Step 1.3: Hypothesis Testing: Test your specific hypotheses

During Hypothesis Testing, students propose a specific hypothesis about smoking that is testable using a question the database. On this page, students access the question of interest, enter their hypothesis and the specific parameters of their query, and then submit their query. The odds ratio and 95% confidence interval are calculated by the database program and provided on a new page. Students are prompted to interpret the data by answering several questions.

1. Select an item for testing

Click "on" a row/item relevant to your OVERARCHING question/hypothesis:

ID Questions
No Study Population Exposed Not exposed OR CI Sample size
2. State your SPECIFIC hypothesis

2-1. You selected:
.

2-2. Your SPECIFIC hypothesis for this question:
2-3. Your definition of "Exposed": 2-4. Your definition of "Not-exposed":
3. Identify answers corresponding to your exposure:
Exposure/Non-Exposure Key Answer
4. Choose Study Population:
5. Estimate Odds Ratio

Report your results and interpretation

Report No: Investigator:
Selected Question:
Your SPECIFIC Hypothesis:
Exposed
Not Exposed :
Study Population: Sample Size :
Odds Ratio :
2 x 2 table
 Cases Controls
Exposed
Not-Exposed
Relationship between the Odds Ratio and Confidence Interval:
Interpret the result:
a) Use the odds ratio in a sentence that describes what it means.
b) Is there an association between the exposure and outcome? How do you know?
c) Do you think this exposure causes people to become a regular smoker? Apply the criteria for causality below to support your answer.
1. Strength of association. A strong association between the exposure and outcome is demonstrated using statistical methods (The larger the odds ratio, the stronger the association). The further the 95% confidence interval is from 1.00, the less likely the odds ratio occurred by chance.
2. Dose-response relationship. An increased dose of the exposure is associated with a greater risk for having the outcome. (For example, having more than 2 passengers is associated with a greater likelihood of getting into a car accident than having only 1 passenger.) May not be applicable to the smoking behavior database study.
3. Temporal sequence. The exposure must occur before the outcome. (Sometimes in case control studies, this can be difficult to verify.)
4. Consistent with other studies. The result should be mostly consistent with what is already known in the field. (If it is not, there is always the possibility that you have discovered something new and unexpected, but there is also the possibility that your study design or assumptions were in some way flawed.)
5. Biological plausibility. The result should make biological sense. (For example, it makes sense that having passengers in the car would cause one to get into a car accident.)
6. Lack of confounder or significant bias. Can the association be explained by another factor? Is there a factor or bias that explains the association?
Manage your report:
Association is not necessarily causation. Epidemiologists determine an exposure(s) to be a cause if the exposure(s) increases the risk of someone becoming a regular smoker. To infer causality, epidemiologists may apply the criteria for causality to case control study results.
Association is not necessarily causation. We are asking if the exposure(s) increases the risk of someone becoming a regular smoker.
Your broad hypothesis from smoker profiles, your intuition/observations, and past research
Your hypothesis that relates to your overarching hypothesis from the Smoking Behavior database questions
All the people included in the study.
Is the way people organize data in a case control study. It tells you how many smokers and how many nonsmokers fall into the exposed or not exposed categories. How you define exposed and not exposed depends on how you drag and drop the answers to this question.
The confidence interval is a tool to help you decide if your result is meaningful to the entire population- people in general. If the confidence interval has the number 1.00 in it, this means that even if your Odds Ratio is bigger or smaller than 1.00, there is not an association between the exposure you identified and regular smokers. However if 1.00 is not inside your Confidence Interval, this means there is an association between the exposure and becoming a regular smoker.
The factor you think might have an influence on someone becoming a regular smoker. For example, believing smoking is not harmful for your health OR having a least 1 parent who smoked.
The factor you think might protect people from becoming regular smokers. For example, believing smoking is harmful for your health OR not having a parent who smoked.
The total number of people who responded to this question, both cases and controls.
An odds ratio of 1 means there is no difference between regular smokers and nonsmokers. For more information, go to here
People who are regular smokers
People who tried or experimented with smoking but never became regular smokers

This report has successfully been saved

The previous note is overwritten with the current note. Do you want save?

This report is deleted. Do you want delete?

The previous note is overwritten with the current note. Do you want save?