PPT Hypothesis testing 21 Aug 2025.pptx

Hypothesis testing

1. Testing of hypothesis Presenter: Dr Shilpi Rani Biswas Moderator: Dr Shoebul Haque (SR) Peer Support: Dr Rashmi Chandra (JR3) Department of Pharmacology & Therapeutics King George’s Medical University, Lucknow U.P. India Email id: [email protected] 8/20/2025 Dr. Shilpi Rani Biswas 1

2. Contents Introduction Types of hypothesis Characteristics of hypothesis Sources of hypothesis Steps of hypothesis testing Summary References 8/20/2025 Dr. Shilpi Rani Biswas 2

3. Specific learning objectives By the end of this teaching-learning session, the co-learner will be able to: Define important terms related to hypothesis testing Differentiate b/w research question & research hypothesis Enumerate steps of hypothesis testing Differentiate b/w type 1 & type 2 error 8/20/2025 Dr. Shilpi Rani Biswas 3

4. Introduction Hypothesis “An idea or explanation of something that is based on a few known facts, but that has not yet been proved to be true or correct” OR It is tentative or expected prediction or explanation of the relationship b/w two or more variables in a specific population 8/20/2025 Dr. Shilpi Rani Biswas 4 Ref https://scispace.com/resources/how-to-write-research-hypothesis-definition-types-examples-and-quick-tips/

5. Research question v/s hypothesis Research question Identifies the problem Open-ended FINER Guides study design, population, outcomes Eg: Does LIPUS(Low intensity pulsed ultrasound) improve pain tendinopathy? Research hypothesis Predictive statement to be tested Declarative Derived from research question Determines statistical testing & analysis Eg: LIPUS reduces pain vs placebo after 12 wks treatment 8/20/2025 Dr. Shilpi Rani Biswas 5

6. Functions of a hypothesis It offers explanations for the relationships b/w those variables that can be empirically tested Facilitates data collection, data analysis & data interpretation Gives direction to an investigation Furnishes proof that the researcher has sufficient background knowledge Leads to formulation of another hypothesis 8/20/2025 Dr. Shilpi Rani Biswas 6

7. Characteristics of a good hypothesis Clear & concise Specific Testable Measurable Relevant Simple Feasible Valid Original 8/20/2025 Dr. Shilpi Rani Biswas 7

8. Sources of hypothesis 8/20/2025 Dr. Shilpi Rani Biswas 8

9. Types of hypothesis Simple hypothesis Directional hypothesis Null hypothesis Associative hypothesis Complex hypothesis Non directional hypothesis Alternate hypothesis Causal hypothesis 8/20/2025 Dr. Shilpi Rani Biswas 9

10. 1.Simple hypothesis Statement which reflects relationships between two variables Example – Smoking is a prominent cause of lung cancer Low level of Hb leads to increase risk of infection 8/20/2025 Dr. Shilpi Rani Biswas 10

11. 2.Complex hypothesis Statement which reflects relationship between more than two multiple variables Example – The combination of smoking , family history & genetic predisposition increases the risk of lung cancer Satisfaction among patients who are older & dwelling in rural areas than those who are younger & living in urban areas 8/20/2025 Dr. Shilpi Rani Biswas 11

12. 3. Associative hypothesis Reflects relationship between variables that occur or exist in natural settings without manipulation Example – There is a positive correlation between physical activity and cardiovascular health 8/20/2025 Dr. Shilpi Rani Biswas 12

13. 4.Causal hypothesis Predicts cause and effect relationship b/w two or more variables Example – Increase sugar intake leads to weight gain Regular yoga practice lower blood pressure 8/20/2025 Dr. Shilpi Rani Biswas 13

14. 5.Directional hypothesis Specifies not only the existence, but also expected direction of relationship b/w variables Example – Sleep deprivation increases, cognitive performance decreases As carbon dioxide levels increase global temperature also increase 8/20/2025 Dr. Shilpi Rani Biswas 14

15. 6.Non-directional hypothesis Indicates existence of relationship b/w variables Example- There is a relationship b/w years of teaching experience & job satisfaction There is a significant difference in patient outcomes between hospitals A & B 8/20/2025 Dr. Shilpi Rani Biswas 15

16. Q. Which of the following is an example of a complex hypothesis? A. Smoking is a prominent cause of lung cancer. B. Low Hb level increases risk of infection. C. Combination of smoking, family history & genetic predisposition increases risk of lung cancer. D. There is a correlation between physical activity and cardiovascular health Answer- C 8/20/2025 Dr. Shilpi Rani Biswas 16

17. 8/20/2025 Dr. Shilpi Rani Biswas 17 Ref; https://images.app.goo.gl/zNgyYPbpxorC8t6L6

