Visualize This: Sexual health data from national survey

It’s been a while since I ran one of these, so you must be dying to do some visualizing. For those new around here, Visualize This is a little fun practice we like to run around here to exercise our visualization skills. I post a small dataset, and then you can try visualizing it. Do you have what it takes?

Deadline: October 27, 2010

For this round, we’ll take a look at results from the National Survey of Sexual Health and Behavior conducted by Indiana University. They asked over 5,000 participants if they’ve engaged in certain behaviors in the past year. They also asked age and gender.

Here are the results of that poll (click to enlarge):

See comments below for the table as an Excel spreadsheet.

The table itself is pretty interesting. I’m sure the first thing you did was to to your age group and gender and scan the column. How does your group relate to the others? Can you represent the data visually? Leave links to your graphics in the comments below. Or if you don’t have time to make something, just tell us what you’d do. Finally, you’re also welcome to email me your attempt if you don’t have anywhere to upload and link to.

One rule: let’s keep it tasteful. I probably don’t need to say this, but given the topic, I probably should. Have fun!

Deadline: October 27, 2010

56 Comments

  • Do you have the data in, you know, a useable *data* format, or do we need transcribe it from the image?

    • I purposely left that to you guys, because transfer and formatting is part of the process. But Andy has taken care of that extra step, so there you go :).

  • Sam – see my comment above yours!

  • I’ve created a dashboard using this dataset – click below to see my attempt, and click around to compare activity/age group:
    http://www.thedatastudio.co.uk/blog/the-data-studio-blog/andy-cotgreave/us-sexual-health-and-behaviour-survey

  • it’s funny…i’ve never in my life – NOT ONCE – been asked to be a part of any of the hundreds of surveys there are. “statistics say…” or “studies have shown…”

    never. i’ve been alive and on this planet for a solid 26 years, and if i haven’t been asked one question EVER in my life about anything there is a statistic for…

    • I know someone who participated in a very long (decades) longitudinal survey. He initially agreed to do it for the substantial compensation and they did ask him very personal questions (sexual activity, drug use etc.).

  • @chris So I take it you didn’t fill in your census form?

    • did they ask you how many times you had done anal sex on the census form? haha, you know what i mean – i’ve never been polled thru my early 20’s on how frequently i drink, whether or not i had tried certain drugs, whether or not i had done this or that. the ‘hey, here’s an interesting fact…’ type statistic.

  • I have reshaped the data to allow better pivoting and posted it on Google Docs. There are four columns:

    Sexual Behavior
    Respondent
    Age
    % Performing Act

    http://bit.ly/csTNuj

  • The percentage data lead me to think about blue & pink pie charts…
    maybe along an age groups x-axis
    or possibly a silhouette of the behavior filled partially (%) w/ color.

  • OK, this was more just for fun than a serious visualization attempt. And it’s more of an infographic anyway. But: http://www.megantaylor.org/projects/sexinamerica/index.html

  • Here is another Tableau Public visualization:
    http://goo.gl/wDDq

    An example of an interesting viewpoint this approach allows for is when only “Received Oral” and “Gave Oral”: are checked. You can also see that the line for Men “Masturbating Alone” is slightly different from the other lines. Additionally, it looks like the percentages peak for Men in the 25-39 age range, while 20-29 for Women.

    • Simple and well done Joe. The age range peaks make sense to me. I’m sure you’ve heard that men prefer younger women. Here is one article that says:

      “It turns out that older men chasing younger women contributes to human longevity and the survival of the species, according to new findings by researchers at Stanford and the University of California-Santa Barbara.”

      http://scienceblog.com/cms/old-men-chasing-young-women-good-thing-14203.html

    • Good stuff. My only trouble with the time series approach is that the age brackets aren’t equal.

      • Thank you for the input. Going off of your comment and the other vizes I have seen since, here is my improved version 2: http://goo.gl/4NYC

        This allows the user to change the line coloring to the viewpoint they are most interested in, and then filter to behavior.

        I also changed the names of the Oral behaviors so they can be grouped and compared more effectively.

