Exposure to Video Games

Question 1

What is aggression?

Answer 1

Wide variety of hostile behaviours motivated by fear or frustrated: physical, verbal, relational

violence - subtype of aggression
involves greater intensity and destruction, manifests in attempt to cause harm

violence can be aggressive, but aggression is not always violent

Question 2

What is the difference between short term and long term effects?

Answer 2

Short term effects: brief exposure to VVG with immediate assessment of outcomes

Long-term effects: accrue from repeated exposures over extended period of time - measured distal from exposure

Question 3

What are the types of methods used?

Answer 3

Correlational designs: is amount of time spent playing VVG associated with scores on a measure of aggressive thoughts?

Longitudinal designs: does amount of VVG use in a sample of children at age 7 predict their aggressive behaviour at ages 16 and 21?

Experimental designs: do people randomly allocated to play a VVG compared to non-VVG report more aggressive cognitions and behaviours such as willingness to administer aversive stimuli

Epidemiological data: relationship between VVG and violent crime rates

Question 4

What does epidemiological data hypothesise?

Answer 4

Positive relationship between VVGs and trends in violent crime rates

Question 5

Does data support the epidemiological data hypothesis?

Answer 5

Negative relationship between VVG sales and violence rate

Significant decrease in crime rates with increasing game sales

Negative relationship between youth violence rates and video games sales between 1996 and 2011

Question 6

What happened in 2018 in Australia?

Answer 6

Introduction of legislation that allowed for release of R18+ games - burglaries continued to decrease; homicides remained stable, sexual assault

but, important to remember that VVG exposure is not going to be only or main risk factor for societal violence

Question 7

What did Anderson et al, 2010 look at?

Answer 7

Outcomes:

• Aggressive behavior
• Aggressive thinking;
• Aggressive mood
• Physiological arousal
• Empathy/desensitization to violence
• Prosocial behavior

Rated methodological quality of papers to create “best practice” subset (208 effect sizes) - took each paper and rated it

Controlled for sex, and Time 1 aggressive behavior in longitudinal designs (“best partials” dataset”) and looked at moderating effects of research design and culture

Looked at experimental, longitudinal and cross-sectional

Also looked at effect size in overall sample regardless of methodological quality

Question 8

What did Anderson et al, 2010 find?

Answer 8

Significant effects of VVG on aggressive behaviour regardless of study design

Effect not moderated by culture, sex and most methodological variables

Weak effect of age – larger effects in younger participants

Effect significant even when methodologically weaker studies included

No evidence that publication bias (using trim and fill approach, only reporting when there had been an effect) had an influence on results

Question 9

What did Hilgard et al, 2017 look at?

Answer 9

Reanalysis of Anderson et al
Had concerns about: Publication bias: significant studies more likely to be published
P-Hacking: questionable research practices (e.g. data stopping when p < .05, outlier management, outcome switching) increases risk of false positives
Selection bias: flexibility/bias in meta-analytic exclusion criteria, “best practice” more often applied to significant studies

investigated effects in unpublished dissertations
used more advanced methods for detecting and correct publication bias

Question 10

What did Hilgard et al, 2017 find?

Answer 10

Funnel plots showed asymmetry indicative of bias and overestimated effect sizes

Anderson reported effect size of .21 but when did adjustment, had to reduce it by .03

Evidence of significant bias in estimates from experimental studies of aggressive behaviour - when Hilgard looked it, they reduced it a lot more

Multiple methods suggest that underlying effects so small to be undetectable in typical sample sizes

Unpublished work showed smaller, non-significant effects (r = .01) and less likely to be classed as “best-practice”

Question 11

What did Ferguson look at?

Answer 11

Outcomes:
aggressive
prosocial behaviour
academic performance
depressive symptoms
attention deficit symptoms
Rated methodological quality of papers to create best practise subset
controlled for sex, time 1 aggressive behaviour in longitudinal designs, personality traits, family environment
Looked at moderators, use of standardises measures, ethnicity, gender, publication status

Question 12

What did Ferguson find?

Answer 12

VVG have minimal negative effect on children’s wellbeing - accounting for less than 1% of variance in aggressive behaviour (0.09)

Effect not moderated by study design, age of the child, length of longitudinal period

Weaker effects in studies with better methodologies and in unpublished research

Some evidence of publication bias in published studies (using Tandem procedure)

Question 13

What other risk factors are implicated?

