Thursday, May 06, 2010

Discussing disparate trends in permissiveness

Over at SocImages, a recent posting (and comment) about "Disparate Trends in Permissiveness" got me thinking about some things. The original data presented were the following two graphs (original source here):

The point was in wondering why (between these two graphics) the general trend in both graphs is downward, and Japan is an outlier when it comes to prostitution. One thing that was immediately apparent to me (and others) is that Japan doesn't necessarily pose as a good comparison group, since there are cultural and linguistic divides that are much greater than between any of the other six countries depicted. The better (and perhaps more interesting) thing to do is to compare it to a closer cultural neighbor.

Going over to the World Values Survey page, I was able to download the massive files in order to do some statistical analysis using SPSS and then exporting the data to Excel in order to graph it. I found a few things while playing around with the data:
  1. The dataset is HUGE! I had to cut it down to only a few items due to its unwieldiness. 
  2. A possible reason why looked at the percentage of people who responded "Never justifiable" to the questions is because they form the largest single percentage among the ten response groups, except in a few cases (such as Norway and Switzerland in their most recent surveys). Looking at other groups would not likely show values greater than 5%-10%, so most of the change in most of the countries is likely to be seen in the "Never justifiable" group.
  3. The sample sizes were quite large, always greater than 1000 people per survey year, so the standard error for all of these surveys should be quite low.
  4. Japan matched much better with South Korea and South Africa than with European countries when it came to these two parameters.
  5. The variance in the responses among the Japanese and South Koreans were generally greater in 2000/2001 than they were in 1981/1982 with regard to justified homosexuality. However, while South Koreans' response variance increased from 1982 to 2001 with respect to justified prostitution, it remained almost unchanged in Japan. (An increased variance would indicate a greater variety of responses given, something not immediately visible when looking only the % response of one extreme position.)
I present the comparative data for Homosexuality and Prostitution for South Korea, Japan, and South Africa in the same manner as done originally by Will:

So.... perhaps the reason why Japan is different than Canada, Germany, Great Britain, Spain, Sweden, and the United States is because it is not culturally similar to these countries. That it is more similar in some respects to South Korea (arguably a closer cultural neighbor) is less surprising.

In one of the comments at the SocImages post, "5.56" states:
Also, given how much the lines on the graph drastically zigzag (society cannot change from 20% acceptance one year to near 40% the next) around, I find it very hard to put much faith in the survey’s legitimacy.
However, much of this can be explained by the fact that Will's depiction only showed the percentage of people who responded with an answer of 1 (i.e., "Never justifiable") as opposed to a 2 or 3 (which, out of a ten-point-scale, are still pretty close to "Never justifiable"). By grouping the other responses together into three classes, one can see how the shifts over time occur. In the following graph, each time period's survey responses are groups into four categories: Never (response=1), Low (response=2-4), Medium (response=5-7), and High (response=8-10):

As one can see, while the "Never justifiable" category does "drastically zigzag" (especially between 1990 and 1995), this is due to concomitant increases in the other categories, especially in the medium and high groups. Similarly, the graph of the change in the average responses does not "drastically zigzag".

Nothing really surprising in the changes in either representation, but it does look suspicious when looking only at the changes in the "Never justifiable" responses (in other words, it pays to understand what it is one is looking at).

1 comment:

Fr. said...

Hi Umlud,

I am still reading through your comments over at SocImages on this topic. Would you mind posting (or sending me) the dataset you cropped out of the -indeed gigantic - WVS one, so that I can feed it to Excel and have a look too?