Violence and graphic sexual content in American movies and television shows have long been a concern.

Now, researchers at the University of Pennsylvania plan to study positive behaviors, such as exercise, healthy eating, wearing a seat belt, showing concern for others, and respecting differences.

Leading the effort is Patrick E. Jamieson, who directs the Annenberg Public Policy Center's Adolescent Health and Risk Communication Institute.

The new study builds on a five-year analysis of more than 900 movies and 1,600 hours of television. Among the numerous findings from that work:

In movies, depictions of suicide have increased steadily since 1950. Youth suicides rose from the 1960s to the 1990s.

The rate of gun violence in popular PG-13 movies has tripled since 1985.

Since 1950, male characters in movies have consistently outnumbered female characters two to one.

The group now has a three-year, $745,000 grant from the Robert Wood Johnson Foundation to identify and examine trends that reflect a "culture of health." We recently spoke to Jamieson about it.

Why look at healthy behaviors?

I want to understand better what TV is actually portraying, and I think we'll get closer to doing that at the end of this project.

The new project - dubbed CHAMPION, for Culture of Health and Media Portrayal in Our Nation - is built out of its predecessor, a very large academic content analysis of risk portrayal, including alcohol, tobacco, drugs, sex, suicide, and violence. It covered the 1950s to the present.

But if you only look at the negative risk portrayals, what is it you're actually capturing? There are positive portrayals on TV, and we know that matters, too. So we will be looking at things like people helping each other, antidrug messaging, what people are eating, positive parent-child interactions.

We train undergraduates to code the programming - teaching them definitions - so we can make sure they see things the same way. We are very rigorous about our reliability testing.

Some of the questions they code are dichotomous . . .. Was the content present or was it not? Others are measured on a five-point scale. For instance, a verbal reference to alcohol would be a "one." Implied consumption would be a "two." Actual consumption would be a "three" . . ..

The amount of coding is massive. For this study, we'll code 3,000 hours of TV programming, but no movies. One of the reasons is that, depending on age group, exposure to TV can be 20 times greater than that for movies. Movies tend to have racier content than TV, but TV seems to be catching up.

The policy center also will analyze shows of particular interest to African American and Hispanic viewers, right?

America is becoming more diverse, so now is a really good time to study cross-ethnic and cross-racial comparisons. African American and Hispanic audiences watch more TV than the U.S. adult average, according to 2015 Nielsen data. We'll be coding Univision - the American Spanish-language network ranked No. 1 in several sweeps. We have coders who are fluent in Spanish, so we'll also be looking at Spanish-language telenovelas, or soap operas, and narconovelas, which are crime dramas.

There's increasing content as well that's popular with African American audiences, such as Blackish and Empire, the show about an African American family in the music business that was such a hit in the 2015 season, and we'll study them.

Any preliminary expectations?

The first thing we'll have is trend lines. For instance, in our previous work, is violence in movies going up or down? We showed it's going way, way up. What gets more fascinating - and complex - is when you start combining the measures. Do movies that feature violence also feature alcohol and tobacco and sex? And does a trend in TV correspond to a similar trend in the real world?

As for the healthy behaviors, I don't really know what to expect, although the healthy behaviors are probably fewer than the negative behaviors. And I know that capturing the positive behaviors is much harder than capturing the negative behaviors. If you see someone smoking a cigarette, that's easy to code. It's harder to code whether a person is intending to help someone.

One of the fascinating parts of the new research will be to create a TV positive vs. negative health risk ratio, and to see if that trend line is going up or down.

Since movies and TV are so influential, should script writers be doing something different?

One of the biggest problems I see is that almost all writers in Hollywood are male. I think there needs to be more female screenwriters. More female directors.

Criticism of the current Oscar nominations - the overwhelming number of whites - reminds me of when Denzel Washington spoke out that African American actors also should be winning Oscars, and the next year, 2002, he won for Training Day. Hollywood needs to balance its male-female, racial and ethnic representations and better reflect America's diversity.

What I'd really like to have happen is for suicide to be never modeled in TV or on film. I'd like it to be oblique - a suggestion, but not actually shown. The scientific literature on media suicide "contagion" is quite compelling. When suicides are portrayed in the media, the suicide rate in the real world goes up.

Given your research, what about your own family's TV watching?

After my wife and I had kids, I watched less TV. Now, I limit my children's TV-watching a lot. I have them watch documentaries; I have them listen to music; they read books. But I don't know how long I can keep that up because they are getting older and more independent.