Here’s something we need to investigate further. While researching a major post on Political Correctness, I have run across a study on the web site of the Pew Research for the People and the Press. This study, “sorts voters into homogeneous groups based on values, political beliefs, and party affiliation.” It then looks at various aspects of their behavior and, using survey results. presents statistical evidence and analysis. It’s a big study with a lot of interesting ideas threaded through it. I was enjoying reading through it and was thoroughly sidetracked for a day or so as I read it. One thing jumped out at me and I wanted to pass it on. Second Draft needs to take a hard look at this.
It is not specifically referred to in the written analysis of the article but there is a glaring (and I do mean red, purple and throbbing) anomaly in the data presented. On page 73 of the report, there is a table entitled “Typology Groups and Media Use”. This chart looks at the kinds of media that each political type relies on for their information. It immediately jumped out at me that the largest single political type had the smallest average use of television as an information source. This was no small artifact. The Liberal typology was almost 40% (37.54, to be exact) larger than the next largest group (Conservative Democrats) and their television usage was nearly 20 % (17.55%) lower than the next lowest group (Upbeats).
I decided to drill down into the numbers. Now, this is a little suspect because the numbers are already averaged out and it is not clear how some of them were derived from the research (for example it is not clear whether the break down of the television numbers into categories like network, local, CNN, etc…, reflect some sort of break down of anwers that were asked independently or is the television average was derived from aggregated usage numbers for all categories). But if the numbers are good to begin with, then my manipulations should not be too far out on a statistical limb.
I totaled up the percentages for each of the media cited as main sources of information by each political type. This should give a rough measure of how broad a range of information sources the average subject in each group is accustomed to using.
Here are the numbers I came up with:
Social Conservatives 269
Pro-Government Conservatives 264
Conservative Democrats 262
Upbeats 256
Disadvantaged Democrats 256
Enterprisers 253
Bystanders 242
Disaffecteds 239
Liberals 230
It may have meaning or not (certainly there are methodological questions that would have to be addressed) but the numbers are tantalizing. According to these numbers, Liberals are at the very bottom of the scale for media variety. This implies that, as a group, they tend to draw on a thinner variety of information sources.
On the other hand, the largest conservative typology (Social Conservatives) has a media variety index that is almost 20% (16.955) higher than that of the liberals.
It is also interesting that Liberals rank highest for Internet usage. Anyone reading this post knows very well that, between search engines and link sharing with friends and colleagues, when you read and explore on the internet you are mainly pursuing sources that you agree with.
So, it makes me wonder if what many of us think might be provable; that conservatives tend to look at more possibilities before making up their minds and liberals tend to stick their index fingers into their ears and shout “La, La, La, La” when they come up against information that does not confirm their preconceived ideas.
5 comments:
Not to be a wet-blanket, but...OK, let's say "we" use 16.955% more sources than "them." This is much closer to 1/6th (16.666) than 1/5th (20%), right?
So in practical terms, if all you know about 2 people was that one subscribed to 12 magazines, while one subscribed to 14...is that really significant? Wd we really assume one is inferior to the other?
To me, what jumps out from this survey is the uniformity of the results...269 at the top, 230 at the bottom, I'd say those differences are insignificant in practical terms.
J.O., my friend, let me see if I can set your blanket on fire- or at least dry it out a little.
I don’t want to get too deeply into this debate because, as I said in the post, there are a lot of methodological loose ends here anyway. But I will insist that there is good reason to pay attention and pursue more information.
Actually, in social science research any difference over 5 percent between two segments of a population will generally turn out to be “statistically significant (i.e. attributable to a real, quantifiable difference in behavior). The difference here is huge.
As for your analysis of the real meaning of the statistical difference, the example you pose is somewhat misleading. The number of media contacts per subject is much lower than 12 or 14 per subject. Actually, one method of computing the difference Note too that there are a lot more Liberals in the sample. I have tried (in the privacy of my own home- because without the original data base all of these machinations could turn out to be illusory) actually puts the high end of the “media per subject” statistic at about 1.3 and the low end (for the Liberals) at somewhat less than 1 (about .6).
We have to remember that we are talking about individual human beings and that a group that is averaging .6 main sources of information is a pretty uninformed group. That means that for everyone that reads a newspaper and watches the news on TV there is a someone who may be catching the back end of the national news as they tune in to find out if Britney is REALLY pregnant. My guess would be that there are, in that group, a great many of the ignorant followers of “popular opinion” whose progenitors have made up the lynch mobs, good Germans, docile sheep and guillotine watchers of human history.
Whatever the percentage, the general belief is that more information is better than less but if they are signing on to the Daily Kos and counting it as a main source of information, maybe not…
Not totally true. I am a liberal, and I read your post. :-)
Welcome, Anonymous and congrats! You are obviously one of the open minded ones.
Yaacov, the 5% you refer to is a level of confidence, not necessarily the percentage difference. Let me explain. If you toss a coin 10 times you may get something like 6 heads and 4 tails a 20% difference. Statistically significant? Not at all. However, with a much larger sample, say 10,000 tosses you will get nearly a 50/50 split and you can be sure at a 95% level of confidence that your results are not due to "chance."
The 95% level of confidence simply states that if you run the same experiment 100 times, you should get the same results 95 times or more.
GM's Corner
Post a Comment