Psychological Correlates of Feelings Toward Labor Unions among Liberals

27Feb/11

Psychological Correlates of Feelings Toward Labor Unions among Liberals

I have been reading a great deal lately about the labor battle in Wisconsin lately. As someone who rarely has had a traditional job, I have never had a well formed opinion about unions and it has been an interesting opportunity to think about the role of unions in society. There have been a great number of polls lately, each of which provides fodder for our innate abilities to confirm what we already believe to be true (confirmation bias). What psychological (as opposed to demographic) variables might lead someone to have warm or cold feelings toward unions?

By the time we can vote, we have developed coherent narratives that help us make sense of our emotions, beliefs, and opinions. In psychology, we often study individual variables and their impact on attitudes, but the real world is more complex and there are a whole host of attitudes, opinions, and dispositions that may have an impact on your opinion about unions. As such, I thought it might be interesting to look at the whole picture of what our yourmorals data shows as the correlates of warm or cold feelings toward unions.

The below chart (click on it to enlarge) is sorted from measures/beliefs that are most associated with warm feelings toward unions to measures/beliefs that are negatively associated with warm feelings toward unions. Warm/cold feelings were assessed using a feeling thermometer scale from 1-7. Our sample is not representative, so any conclusion that you may draw would be based on the idea that the psychological associations in our overly educated, liberal leaning, internet user sample would hold for other groups. To help isolate psychological variables, I ran the analysis on only those who self-identified in our sample as liberal, effectively holding that variable somewhat constant (I say somewhat because within this sample, some people were more liberal than others).

I would love to hear what others see in these patterns, but my initial impressions are:

A lot of what is associated with being liberal is associated with being pro-union. It is likely a mistake to try and figure out which comes first as people certainly adhere to their party positions, but people also certainly gravitate toward their parties due to psychological variables. It is all tied together and research supports both relationships. As such, it may make sense that Wisconsin Governor Scott Walker's decision to not only try and reduce pay, but effectively try to end all union representation for public workers, meets with such vehement opposition.
Other oriented connections appear even more related to feelings about unions beyond what one might expect from simple liberal partisanship. For example, identification with country is actually negatively associated with liberalism, but is positively associated with feelings toward unions. All measures of connection to others seem to have positive relationships. The Big 5 personality dimension of agreeableness (e.g. being trusting) has an almost equal relationship as the dimension of openness to experience, which is usually the dominant predictor of liberalism among Big 5 dimensions.
Dispositional emotional reactivity appears to be a predictor of how liberals feel about unions. Liberals who are empathizers (on Baron-Cohen's measure) who care about the less fortunate, feel emotional when perceiving beauty, and are also slightly more prone to depression tend to be those who feel warm toward unions.
In contrast, rationality, a liberal hallmark, is not related to feeling toward unions. Belief in scientific causation is strongly associated with liberalism, but not related to feelings toward unions among liberals. Experiential thinking appears slightly positively correlated with positive feelings toward unions among liberals even as it is negatively correlated with liberalism in our wider dataset. Rational thinking is not correlated with feelings toward unions, even as it generally is associated with being liberal.

Overall, the impression I get from the pattern is that it is the bleeding heart liberals, as opposed to the more rational, scientific liberals, who likely feel more connected to the ongoing protests in Wisconsin. But I welcome alternative ideas/interpretations as well as ideas about how these results might not hold in other populations, as the interaction would likely prove instructive.

- Ravi Iyer

Comments

comments

Old Comments (6) Trackbacks (0)

wreaver
February 28th, 2011 - 21:14

Ravi,

Why not create a Principal Component Analysis?

First, create a palette of colors for each choice for an individual’s feelings towards a union. For ex: if the choices were 1 to 5, then maybe something like 1=yellow, 2=orange,3=red, 4=purple, 5=blue. (You could also use different kinds of symbols instead of colors.)

Then on the Principal Component Analysis graphs, for each individual, make their “dot” with the color appropriate for their choice for their feelings towards unions.

Perhaps the eigenvectors/eigenvalues will tell us something.
Tom Cannon
March 1st, 2011 - 12:32

I’d bet that many people have negative feelings about unions that resulty from being forced to join a union and pay dues (as I was when working one summer in a large manufacturing plant), working for a company that was going through the process of being organized by an industrial union (ditto above), or having a friend or family member who expressed negative feelings about an experience they had with a union (again, ditto above).
Ravi
March 5th, 2011 - 23:39

Thanks for the suggestion. Would those be feedback graphs that people would get after filling out measures on yourmorals? I’m not sure what you mean by graphs “for each individual”. Certainly, I could condense the above measures into fewer measures and probably should come up with a visual way to do that.
wreaver
March 7th, 2011 - 23:11

Ravi,

If you are not familiar with a Principal Component Analysis, see here:
http://www.cs.otago.ac.nz/cosc453/student_tutorials/principal_components.pdf

Lets say there are “n” people who answered the YourMorals quiz who self-identified as liberal.

And let’s say there are “m” questions you care about, from the YourMorals quiz. (I didn’t count how many questions you have above. The number up there would be “m”.) And we’ll (re)number them from 1 to m; so we’ll have question #1, question #2, question #3,… question #m.

So then, for example, let’s say that the answers person number “i” gave is represented by:

X{i} = [1, 7, 2, 2, 5, 4, 5, 1, 3, ... 4]

(I just made up the number “1, 7, 2, 2, 5, 4, 5, 1, 3, … 4″, so don’t focus on the values… there just there for illustrative purposes.)

On the left of the equal sign we have “X{i}” which represents the answers of the i’th person. (Thus “X{7}” would be the answers of the 7th person; and X{14} would be the answers of the 14th person.) Note, we’re going to go from X{1}, X[2],… X[n] since we have “n” people.

On the right of the equal sign we have “[1, 7, 2, 2, 5, 4, 5, 1, 3, ... 4]” is a vector in mathematical terms. (In programming terms, you could think of this as an array.) This is a vector that has “m” numbers in there. I.e., this vector has “m” dimensions. (Remember, “m” is the number of YourMorals questions you care about.) So, the first number in there represents the answer for questions #1, the second number in there represents the answer for question #2, etc.

You would use X{1}, X{2}, … X{n} as the input to calculate your Principal Component Analysis.

Now (and this could be the tricky part) you’ll want to somehow classify X{1}, X{2}, … X{n} to tell the software that creates the graph generated from the Principal Component Analysis to color the point of each X{i}differently, depending on their “feelings towards a union”.

What you’ll look for in the graphs generated from the Principal Component Analysis is some kind of stratification of the data. If we see that, then we can look at what that dimension (that the data is stratifying along) means. (Each dimension will be an eigenvector, and we may be able to draw some meaning from it.)

(Apologies if you already know this stuff.)

If any of that wasn’t clear let me know, and I can clarify. Also, if you need some help with the analysis (and don’t have a math guy around) let me know and I can lend a hand. (These types of things are trivial for me.)
Ravi
March 12th, 2011 - 07:11

@ tom, that makes sense. we are all certain to generalize from our experiences.
Ravi
March 12th, 2011 - 09:41

I am familiar with Principal component’s analysis, but I think I just misunderstood what you meant by “for each individual”. Perhaps you mean that I could color code the graphs by which measures load on the same factor? Yes, that would be an improvement, and the next time I do such a post, perhaps I’ll come up with such a color scheme. Thanks!