Speech does not cause harm or violence

From The Ecology of Schools: Fostering a Culture of Human Flourishing & Developing Character by Christopher J. Ferguson, Ph.D.

It has been a longstanding article of faith among enemies of freedom and, in particular, freedom of speech, that speech can cause actual measurable harm. This is manifested in speech codes, in legislation against hate speech, in ratings systems, in educational circles where books are deemed dangerous because of their content, concern about micro-aggressions, etc. It is a pernicious evergreen that blossoms in authoritarian soil.

It is an interesting empirical question. Do people who read violent books consequently behave more violently? Do people who hear facts or opinions with which they disagree suffer measurable harm? Does watching a violent movie make you more violent? There is also the corresponding claim - does watching pacific, pastoral or pablum make you more amiable and inoffensive?

Ideas have always frightened those vested in an establishment. They are especially frightening to those in authoritarian states where the claimed reality is distant from obvious reality. Hence the restrictions on speech in authoritarian states (and among those of an authoritarian stripe such as college administrators and postmodernist advocates.)

Since at least the nineteen fifties there has been efforts to mount credible research to answer those questions. Efforts, but not so much results. You have to take into account the direction of causal flow (what happens if violent content has no causal effect but violent people like watching violent content?), predispositions, define verbal or visual violence, define what constitutes measurable effect, etc.

While there have been many efforts, the results have been muddied. Advocates for controlling speech tend to conduct flawed studies which support their positions, and unbiased researchers have a hard time measuring hard-to-measure phenomenon and to control the confounding variables. As the efforts have continued and as study controls become more robust, it appears that, as suspected by many, there is no systemically and reliably measurable association between violence, seen in movies or heard in songs or read in books, and actually manifested violent behavior. From the above summary of the research:

Several things do appear to be clear from meta-analyses:

1. Effects range somewhere between very small to nonexistent.

2. Study results are influenced by systematic methodological flaws, such as used of poorly validated aggression measures, researcher expectancy effects and publication bias. These flaws tend to result in outcomes that are biased toward exaggerating effects.

3. The closer outcome measures approximate actual violence, the closer effects come to zero.

4. Controlling for theoretically relevant third variables (e.g. gender, personality, family environment, genetics) brings effect sizes close to zero. This suggests that violent media has little unique risk for negative outcomes.

5. Research evidence for effects has been getting weaker, not stronger, particularly over the prior 10 years. This is likely due to increased rigor in the field.