  • I’ve made two versions of a surface-map type of visualization. The first version combines both gender data into one graphic whereas the second separates them into two panels. The combined version seems to speak more in terms of “no-man’s-islands”(in blue) or “pathways” (in red) of activities for each gender (approaching a “cloud” of nirvana) though the second paneled version is neater.

    http://web.me.com/leonardodioko/FD_challenge/One_graph.html
    http://web.me.com/leonardodioko/FD_challenge/Two_graphs.html

  • I finally figured out a way to share mine… I wanted to see
    1. how sexuality changed with age, and
    2. whether activities differed by sex.

    There was a strong change in sexuality with age regardless of activity. There were some differences among activities across age groups. However, the only major difference between sexes was for masturbation alone.

    I also initially plotted average sexual activity separately by sex to examine my hypothesis that older women reach their sexual peak at a later age (in contrast to Joe’s hypothesis), but this was not the case. I couldn’t justify saying that there was any general difference between sexes.

    http://i1228.photobucket.com/albums/ee452/berdaniera/Sexualhealth-1.png

    Some notes on my methods:
    – I summed oral sex across the gender of the giver/receiver, and then averaged across giving/receiving (no big differences within sex/age category), which gave me a combined “gave or received oral sex.”
    – Standard errors on the activities come from using the age categories as replicates.

  • Here is my attempt: http://imgur.com/PMYpl.png

    I have classified each type of sexual behavior further based on (a) act, (b) role and (c) type. This in my opinion allows us to get a better visualization of the data. Based on the graph attached in the link, here are some of the interesting observations:

    1. Sexual Activity Peaks around 28 for Women and 30 for Men.
    2. Straight Men > 30 2. for age > 30, men overestimate the extent of
    a. oral given to women
    b. oral received from women
    c. vaginal intercourse
    d. anal sex
    This is evident from the fact that the blue line stays above the red line always

    There are several other interesting conclusions that one could make based on this visualization. I would appreciate comments to make this graphic better. I am in the process of creating a github repository where I would post the code and dataset.

  • Here are two that I received via email.

    This one is from Aaron:

    https://flowingdata.com/wp-content/uploads/2010/10/Screen-shot-2010-10-22-at-1.15.02-PM.png

    He notes:

    “There was a strong change in sexuality with age regardless of activity. There were some differences among activities across age groups. However, the only major difference between sexes was for masturbation alone.

    I also initially plotted average sexual activity separately by sex to examine my hypothesis that older women reach their sexual peak at a later age, but this was not the case. The cougar effect is not seen in this data.

    Some notes on my methods:
    – I summed oral sex across the gender of the giver/receiver, and then averaged across giving/receiving (no big differences within sex/age category), which gave me a combined “gave or received oral sex.”
    – Standard errors on the activities come from using the age categories as replicates.”

    And this one is from Mike D. He went with a paneled time series:

    https://flowingdata.com/wp-content/uploads/2010/10/Screen-shot-2010-10-22-at-1.15.32-PM.png

  • I have applied Principal Component Analysis and have used biplot as indicator to create a map, which shows distribution of various sexual activities and showing separation of men and women by age on the map.
    The 1st PC (horizontal axis) separates data by gender-specific activities: in the middle there are activities common to both genders, to the left – men-specific and to the right – women-specific activities.
    The 2nd PC (vertical axis) separates by degree of “advancement” in type of activity, masturbation being the most conservative one common to very youngest and oldest men and women, and anal sex being the most rare mostly common for the most sexually active age group of both men and women.

  • Didn’t quite get the chance to do as much as I wanted with this, since I was learning Protovis as I went, but here’s what I’ve got: http://www.mjs-svc.com/rand-bin/sex/

    • This one is really nice – I think you’ve done some lovely stuff with Protoviz. How would you rate learning how to use it? Is it easier/harder than some of the other equivalents (Google Charts, etc)?

      • Much harder. The biggest learning block, however, was the state of the documentation. Although class documentation is in place, the main method for learning how to actually construct a visualization seems to be deconstructing existing visualizations, as very little was written in the docs. That’s why I ran out of time! Thanks for the comment, though!