Answer 13

Ferguson et al, 2018

Explored correlation between exposure to family violence, personality, trait aggression, video game playing and violent criminal behaviour in undergrads
Violent criminal behaviour significantly associated with:
- Trait aggression (r = .33)
- Physical abuse (r = .27)
- Spanking (r = .19)
- Verbal abuse (r = .18)
- Male gender (r = -.19)
- But not exposure to violent video games (r = 0.11)  not significantly correlated

Question 14

What did the regression analysis show about risk factors? Ferguson et al

Answer 14

Trait aggression and exposure to physical abuse strongest predictors of violent criminal behaviour
Violent not significant alone, but had an impact on trait aggression
Direct exposure to VVG did not predict violent criminal behaviour
Significant interaction between trait aggression and VVG exposure on violent crime
Exposure to VVG had little effect on crime behavior, except in players in top quartile for trait aggression - group of people prone to violent behaviour and enjoy playing violent games
Not games, exposure to physical abuse as a child and trait aggression
Consistent with a Catalyst Model
Subtype of aggressive individuals who are prone to violent behavior and enjoy VVG

Question 15

What are the 8 methodological limitations?

Answer 15

Diversity in video games
Mismatched control conditions in experiments
No pre-test measures of aggression
Overreliance on bivariate correlations
Selective interpretation and citation bias
Unstandardised aggression measures
Moderators
Lack of clinical/real world validity

Question 16

What is the problem and solution with the diversity in video games?

Answer 16

Huge variability within violent video games, differ in the dimensions (realistic vs fantasy, difficulty and pace of action, target of violence)

Solution:
routinely code these factors and conduct moderator analyses in meta-analyses
manipulate in future studies to determine impact

Question 17

What did Hull et al, 2014 look at in terms of diversity in video games?

Answer 17

Looked if type of protagonist impact aggression
Tracked use of mature-rated risk glorifying VVG and deviant risky behaviours in 5000 teenagers over 4 years
Greater game play associated with increases in sensation seeking and rebelliousness
Predicted change in alcohol, smoking, delinquency

Question 18

What did Hull et al, 2014 find in terms of diversity in video games?

Answer 18

Effects stronger in youth who played VVG with an antisocial protagonist (e.g. GTA) compared than heroic protagonist (e.g. Spider Man)

If character looks like them, they aggresse more  they can more strongly identify with characters who look like them

Question 19

What is the problem with mismatched control conditions in experimental studies?

Answer 19

Violent and non-violent games often not matched for:
◦	Competitiveness, 
◦	Difficulty level
◦	Frustration
◦	Pace of action

Solution:
Use modified game paradigms where researchers manipulate violent content of a single video game by changing underlying code to reduce effect of confounding variables
Change code to modify the content

Question 20

What does Przybylskilet et al believe causes aggression and violence?

Answer 20

Aggression and violence result from threatened or actual thwarting of basic psychological needs:
need for competence
need for autonomy
need for relatedness

Question 21

What did Przybylskilet et al look at?

Answer 21

Hypothesised that games that undermined players’ feelings of competence would lead to increases in aggressive thoughts, feelings and behaviour
Manipulated complexity of game interface, complexity of controls, opportunity to build experience, degree of challenge and violent/non-violent content of games
Self-report, cognitive and behavioural indicators of aggression

Question 22

What did Przybylskilet et al find?

Answer 22

Gaming that undermined need for competence associated with short term shifts in:
aggressive feelings
accessibility of aggressive thoughts
aggressive behaviours
• Importantly, effects were independent of the degree of violence in the games
• Suggests that how challenging or frustrating a game is, rather than its violent content might be more predictive of aggression

Question 23

What is the problem with no pre test measures of aggression?

Answer 23

Unclear if higher post-test aggression scores in VVG condition reflects an increase or a differential reduction
Some evidence that all video games reduce hostility from pre-post-test

Solution: 
Measure aggression (other outcomes) pre- and post- VVG exposure

Question 24

What is the problem with over reliance on bivariate correlations?

Answer 24

Some studies only report raw correlation between VVG exposure and aggression
Fail to control for third variables that may influence both VVG use and aggression levels e.g.
◦ gender
◦ trait aggression
◦ family environment
Might see a significant effect, but without controlling for other variables

Solution:

• Measure key confounds routinely
• Present controlled analyses as standard
• Pre-registered analyses

Question 25

What is the problem with selective interpretation and citation bias?

Answer 25

Focus on results (e.g. raw bivariate correlations) that confirm association despite better controlled analyses not showing an effect
Null findings not cited leading to distorted perception of research field  researchers want to find an effect
File drawer effect – null findings not published

Solution:

• Cite both positive and null findings
• Present rigorous, robust, pre-registered analyses
• Include grey literature/unpublished work in meta-analyses where possible

Question 26

What is the problem with unstandardised aggression measures?

Answer 26

Multiple ways to score the same measure(s)

Researchers may pick the approach that gives the “best” (i.e. significant) results

Solution:

• Community agreed approaches for scoring common tasks
• Preregister outcome measures, scoring approach and approach to data cleaning

Question 27

What is the competitive reaction time task?

