rival causes and statistics
TRANSCRIPT
RIVAL CAUSES AND DECEPTIVE STATISTICS
Review of Browne and Keeley (2012)
As SLPs
We need to be reasonably sure, at the very least, that what we are doing is having a positive impact.
Often, the perfect causal relationship does not exist.
Other factors, often known as rival causes, can be equally responsible for the change in our clients.
When evaluating information about treatment programs and research studies, rival causes are also important.
May think about these as nuisance variables, or important variables to consider.
What are the rival causes?
Rival causes imply some alternative interpretations for the interpretation made by the researcher for why events turned out as they did.
A plausible alternative explanation that can explain why a certain outcome occurred.
Need to look deeply at the evidence and try to understand the causal relationship that the researcher or writer is hoping to have us accept.
You should view your therapy the same way.
Things to consider
1. Many kinds of events are open to explanation by rival causes.
2. Most communicators will provide you with only their favored causes…Must analyze these.
3. Generating rival causes is a creative process. The rival causes will not be obvious.
4. The certainty of a cause is inversely related to the number of possible causes.
Questions to Ask
Can I think of other ways to interpret the evidence?
What else might have caused this act or these findings?
If I look at this from another point of view, what might I see as important causes?
If this interpretation is incorrect, what other interpretation might make sense?
Comparing groups who receive different treatments is not as straightforward as we might hope
Oversimplifcation fallacy leads us to determine results based on causal factors that are insufficient to account for the event or by overemphasizing the role of one or more of these factors.
Confusing causation with association
Problems Determining Causation
Confusing “after this” with “because of this” Post Hoc- Assuming that a particular event, B, is
caused by event A. Could be that the sequence is a coincidence or
due to other factors. Explaining Events
Fundamental attribution error-Overestimate the importance of personal tendencies relative to situational factors in interpreting the behavior of others.
Victim blaming is an example of this. Could also related to wrongful theories.
More Causation Problems
Evaluation of Causes
Do the causes make logical sense? Are the causes consistent with other
knowledge that you have? Are the causes important for explaining
or predicting events?
In Your Own Communication
Be open to other potential causes. Be clear with your clients what parts of
your treatment are predicting the outcome.
Be aware of rival causes, attempt to control or address their impact on what you are doing.
Are the Statistics Deceptive?
What meaning can we take away from the use of statistics? Often times, researchers can use statistics to
deceive. Also, statistics will have inherent limitations. Use of statistics require certain rules to be
followed. Certain statistics are simply inappropriate for certain kinds of data.
Some times they are ill-defined Do not fit the phenomenon being studied. Lack of operational definitions serve to confuse
things.
Ways to deceive
Unknowable and biased Incomplete reporting of data- Need to look for
the information that is missing. Making choices as to what data is reported.
Confusing averages Mean Median Mode Range
Ways to deceive
Concluding one thing, providing data for other finding Use statistics that prove one thing, but then claim
to have proved something different. Generally, researchers are supposed to be “blind”
to data until the results are found. Readers should attempt to do this, too.
Ask---”Do the statistics match the conclusions??” Deceiving by omitting information
Statistics are often incomplete. “What relevant comparisons were omitted?
What didn’t they explore?”
Clues Try to find out as much as you can about how the
statistics were conducted and findings were attained. Be curious about the data, especially averages and
percentages. Look for the range and other data that is often omitted.
Be alert to researchers using statistics to make one conclusions, though they have proved something else altogether.
Blind yourself to the researcher’s statistics and compare the evidence with what is actually provided.
Form your own conclusions from the statistics. If it does not match the researcher’s conclusion, then something is wrong.
Determine what information is missing.