Who wants to eat some cake?
(Check your assumptions, first!)
If you stopped to read this post or watch the video, that title line created some expectations in your mind. You "tested" that assumption by reading/watching more.
The actual content will validate if your hypothesis was true. The action you take towards this post will impact your feed. And the feed you see is your reality based on multiple actions.
So, every assumption you make impacts your actions. and through that, it impacts how your reality works. If you fail to validate your assumptions consistently, your reality can be very different from your vision.
But how to validate assumptions meaningfully? Science & Statistics have the answer: Hypotheses Testing!
Null Hypothesis:
- Usually the opposite of our assumption.
- No relationship between X & Y (status quo).
Alternate Hypothesis:
- The assumption we are making (X impacts Y).
- This is what we are actually testing against the Ho.
Hypothesis Cake Tiers:
1. Determine a clear assumption and goal to test.
2. Data collection and cleaning of samples.
3. Define the hypotheses clearly (Ho vs. Ha).
4. Set significance level (alpha); standard is 5%.
5. Select the correct test to use for your goal (t, Z, F, etc.).
6. Critical Values are based on the alpha and sample size.
7. Calculate your test statistic precisely (data team can help).
8. Compare the test statistic with critical values.
9. Conclude whether you can Reject or Fail to Reject Ho.
10. Actions should be based on the conclusion you derived.
And the cherry on top?
Reliability!
Actionable Insights:
1️⃣ Practice writing out your assumptions clearly.
2️⃣ Align your hypotheses with business goals.
3️⃣ Seek help from your data team for tests.
4️⃣ Take time to understand conclusions.
5️⃣ Actions should depend on results.
Hope you enjoyed this "assumptions" to "actions" foray!
🔔 Dr. Kruti Lehenbauer of Analytics TX, LLC
hashtag
#PostItStatistics
hashtag
#DataScience
hashtag
#Economics
P.S: Would you like to learn more about a specific test?