18. 7. Null hypothesis (H ) No effect or no difference between variables/parameters It predicts that changing one variable (independent variable) will have no effect on the variable being measured (dependent variable) Example- The number of calories consumed has no effect on weight There is no significant difference in blood pressure between patients taking Drug A and those taking Drug B 8/20/2025 Dr. Shilpi Rani Biswas 18

19. 8. Alternate hypothesis (H 1 ) An effect or difference exists between variables/parameters States that the parameter has a value that somehow differs from the null hypothesis Example- A low carb diet results in greater weight loss compared to a low fat diet Drug A is more effective than Drug B in reducing blood pressure 8/20/2025 Dr. Shilpi Rani Biswas 19

20. Q. Which of the following best describes the Null Hypothesis (H₀) ? A. It states there is no effect or no difference between variables. B. It states that there is always a significant effect. C. It specifies the exact probability of error. D. It predicts the expected direction of the relationship. Answer: A 8/20/2025 Dr. Shilpi Rani Biswas 20

21. Which of the following is a correct Alternate Hypothesis (H₁) for: “The mean hemoglobin in a population is claimed to be 12 g/dL”? A. H₁: μ = 12 B. H₁: μ ≠ 12 C. H₁: μ ≤ 12 D. H₁: μ ≠ 0 Answer: B 8/20/2025 Dr. Shilpi Rani Biswas 21

22. Hypothesis testing ( significance testing) A measure in statistics by which a researcher tests an assumption regarding a population parameter Conclusion It’s an inferential statistics - Purpose : To determine whether there is enough statistical evidence in favor of a certain belief about a parameter Indicates the expectations of the researcher regarding certain variables  Most specific way in which an answer to a problem can be stated Sample data Population 8/20/2025 Dr. Shilpi Rani Biswas 22

23. Hypothesis testing procedure 8/20/2025 Dr. Shilpi Rani Biswas 23

24. Steps of hypothesis testing 8/20/2025 Dr. Shilpi Rani Biswas 24

25. 1) Set/formulate hypothesis No significance difference b/w variables μ 1 = μ μ 1 = μ 2 p1 = p p1 = p2 Significance difference b/w variables exist μ 1 ≠ μ μ 1 ≠ μ 2 p1 ≠ p p1 ≠ p2 Null hypothesis(H ) Alternate hypothesis (H ) or (H A ) 8/20/2025 Dr. Shilpi Rani Biswas 25

26. 2) Select a level of significance ( α ) Maximum probability of rejecting the null hypothesis when it is true Maximum probability of making wrong decision Prechosen probability 8/20/2025 Dr. Shilpi Rani Biswas 26

27. 8/20/2025 Dr. Shilpi Rani Biswas 27 Zone of acceptance and rejection

28. Critical Z-scores for confidence levels Confidence level (%) Critical Zα/2​ score (Two-Tailed) Critical Zα -score (One-Tailed) Application 90% 1.645 1.28 Exploratory studies or less stringent precision requirements. 95% 1.96 1.645 Most common in research for balance of precision and certainty. 99% 2.576 2.33 Critical studies where high certainty is required. 99.9% 3.291 3.09 Rarely used; for extremely high precision requirements. 8/20/2025 Dr. Shilpi Rani Biswas 28

29. Interpreting the Level of Significance : If p-value (probability value obtained from test ) < level of significance : null hypothesis is rejected If p-value greater than α: null hypothesis is not rejected Example : Suppose you set α=0.05 and your test results in a p-value of 0.03. Since 0.03 < 0.05, you would reject the null hypothesis 8/20/2025 Dr. Shilpi Rani Biswas 29

30. Type 1 & type 2 errors 8/20/2025 Dr. Shilpi Rani Biswas 30 Decision Reality Ref-https://carreersupport.com/type-1-vs-type-2-error/

31. Q. Which error occurs when the null hypothesis is true but rejected? A. Type I error ( α error) B. Type II error ( β error) C. Sampling error D. Random error Answer: A 8/20/2025 Dr. Shilpi Rani Biswas 31

32. Q. The level of significance (α) represents: A. The maximum probability of accepting H₀ when it is false B. The maximum probability of rejecting H₀ when it is true C. The minimum probability of making a Type II error D. The probability of H₁ being true Answer: B 8/20/2025 Dr. Shilpi Rani Biswas 32

33. Step 3 : Setting a test criterion Parametric test Applied when the distribution is normal Quantitative data Metric scale(interval, ratio) Questions has mean or standard deviation Non-parametric test Applied when there is skewed distribution Qualitative data Nominal or ordinal scale Question has percentage or proportion 8/20/2025 Dr. Shilpi Rani Biswas 33

34. 8/20/2025 Dr. Shilpi Rani Biswas 34

35. 1 group Paired T test Parametric 8/20/2025 Dr. Shilpi Rani Biswas 35

36. 1 group Sign test Mc Nemar test Non parametric 8/20/2025 Dr. Shilpi Rani Biswas 36

