-factor ANOVA. Understanding this design, and why the age variable largely failed to produce the expected effects, is the key takeaway.
Study Two added Corey's age (young vs. old) as a second variable alongside rudeness (rude vs. polite). Rudeness still had a strong main effect on impressions, but age had almost no impact on whether participants thought Corey was rude. There was a small interaction effect on the rudeness rating and a small main effect of age on anger (younger Corey provoked slightly more anger), but overall the researchers' hypothesis that age would be a significant factor was not supported.
Factorial ANOVA (two-way ANOVA)
An extension of the one-way ANOVA that tests the effects of two or more independent variables simultaneously, including their interaction. In simple terms, it lets you ask whether rudeness and age each have their own effect and whether the combination of the two produces something different from either alone.
Main effect
The independent effect of one IV on the DV, averaged across all levels of the other IV. For example, the main effect of rudeness compares all RD participants against all PD participants, regardless of whether Corey was young or old.
Interaction effect
Occurs when the effect of one IV on the DV depends on the level of the other IV. Think of it as: "Does the impact of rudeness change depending on whether Corey is young or old?" A significant interaction means the answer is yes.
2×2 factorial design
A design with two IVs, each at two levels, producing four conditions. Here: Rudeness (RD vs. PD) × Age (Young vs. Old), yielding RDY, PDY, RDO, PDO.
Phi coefficient (φ)
An effect-size measure for 2×2 chi-square tests (the equivalent of Cramer's V for larger tables). A low Phi value (.023 in this study) indicates the two categorical variables are barely related.
Between-subjects design
An experimental design where each participant is exposed to only one condition. Both studies in this paper use this approach.
N = 581 (234 male, 347 female)
Age range: 18 to 71, mean 26.62 (SD = 9.588)
Ethnicity: 57.5% Hispanic, 24.8% African American, 14.3% Caucasian, 2.4% Asian American, 1.0% Native American
Recruited online via Qualtrics
2×2 between-subjects factorial design
IV 1: Rudeness condition (Rude Disagree vs. Polite Disagree). The Polite Agree condition from Study One was dropped.
IV 2: Corey's age (Young, under 25 vs. Old, 25 and over), manipulated through the age displayed on Corey's Facebook profile
Four conditions: RDY (Rude Disagree Young), PDY (Polite Disagree Young), RDO (Rude Disagree Old), PDO (Polite Disagree Old)
Consent was collected online (Qualtrics) rather than orally
Survey expanded to six parts (added a manipulation check question about Corey's age)
Same Facebook page stimulus, same DVs, same rudeness manipulation
χ²(1) = 0.296, p > .05, Phi = .023
There was no significant association between the age condition and participants' recall of Corey's age
This is a weak result: participants were not reliably distinguishing whether Corey was young or old, which undercuts the age manipulation
Main effect of rudeness: F(1, 577) = 109.611, p < .05. RD (M = 4.23) rated significantly ruder than PD (M = 3.01). This replicates Study One.
Main effect of age: F(1, 577) = 2.147, p = .143. Not significant. Young (M = 3.70) and Old (M = 3.49) did not differ meaningfully.
Interaction effect: F(1, 577) = 4.233, p = .040. Slightly significant. The four-condition means were: PDY = 3.18, PDO = 2.75, RDY = 4.20, RDO = 4.28. The interaction suggests that the gap between polite-young and polite-old was larger than the gap between rude-young and rude-old, but the overall pattern is modest.
Main effect of rudeness: F(1, 577) = 33.475, p < .05. RD (M = 3.29) provoked significantly more anger than PD (M = 2.59).
Main effect of age: F(1, 577) = 5.035, p = .025. Significant. Young (M = 3.05) provoked slightly more anger than Old (M = 2.76). This is the one place where age had a detectable effect.
Interaction effect: F(1, 577) = 0.457, p = .499. Not significant. The rudeness-by-age combination did not produce a unique pattern for anger.
The core finding is consistent: rudeness drives negative impressions. In both studies, participants in the RD condition rated Corey as significantly ruder and felt angrier.
Age was largely irrelevant to rudeness judgments. The researchers expected younger users to be perceived as ruder, but the data did not support this.
The one age finding (younger Corey provoked slightly more anger) may reflect a perception that rudeness from a younger person is more disrespectful, but the effect was small.
The PD condition continued to occupy an ambiguous middle ground, consistent with Study One.
The age manipulation may have been too subtle. The age was listed in Corey's profile details, and many participants did not notice or remember it.
The sample was drawn primarily from FIU students and their networks, limiting generalisability.
A more diverse sample across different age groups and cultural backgrounds could reveal effects that this sample could not detect.
Future research could make the age cue more prominent, or could examine how personalised rudeness (directed at the participant rather than a third party) changes responses.