Answer 27

Reaction time game against an “opponent” (Taylor, 1967)
Winner of each round (25 trials) punishes the loser by administering a loud noise blast
Participants set volume and duration of noise blast
- louder and longer blasts indicator of higher levels of aggressiveness
Lack of standardization of test procedure and data analysis approaches (Elson et al., 2014) - no agreed way to score measure of aggression

At least 140 ways to score CRTT:

• Sum of volume and duration setting in 1st trial
• Mean volume and duration scores of winning and losing trials
• Volume multiplied by log transformed duration on each trial then averaged across 25 trials
• Mean volume x mean duration
• Standardized volume + standardized duration

Question 28

What did Elson et al., (2014) find when they reanalyzed CRTT data collected after playing VVG vs. non-VVG?

Answer 28

Calculated CRTT aggression scores using 13 different approaches and tested for significant difference in scores between VVG and non-VVG groups

Results showed different patterns of results, depending on how you’d coded your outcome measure

Depending on measure, possible to make it look like violent games increased, decreased or had no effect at all

Question 29

What is the problem with moderators?

Answer 29

Unclear if effect stronger in particular ’at-risk’ subgroups e.g.
children with pre-existing mental health problems
high levels of trait aggression
not many studies have looked at high risk populations

Solution:
Research studies needed that include high risk sub-samples that allow for moderation analyses

Question 30

What happens to at risk populations?

Answer 30

In young adults, personality traits such as anger, aggressiveness and psychoticism may moderate association between VVG and aggression - raising the risk

Adults and children with elevated mental health symptoms/disorders are not more susceptible to VVG effects:
Adults with ASD not differentially affected by acute exposure to VVG
Children with pre-existing antisocial traits, clinically elevated depression and attention deficit symptoms not adversely influenced by VVG
Children with higher trait aggression did not show stronger effects of VVG on aggression

Question 31

What is the problem with lack of clinical/real world validity?

Answer 31

Lab aggression measures (e.g. noise bursts, hot sauce) over-generalised to real-world violence (e.g. mass shootings)
Extrapolate from brief lab-based exposures and outcomes to long-term real-world effects
Small effect sizes – explaining 0.4-3.2% of variation in minor forms of lab-based aggression but people lead to the assumption that it will lead to real world violence

Solution:

• Avoid overgeneralising and improve communication around size of effect and methodological limitations
• Longitudinal designs with repeated experimental or chronic exposure

Question 32

What have improvements in methodology shown? (study design) Kuhn et al

Answer 32

What are the effects of frequent, habitual VVG play? How long do these effects persist?
Longitudinal RCT comparing VVG group vs. non-VVG group vs. passive control
Played game every day for 2-month period (>30h exposure to game)
Assessments at pre-, post- and 2mth follow-up
Self-report and behavioural aggression measures
Impulsivity and risk taking
Pro-social behaviour and empathy
Anxiety and depression symptoms
Executive control

Question 33

What have improvements in methodology shown? (results) Kuhn et al

Answer 33

Minimal evidence that participants assigned to VVG condition showed significant emotional, cognitive or behavioural changes:

• Compared to non VVG condition
• Compared to passive control group
• Between pre and post-test (i.e. proximal effects)
• Between pre- and follow-up (i.e. long-lasting or evolving effects)

No detrimental effects of extended VVG play in a study design that moves beyond short-term priming effects of acute exposure
• Similar study designs needed with child samples

Question 34

Can playing video games have a positive effect?

Answer 34

Growing recognition of benefits of video games (including VVG) for children’s development  cognitive, motivational, social and emotional

Question 35

What are the cognitive benefits of playing video games?

Answer 35

Enhanced visuo-spatial cognition
Playing shooter games (vs. control condition) leads to:
- More accurate, efficient attention allocation
- Improved spatial and mental rotation skills
- Problem solving skills central to all genres of video games as have to solve complex challenges
- Enhanced creativity and openness to experience
- Mechanisms unclear: quick decision-making in response to unpredictable changes, hand-eye coordination, navigate visually rich 3D spaces, largely correlational studies

Question 36

What are the motivational benefits?

Answer 36

Help children learn the value of persistence in the face of failure and that they will eventually get rewards

Provide intermittent reinforcement to encourage players not to give up despite increasing levels of challenge

Help players learn that intelligence/skill is incremental and can increase with time and effort

Effects not specific to VVG and unclear whether benefits carry over into the real world

Question 37

What are the social benefits?

Answer 37

Multiplayer online (shooter) games can reduce isolation, build community and provide opportunities to discuss sensitive information

Promote prosocial behavior:

• Players who play VVG that encourage cooperation more likely to show helpful behaviours than those who play non-VVG
• Playing VVG in groups reduces hostility compared to playing alone
• Playing a VVG cooperatively compared to competitively increases prosocial behavior outside of the game and reduces effects of outgroup membership

Question 38

What are the emotional benefits?

Answer 38

Effective way of generating positive emotions and letting off steam, managing negative emotions

Emotion regulation:

• Have to quickly deal with full range of emotions incl. anxiety and frustration otherwise play impeded
• Playing different characters promotes ability to flexibly reappraise and respond adaptively to changing emotional experiences
• Moderate game play (compared to non and excessive) associated with better mental health and life satisfaction