They say two people can’t see the same mirage. I wonder who the hell started saying that?

Last fall, in the article, Polls Are Not Rigged, But They Also Aren't 'Scientific', prolific science writer, Alex Berezow, argued that political polling is more like an art or craft that has been honed over the years. That is why they often work so well, but they are not really scientific. This set off an amazing firestorm among the readers.

The responses were divided into two categories of people. One category thought polling was indeed scientific and they were adamant about their reasoning. The other group though it was clear that sometimes social science is just social studies, and anyone can see that. As a social psychologist, I was struck by this. These are all smart people, and it was fascinating to see all of these smart people lining up in two distinct groups to profess their sincere views. I know it is possible for people to have different opinions, but ...

**Can two people
see the same mirage?**

Consider the
two groups of people in this case – the *polling
is science* versus the *polling is not
science* groups. If you assume that two people cannot see the same mirage,
then one of these groups is right, and the other group is full of people who
are all seeing the same mirage. But, that means it is possible for two or more people
to see the same mirage. Even though paradox is not allowed in contemporary
science, the appearance of paradox is arguably a sign of strength in science. The
fact that we have this saying about two people not seeing the same mirage, and
that it actually reveals a social paradox of life, makes things all the more
interesting. Philosophically, we must be entering juicy territory.

So, not only can two or more people see the same mirage, given two or more groups of people, it is actually possible for all the people in each group to see the same mirage, even while both groups are looking at exactly same thing. Sound crazy?

What about the Democrats and the Republicans? They are each looking at the same world, and they are each sharing their own brand of mirage. An Independent is a person who does not want to buy into the mirage of either major party. Furthermore, Republicans and Democrats each represent their own weltanschauung, but they will each use the weltanschauung of statistics to their own advantage because Republican pollsters always poll in a way that favors Republicans, and Democrats do the same. That is the most scientifically reliable result in the world of polling – I can scientifically predict that Republican and Democrat pollsters will always create scientific polls that favor their own candidates.

Back to Berezow’s article, there was one response that stood out in a strange way. Hank Campbell made several observations stating that polling is clearly not scientific. At one point he said this: “Science has a theoretical foundation. What is the theoretical foundation for the result of that football game you will bet on? What is the theoretical foundation for political polling? None, it is an educated guess - which is not science.”

Wait a minute – what? Back up …

*What is the **theoretical foundation for **political polling?*

That was a moment for me.

Like hearing a big GONG. Not in my head – in the Grand Canyon, echoing and echoing, on and on. We always have everything in science divided into theory and methods. Theory explicates the underlying logic of the object or the subject of study, and it forms the conceptual framework for research. Then, research is about methodology, the unceasing attempt to minimize researcher bias and measurement bias in the course of testing theoretical hypotheses. They are like steps you take, walking up the grand spiral staircase in the tower of science, ascending to higher levels of understanding.

But …

GONG ...

*There
is no theory section in the methods section.*

When we study quantitative methods, we learn the mathematics of statistics.

There are reams written, untold numbers of tomes written as critical expositions about methodology. There are volumes written about statistical probability theory, but, they all assume the integrity of statistical methods. In other words, statistics is largely derived from probability theory, however, they are qualitatively distinct. Statistics deals with making inferences from data, and probability deals with random processes underlying the data. The whole of statistical probability theory is replete with formal theories and volumes of mathematical engineering and testing.

**Why Statistics Works When It Works**

Yet, the truth is that statistics only works when it works.

In these cases, the stats themselves are simply assumed, much like Euclidian premises for a formal theorem. This is quite striking because science is almost always divided into theory and methods. But, in the case of statistics - and polling is a specific application of statistics – all the theory backs up the statistics when it works, and when it does not work, then all the theory in the world does not help or apply. We never discuss the theory of statistics, we just assume mathematicians figured it out and they are the smartest people – end of story.

Therefore, the theory of statistics is ultimately reified by its application, but there is never any critical discussion of the theory of quantitative methodology or statistics, there is only the learning of the math of statistics. There is no theory section in a methodology text book discussing the theoretical aspects of the methodology itself.

Of course, I am sure someone will offer up the bell-shaped curve as a theoretical framework for statistics. It is true that statistics and probability both believe they are considering data that conforms to the bell-shaped curve … most of the time. It is a fascinating idea, the average number of incidents comprises the largest number of incidents, and the farther one goes in either direction away from the peak of the curve, the less incidents are registered. Statistics is largely based on this idea.

In this figure, we see the basic bell-shaped curve, also referred to as the normal curve, as it represents a normal distribution.