14 – Establishing Cause

Why Should I Care?

Something is a problem. We can reduce the negative impacts, but that is called a “band-aid” solution. It does not fix the root of the problem. If only we could find the root cause of the problem, then the solution would actually fix the problem.

Definitions

Cause and effect: Understanding phenomena in terms of systematic connections, which can be discovered through the research process.

Cause: The independent variable whose importance was confirmed by empirical testing.

Effect: The dependent variable whose variations are explained by a cause.

3 Criteria for Causation

Temporal Order: The independent variable has to move before the dependent variable.

Association / Correlation: The idea of a relationship between variables such that changes in the value or strength of one or more independent variables are associated with changes in value or strength of the dependent variable.

Elimination / Reduction: Elimination of other possible factors that could have been in your hypothesis but have not been observed to be associated to the dependent variable. 

How to Establish Causality

If you plan on conducting an explanatory study, you hold the ambition of “confirming” if your hypothesis is right or wrong. This is called establishing causality, or a “cause and effect” relationship.

There is no such thing as “proof”; that is too ambitious for eternally skeptical scientists.Remember that the aim of science is not only to document life on earth, the aim of science is to explain why and how things happen to help humans solve important problems. Identifying and confirming a causal relationship is the “holy grail” of science.

Student’s words   “A precedes B” “A has to be related to B” “B has no other explanation than A”   Souleymane Camara, RM class of Winter 2017

Example – What is the root cause of sadness?

If we knew, we could solve that problem more effectively.

Hypothesis A:                            Sadness may be caused by social isolation   Hypothesis B:                            Sadness may be caused by lack of physical activity. Hypothesis C:                            Sadness may be caused by a genetic predisposition. Dependent variable:                   Sadness Sadness depends on how isolated people feel, how often they play sports, or their genetic makeup. Independent variables:               Isolation, Inactivity, Genetics   Isolation, Physical Activity, and Genetics do not depend on how sad people are.

The key to establishing cause and effect between Sadness and Isolation, we must show three criteria:

• Temporal Order            
• Correlation                               
• Reduction                                

Let’s say a scientific study is conducted. And here are the results.

• The people felt sad before they isolated themselves from their friends
• The people ended up isolated, and even sadder
• The people were inactive before they became sad
• The people were even sadder, the less activity they did
• The people’s genetics were set before they were sad.
• The people’s genetics had no impact on how sad they were.

Independent Variables	Temporal Order	Association / Correlation	Possible Cause
Isolation	NO	YES	NO
Inactivity	YES	YES	YES
Genetics	YES	NO	NO

Interpretation

According to this fictional study, we see that Isolation cannot be cause of sadness, because people were sad before they isolated themselves socially. Therefore Temporal Order is not achieved. This variable is not a cause, we can eliminate it from the list of possible causes.

Second, Inactivity could be a cause. It happens before the occurrence of sad feelings, and it is correlated with the sadness data.

Third, we see that Genetics can be eliminated. It does respect temporal order since our genetics are set at birth, but the data do not show a particular gene to be correlated with increasing sadness.

Finally, we can eliminate Isolation and Genetics. Leaving only Inactivity. According to this fictional study, Inactivity would be the only possible cause of sadness.

MORE RESEARCH IS NEEDED BEFORE THIS RESULT BECOMES A LAW OF SCIENCE.

Strength of Variables

When many variables are established to be in the right temporal order, and observed to be correlated, it is not possible to identify a true cause. Further analysis must try to isolate the effect of each variable.

Researchers may still try to evaluate the strength of each factor.

Contributing Factor. This variable is important. When it is present, it helps to strengthen the “effect” measured. For example, a good night’s sleep helps to perform better on tests. But this factor is not enough.

Necessary Condition. This variable is very important. Without it, the “effect” is very weak. However, it is not the only factor which determines the effect. For example, it is very important to study for tests.

Sufficient Condition. This variable is the most important. Most studies are not able to achieve this level of scientific power. This is pure causality. Without this factor, the “effect” is not there. With this factor, the “effect” is very strong. For example, mastering class material is the only condition that guarantees success on a test.

Identify the strength of the independent variable.

When a student sleeps well, then usually his test scores are improved.  Contributing factor.

If a student studies at least 12 hours to prepare for a test, then his test scores are usually stronger. Necessary condition.

 If, and only if, a student achieves mastery of academic content, then his test score will be very strong. Sufficient condition.

A fictional hockey parable Our study of the Club de hockey Canadien de Montréal, Inc. evaluated the importance of many factors in determining the ultimate level of success, winning a National Hockey League Championship and its prestigious Stanley Cup. Our data shows that “team cohesion”, “team spirit”, “quality of management”, and “player health” were contributing factors to the success of the team in the end-of-season tournament. Other factors are qualified as necessary conditions. Of course, the “regular-season standing” is necessary to be admitted to the “playoffs”, since only the top 16 teams are included. Also, having a “top 10 goaltender”, and a “top 10 goal scorer”, were strongly correlated to Stanley Cup wins. However, our data suggests that the only sufficient condition, albeit once other factors are found to be present, was the variable “Large Group of Under-Paid French-Canadian Hockey Players”. In effect, the Canadien have never won a Stanley Cup without at least a dozen under-paid French-Canadian members on its roster. Other variables such as “Beer Magnate Owner”, “Quebec Political Climate”, “Economic Conditions”, “Canadian Exchange Rate”, “Butterfly Goaltending”, and “Curved-Blade Sticks” were not correlated with the dependent variable: winning the Stanley Cup.