The finding that age does not meaningfully change how observers judge online rudeness has implications for both platform design and public discourse. It suggests that "tone policing" interventions (nudging users to rephrase rude replies before posting) would be effective regardless of the user's demographic. It also challenges the assumption that older adults get a "free pass" for harsh language online, or that younger users are disproportionately blamed for incivility.
Younger people are always perceived as ruder online. This study found no significant main effect of age on rudeness perception. The stereotype did not hold up under experimental conditions.
A non-significant result means the study failed. A null finding on the age variable is informative. It tells us that age is not the driver of rudeness judgments that the researchers (and common wisdom) expected.
A significant interaction effect always means a large or important effect. The interaction effect on rudeness (p = .040) was only just significant and the pattern was modest. In context, it did not change the overall conclusion.
If the manipulation check fails, the whole study is invalid. The weak age manipulation check suggests the age cue was too subtle, not that the entire design is useless. The rudeness manipulation still worked, and the rudeness results still stand.
⚠️ Be able to explain the 2×2 factorial design and name all four conditions (RDY, PDY, RDO, PDO).
⚠️ Know the difference between a main effect and an interaction effect. You may be asked to interpret the ANOVA table for this study.
⚠️ The key contrast: rudeness had a strong main effect (F = 109.611), age did not (F = 2.147, p = .143). Be ready to explain what this means in plain language.
⚠️ Understand why the weak age manipulation check (Phi = .023) is a limitation. If participants did not register Corey's age, the age variable could not have had its intended effect.
⚠️ Be prepared to discuss the limitations and suggest improvements. A common exam question: "How would you strengthen the age manipulation?"
True or False: Study Two used a 3×2 factorial design.
False. It was a 2×2 design (Rudeness × Age).
Fill in the blank: The four conditions in Study Two were RDY, PDY, RDO, and ______.
PDO (Polite Disagree Old)
True or False: There was a significant main effect of age on "Corey seems like a rude person."
False. The main effect of age on rudeness was not significant (p = .143).
Fill in the blank: A Phi coefficient of .023 indicates a ______ effect size.
Very low (negligible)
True or False: The interaction effect on anger was significant.
False. The interaction for anger was not significant (p = .499).
Q: Describe the 2×2 factorial design used in Study Two, including both IVs and all four conditions.
A: Study Two crossed two IVs: Rudeness (Rude Disagree vs. Polite Disagree) and Age (Young, under 25 vs. Old, 25 and over). This produced four conditions: Rude Disagree Young (RDY), Polite Disagree Young (PDY), Rude Disagree Old (RDO), and Polite Disagree Old (PDO). Each participant was randomly assigned to one condition.
Q: What did the main effect of rudeness tell us in Study Two, and how does it compare to Study One?
A: The main effect of rudeness was significant for both DVs. For rudeness perception, F(1, 577) = 109.611, p < .05, with RD participants (M = 4.23) rating Corey as much ruder than PD participants (M = 3.01). This replicates the core finding from Study One: how you respond matters far more than other factors.
Q: Why did the researchers' hypothesis about age fail to find support?
A: The researchers expected younger Corey to be judged as ruder, based on stereotypes about adolescent immaturity. However, the main effect of age on rudeness ratings was not significant (p = .143). A likely contributing factor is that the age manipulation was too subtle (Corey's age was listed in profile details but not emphasised), as confirmed by the weak manipulation check (Phi = .023). Participants may not have noticed or remembered Corey's age.
Q: Explain the difference between the main effect of age on rudeness perception and the main effect of age on anger. Why might these differ?
A: Age had no significant main effect on rudeness perception (p = .143) but did have a small significant main effect on anger (p = .025), with younger Corey provoking slightly more anger. This suggests that while participants did not think a younger person's response was inherently ruder, they may have felt more emotionally provoked by it, possibly perceiving rudeness from a younger person as more disrespectful or socially inappropriate.
Q: Name two limitations of this research and suggest one improvement for each.
A: (1) The age manipulation was too subtle, with many participants not noticing Corey's listed age. Improvement: make the age more salient, for example by having Corey reference their own age in a post or by using a profile photo that clearly suggests a particular age group. (2) The sample was mostly FIU-affiliated, limiting diversity. Improvement: recruit from multiple universities and geographic regions, with deliberate oversampling of older adults to better test age-related hypotheses.
Study Two connects to age stereotyping in social psychology, specifically whether expectations about generational behaviour ("kids these days") translate into measurable perception differences. It also ties to research on ecological validity: the weakness of the age manipulation raises questions about how well laboratory-style online experiments capture real-world social media dynamics. The factorial ANOVA design is a staple of PSY 3215, and this study provides a concrete example of interpreting main effects, a null main effect, and a marginally significant interaction.
Factorial ANOVA, 2×2 design, main effect, interaction effect, Phi coefficient, age stereotyping social media, rudeness perception online, Facebook rudeness age, manipulation check failure, between-subjects factorial, online incivility age, adolescent vs. adult social media behaviour, Study Two PSY 3215, FIU psychology, Qualtrics online experiment