Here is an ok thesis:
Among Wisconsin 9th graders between 2010 and 2015, students that had longer hair were also more likely to attend college 4 years later compared to students with shorter hair.
Here is the MADLIB:
POPULATION: Wisconsin 9th graders between 2010 and 2015
UNIT OF ANALYSIS: students
TREATMENT: Longer hair
OUTCOME: attend college four years later
CONTROL_GROUP: shorter hair
CONFOUNDERS: none
Among POPULATION, the UNITS-OF-ANALYSIS with higher TREATMENT also have higher/lower OUTCOME compared to UNIT with CONTROL_GROUP after controlling for CONFOUNDERS.
Spelling it out like this clarifies the results of the data analysis.
It identifies the population.
The unit of analysis.
The groups (treatment/control) and the outcome.
The results of the data analysis become clearer when this information is fully stated in one sentence, targeted for a general audience.
In this case, these results are deceptive.
We can improve the original thesis (and perhaps make it less deceptive) by identifying the variables that were “self selected” (or marking the decisions by individuals).
Here is a better thesis:
9th graders that choose to have longer hair are also more likely to attend college 4 years later.
Do you know why? Give a “general audience” explanation for why this result is deceptive.
This research can be improved by controlling for a confounding variable.
Here is the best thesis:
Among Wisconsin 9th graders between 2010 and 2015, students that chose to have longer hair were not more likely to attend college 4 years later after controlling for self-reported gender.
Again, here is the mad lib:
Among POPULATION,
the UNITS-OF-ANALYSIS
with higher TREATMENT
also have (DO NOT HAVE)
higher/lower OUTCOME
compared to UNITS
within CONTROL_GROUP
after controlling for CONFOUNDERS.