37. Step 4 : Perform computations Test statistic Calculated from sample data Summarize sample information relevant to the hypothesis test Depends on distribution ( e.g : z, t etc ) Represents observed difference Critical value Fixed from a statistical distribution Determine rejection region for the null hypothesis Depends on α & test type Represents boundary beyond which the null hypothesis is rejected 8/20/2025 Dr. Shilpi Rani Biswas 37

38. Step 5: Conclusion/Decision making Compare the calculated test statistic & critical value Make a decision about null hypothesis & draw a conclusion about the population parameter 8/20/2025 Dr. Shilpi Rani Biswas 38

39. Question The average height of all residents in a city is 168cm with a population standard deviation of 3.9. A doctor believes mean to be different. He measured the height of 36 residents & found the average of 169.5cm. State the null & alternate hypothesis At confidence interval of 95%, is there enough evidence to reject the null hypothesis. Given – Population mean - 168 cm Sample mean – 169.5 cm Level of significance – 5%(0.05) Standard deviation – 3.9 Sample size - 36 8/20/2025 Dr. Shilpi Rani Biswas 39

40. Step 1 - State the hypothesis H - There is no significant difference between height of sample & population μ = 168 cm H 1 - There is significant difference between height of sample & population μ ≠ 168 cm Step 2 : Set up a level of significance C.I- 95% , α – 5%(0.05) 8/20/2025 Dr. Shilpi Rani Biswas 40

41. Step 3 : Select a statistical test Mean – Parametric test Z test or T test ? Check for sample size n<30 – T test n>30- Z test (36) 8/20/2025 Dr. Shilpi Rani Biswas 41

42. Step 4 : Perform computations Z value/score =  Difference in mean/ Standard error of mean = 169.5-168/0.65 = 2.3 (calculated test standard) SEM = SD/√N = 3.9/ √ 36 = 0.65 8/20/2025 Dr. Shilpi Rani Biswas 42

43. Determine Critical value & fix decision boundary 8/20/2025 Dr. Shilpi Rani Biswas 43

44. Step 5: Conclusion/Decision making Compare the test static calculated & critical value Make a decision about null hypothesis & draw a conclusion about the population parameter 8/20/2025 Dr. Shilpi Rani Biswas 44

45. Test standard – 2.3 falls under area of rejection or p < α Reject the null hypothesis Draw a conclusion “ what doctor believed was true” 8/20/2025 Dr. Shilpi Rani Biswas 45

46. Further reading Difference b/w bivariate analysis & multivariate analysis Significance of difference in mean 8/20/2025 Dr. Shilpi Rani Biswas 46

47. Summary Hypothesis is tentative or expected prediction or explanation of the relationship b/w two or more variables in a specific population Types of hypothesis: simple, complex, null, alternate, directional, non-directional, associative & causal H 0 - no effect or no difference between variables/parameters H A- an effect or difference exists between variables/parameters Steps of hypothesis testing 8/20/2025 Dr. Shilpi Rani Biswas 47

48. Specific learning objectives achieved By the end of this teaching-learning session, the co-learner are able to: Define important terms related to hypothesis testing Differentiate b/w research question & research hypothesis Enumerate steps of hypothesis testing Differentiate b/w type 1 & type 2 error 8/20/2025 Dr. Shilpi Rani Biswas 48

49. References Sarkar S, Srivastava V, Mohanty M. Postgraduate PHARMACOLOGY. 2 nd ed. Paras Medical Publisher; 2024 Ranganathan P, Cs P. An Introduction to Statistics: Understanding Hypothesis Testing and Statistical Errors. Indian J Crit Care Med. 2019 Sep;23( Suppl 3):S230-S231 Christian L, Jeff C, Munro C Matthew E Peters  Nyaz D Journal of Neurology, Neurosurgery & Psychiatry   2020;( 91)   586-592 Banerjee B. Mahajan’s Methods in Biostatistics for Medical Students and Research Workers. 9th ed. New Delhi: Jaypee Brothers; 2018. p. 1–10. Greenland S, Senn SJ, Rothman KJ, Carlin JB, Poole C, Goodman SN, Altman DG. Statistical tests, P values, confidence intervals, and power: a guide to misinterpretations. Eur J Epidemiol . 2016 Apr;31(4):337-50 8/20/2025 Dr. Shilpi Rani Biswas 49

50. Audience questions 8/20/2025 Dr. Shilpi Rani Biswas 50

51. Questions Enumerate types of hypothesis What are sources of hypothesis ? What do you understand by null & alternate hypothesis ? What do you understand by level of significance ? Differentiate b/w research question & research hypothesis 8/20/2025 Dr. Shilpi Rani Biswas 51

52. THANK YOU 8/20/2025 Dr. Shilpi Rani Biswas 52