  • Here is my viz: http://bit.ly/a7OEp6

    Note that I concatenated Sex & Behavior to determine which acts are the most frequent and by whom.

    Some points this viz shows quickly:
    1. Men masturbate their whole lives.
    2. The age range from 20-39 are clearly the peak sexual activity periods, with 25-29 standing out the most.
    3. There is very little women on women and men on men oral sex. So much for that woman on woman stuff we hear about.
    4. We run out of steam in our 70s.

  • Mark Bulling October 27, 2010 at 3:31 pm

    here’s my entry: http://www.flickr.com/photos/everheardofaspacebar/5121892116/lightbox/

    on top of the above comments, comparing men and women there seem to be some areas of over/under claim or asymmetric promiscuity.

    • Looking back at old posts, missed this when it came out. Mark, I definitely think you did one of the best jobs with this data. Most had a hard time categorizing the 9 categories of sex acts in a way that easily compared the acts between each other to discern interesting trends, but yours does a great job of this!

  • I uploaded my attempt to Google Documents.
    https://docs.google.com/leaf?id=0BwfS_L6-lf1KNGNhNWUwMGEtZWIwMi00ZmVjLWIwN2YtNDQ3YjQ5OThlMjBk&hl=en
    The link will take you to where you download it as a jpeg. It’s a very high resolution product of Photoshop.

    I wanted to make it as easy as possible to compare across sexual behaviors within age groups compare gender with gender as easily as possible, and compare sexual behavior within gender. That’s what led me to the setup you’ll see.

    Regarding data analysis, it does not elaborate much upon the original data. My main focus was making the data most accessible, so I focused mainly on visual details. This is because I don’t own any data analysis/graphing software beyond excel (sad, I know). I dissected pie charts from the data on excel and copy/pasted into PS. Most of the work was just in PS.

    The dark, neutral background helps the light blue and purple, which represent the genders, stand out; and the light yellow is a good third color that doesn’t clash with any of the other three but helps to tie them together. Basically, I let condom packaging inspire my color choices. It was the best place I could think to look.

    I wanted to have another feature that placed the same sexual behavior lines from each gender on top of each other, to facilitate that comparison, but I didn’t have enough time. There certainly are a lot of different features that could go into a more elaborate version of what I’ve done.

  • Did that last one post? If not, that’s too bad, I had a nice long explanation of my work. I’m too tired to type it up again. Nutshell: I focused mostly on visual aspects rather than data. Organization tried to make the comparisons within behavior categories across age groups within gender. Did that make sense? Group by gender, then by age and behavior cat. That way you can look up an age cat to compare across behavior, and down a behavior cat to compare across age, while not crossing genders. Still, the genders can compare because they’re aligned with each other. I would have liked to add another section where I put gender lines on top of each other for comparison, but this has already taken too long.

    Enjoy.

    Download high-res jpeg here:
    https://docs.google.com/leaf?id=0BwfS_L6-lf1KNGNhNWUwMGEtZWIwMi00ZmVjLWIwN2YtNDQ3YjQ5OThlMjBk&hl=en

    • One more point I’d like to make, even though it doesn’t concern my visualization. Most people who have commented on the differences between certain statistics as reported by males and females, like comparing men saying they’ve given oral to a woman and women saying they’ve received oral from a man, have assigned the difference to a factor of bad reporting by the respondent.

      Misreporting hardly seems to be the best explanation. Far more likely is what’s labeled above as “asymmetrical promiscuity” (I love the vocab there). If you have five men and five women, and one of those men has performed oral sex on all five of those women, then your statistics will show that 100% of women have received oral from a man and 20% of men gave oral to a woman. I believe it’s safe to assume that on average at least some males give more than women receive, vice versa. Such an assumption requires that the statistics differ to some degree. It would be crazy if they were identical.

  • I’m super late, but here’s mine which I did using Bime.
    http://shubhabala.com/?p=810

  • The super funny thing about this lies in the age group 20-24.. where its clearly reflects that while majority of girls(80%) were having the real thing whereas, most of the boys(83%) were busy with themselves alone :)

  • many thanks to the autho. I like the way you start and then conclude your thoughts