[From Rick Marken (2008.12.07.1030)]
Goldstein (2008.12.07.1125 EDT)
Hi David. I take it this is Paul’s reply to Bill. Mind if I take a crack at it?
It is of course true to state that not all people who consume violent
media become violent. It also is true to state that not all people who
smoke cigarettes develop lung cancer, or that not all people exposed to
lead as children develop intellectual deficiencies. Yet I would argue that
the majority of the public accepts the established findings from public
health research linking smoking and lung cancer, and linking lead exposure
and intellectual problems. What the public fails often to accept -
unfortunately, given that the science often has been of higher quality -
is the link between violent media exposure and aggression. Particularly
vexing here is the fact that the effect size estimates for the violent
media/aggression link are higher than the lead-IQ link, and almost the
same as the smoking-lung cancer link (see Bushman & Anderson, 2001,
These are group level findings and they are fine for making policy level decisions. I think the PCT complaint about such studies would be that they don’t help you understand behavior (like aggression) at the individual level. This research is fine as sociology or policy research; but I think it’s worthless as an approach to understanding individual human behavior.
That being said, I should go further and note that all media violence
researchers acknowledge the obvious fact that before there were violent
media (at least in their modern manifestations), there still were violent
individuals. The history of the world is pretty much a history of
violence. But that doesn’t dampen the impact violent media have today.
Sure, at the group level there may be some detectable “effect” of violent media on the incidence of aggression. But then someone (policy makers, government leaders) have to decide what to do about it. There are trades. Do we keep violent media – stuff like the Old Testment, The Illiad, Grimm’s Fairy Tales, just to mention some of the most grisly – away from children in the hopes of reducing the incidence of aggression a bit or do we allow access to these media and try to reduce aggression using other means – like by reducing poverty, another variable that is known to have a group level statistical association with aggression.
research has shown – clearly, and within the strictures of longitudinal
design, conclusively as well – the following:
- Childhood violent TV exposure leads to increases in aggressive
behavior, even while controlling childhood aggressiven or
violent media, even when controlling childhood violent media exposure
(findings from Eron et al., 1972, American Psychologist; Huesmann et al.,
2003, Developmental Psychology).
What are the r squared values? If they’re not close to .99 these group level results are not useful at the individual level.
As to the processes, cognitive or otherwise, that have been theorized to
account for these links, the research centers on social-cognitive
information processing mechanisms and structures (see Huesmann, 1988,
1998). Violent media “writes the script” for violent behavior, through
cognitive scripts and schemas that become internalized and over time
direct habitually aggressive responding.
This is an individual level theory. And it’s an open loop causal theory, which is consistent with the general linear model of statistics that is used to test it. If the results of these statistical tests don’t produce consistently high r squared (goodness of fit) values then the model should be rejected. We have a better alternative: closed loop control theory.
Finally – your allusion to “simple minded correlational studies” is a
drastic underselling of the violent media research. The studies conducted
by the top investigative teams in this area are well-controlled,
well-designed, and thoughtfully implemented. The study we most recently
published (Boxer, Huesmann, Bushman, et al., online-2008 and in press for
2009, J of Youth and Adolescence) was correlational given the target
sample (incarcerated adolescents, difficult to sample prospectively) but
included statistical controls meant to reduce the threats to external
validity by this design feature. For example, we used multi-source
confirmatory factor modeling to estimate violent and generally aggressive
behavior, and used a cumulative risk method to estimate exposure risk.
It’s still correlational. And it’s “simple-minded” only in the sense that it’s an open-loop model which doesn’t take the obvious feedback connections involved into account. For example, what one views has an effect on what one does while, at the same time, what one does has an effect what one views. We live in a closed-loop of cause and effect. PCT explains why behavior in this closed loop can appear to be caused.Your father-in-law can explain it to you, if he can ever tear himself away from playing with those adorable grandchildren;-)
If you take even a cursory look at the broad narrative and meta-analytic
reviews of the media violence literature (e.g., Anderson et al., 2003,
Psychological Science in the Public Interest; Bushman & Huesmann, Archives
of Pediatrics), you will see that the majority of studies upon which the
general conclusion of violent media effects on aggression rests are very
well designed experimental and longitudinal investigations. And the
quality is most certainly going up, as evidenced by recent work by Polman
and colleagues (2008, Aggressive Behavior, naturalistic post-exposure
sampling of behavior), and Bushman and colleagues (Konijn et al., 2008,
Developmental Psychology, more ecologically valid exp design).
Again, these are probably fine as the basis for policy decisions regarding what should be broadcast during prime time. But they don’t help us understand human behavior very well. Indeed, when these group results are used as a basis for discussing individual behavior they lead to rather prejudicial conclusions, like that exposure to violence causes aggression. It doesn’t. In fact, external events never cause behavior in a closed-loop system. PCT shows that external events are just potential disturbances to controlled inputs. If a person is controlling a variable that is disturbed by the external event, they will react (to prevent the disturbance from pushing the variable from its reference) and it will appear that the external event has caused the reaction. But that’s not what is actually going on. In the case of aggression, a violent image will appear to cause aggression in people who are controlling for things like imitating an adult model. People who are controlling for other things, like treating people with respect, will react to aggressive images in a very different way, if at all.
To assert that the media violence research is “simple minded” is wrong and
degrading, and to summarize all the work as “correlational” is factually
“Simple minded” was probably a poor choice of words. I can see that it might sound degrading but I it’s certainly not wrong. And most of this stuff is correlational (statistical controls do not an experiment make) but the real problem is that , if the research is aimed at understanding individuals, it is based on the wrong model – the open loop causal model of statistics. The fact that this is the case is proved by the fact that the fit of the model to the data (measured by r squared and equivalent measures of the proportion of variance in the data accounted for by the model) is typically about .3 and rarely greater than .5. I thinkit’s about time that the basic model underlying research on media violence – indeed, the basic model underlying all research in psychology, which is the general linear model of statistics – is what is wrong and should be rejected. It’s time to try PCT.
Richard S. Marken PhD