Understanding the p-value

Understanding the p-value in Statistical Analysis

Introduction

In the realm of statistics, the p-value has become a cornerstone of hypothesis testing and inferential statistics. It serves as a critical metric that helps researchers draw conclusions about data and make informed decisions in various fields, from medicine to social sciences. This document will explore the p-value's definition, its significance in hypothesis testing, how to interpret it correctly, its limitations, and common misconceptions surrounding it.

1. What is a p-value?

The p-value, or probability value, is a measure that helps quantify the evidence against a null hypothesis in statistical tests. Formally, it represents the probability of observing a test statistic at least as extreme as the one computed from the sample data, given that the null hypothesis is true. In simpler terms, it indicates how compatible the sample data is with the null hypothesis.

Mathematically, if we denote the null hypothesis as $H_0$ and the alternative hypothesis as $H_a$, the p-value can be defined as:

$$p = P(\text{data} | H_0)$$

Where:

• $P(\text{data} | H_0)$ is the probability of obtaining the observed data (or something more extreme) under the assumption that the null hypothesis is true.

2. Importance of p-values in Hypothesis Testing

In hypothesis testing, researchers typically start with two competing hypotheses: the null hypothesis ($H_0$) and the alternative hypothesis ($H_a$). The null hypothesis usually states that there is no effect or no difference, while the alternative hypothesis suggests that there is an effect or a difference.

The steps in hypothesis testing generally involve:

1. Formulating Hypotheses: Stating the null and alternative hypotheses.
2. Choosing a Significance Level ($\alpha$): This threshold is often set at 0.05, meaning that a p-value less than 0.05 will lead to rejection of the null hypothesis.
3. Collecting Data: Performing experiments or observational studies to gather data.
4. Calculating the Test Statistic: Using statistical formulas to compute a test statistic (e.g., t-statistic, z-statistic) based on the collected data.
5. Calculating the p-value: Determining the p-value associated with the calculated test statistic.
6. Making a Decision: Comparing the p-value to the significance level $\alpha$ to decide whether to reject or fail to reject the null hypothesis.

3. How to Interpret p-values

Interpreting p-values correctly is crucial for making valid conclusions from statistical analyses. Here are some key points about interpreting p-values:

• Small p-value ($p < \alpha$): This suggests strong evidence against the null hypothesis, leading researchers to reject $H_0$. For instance, if a study reports a p-value of 0.03 and $\alpha$ is set at 0.05, the researchers would reject the null hypothesis, suggesting that there is a statistically significant effect.

• Large p-value ($p \geq \alpha$): This indicates insufficient evidence to reject the null hypothesis. However, it does not prove that $H_0$ is true; it simply indicates that there is not enough evidence to support $H_a$.

• p-value does not measure the size of an effect: A small p-value indicates statistical significance, but it does not provide information on the magnitude or practical significance of the effect. For example, a study may report a p-value of 0.001 for a new drug, indicating statistical significance, but the actual effect size may be minimal and not clinically relevant.

4. Common Misconceptions about p-values

Despite their widespread use, p-values are often misunderstood. Here are some common misconceptions:

4.1. A p-value is the probability that the null hypothesis is true.

This statement is incorrect. The p-value is the probability of obtaining the observed data (or something more extreme) given that the null hypothesis is true. It does not provide a direct probability of the null hypothesis itself.

4.2. A p-value of 0.05 is a definitive threshold.

While 0.05 is a common significance level, it is not a hard and fast rule. Different fields of study and research questions may warrant different thresholds. Moreover, reliance on a strict cutoff can lead to arbitrary conclusions, known as the “threshold effect.”

4.3. p-values can be used to prove or disprove a hypothesis.

p-values do not provide evidence to prove a hypothesis; they merely offer evidence to support or reject it. A p-value can suggest whether the data aligns with the null hypothesis, but it does not confirm the truth of any hypothesis.

5. Examples of p-values in Practice

To illustrate the concept of p-values, let's explore a couple of practical examples.

Example 1: Clinical Trial

Imagine a clinical trial testing a new medication for hypertension. The null hypothesis states that the medication has no effect on blood pressure compared to a placebo. After conducting the trial, the researchers find a p-value of 0.02.

• Interpretation: Since 0.02 is less than the common significance level of 0.05, the researchers reject the null hypothesis. This suggests that there is statistically significant evidence to support that the new medication has an effect on lowering blood pressure.

Example 2: Educational Intervention

In an educational study, researchers investigate whether a new teaching method improves student test scores compared to a traditional approach. They set up a controlled experiment and calculate a p-value of 0.15.

• Interpretation: In this case, the p-value of 0.15 exceeds the significance level of 0.05, leading to the conclusion that there is insufficient evidence to reject the null hypothesis. The new teaching method does not show statistically significant improvement over the traditional method based on this study.

6. Limitations of p-values

While p-values are valuable tools in statistical analysis, they have several limitations that researchers should be aware of:

6.1. Misinterpretation of results

As mentioned earlier, p-values are often misinterpreted. A common mistake is equating a p-value of 0.05 with a “true” result, whereas it merely indicates statistical significance.

6.2. Dependence on sample size

P-values are sensitive to sample size. In large samples, even trivial effects can yield small p-values, while in small samples, substantial effects may not achieve significance. This phenomenon can lead to misleading conclusions.

6.3. Lack of context

p-values alone do not provide the full context needed for informed decision-making. Researchers should complement p-values with effect sizes, confidence intervals, and additional context about the study design and implications.

7. Moving Beyond p-values

Due to the limitations and misconceptions associated with p-values, many statisticians advocate for a more holistic approach to data analysis. Here are some alternative strategies:

7.1. Effect sizes

Calculating effect sizes provides valuable information about the magnitude of an effect, offering a more nuanced view beyond mere statistical significance. Effect sizes can help interpret the practical significance of findings.

7.2. Confidence intervals

Confidence intervals (CIs) provide a range of values within which the true population parameter is likely to fall. CIs can help convey the uncertainty surrounding an estimate, offering more context than a p-value alone.

7.3. Bayesian statistics

Bayesian approaches provide a framework for updating the probability of a hypothesis based on new evidence. This method can offer a more intuitive understanding of uncertainty and the likelihood of different hypotheses.

8. Conclusion

The p-value remains a fundamental concept in statistical analysis, particularly in hypothesis testing. While it provides a measure of evidence against the null hypothesis, it is essential to interpret it correctly and be aware of its limitations. Researchers should complement p-values with effect sizes, confidence intervals, and alternative statistical approaches to gain a comprehensive understanding of their data.

By fostering a more nuanced view of statistical significance, researchers can enhance the robustness of their findings and contribute to more informed decision-making in their respective fields.

Interactive p-value Visualizer

Observed Z-score: 0.00

Two-tailed Right-tailed (> Z) Left-tailed (< Z)

Monday, April 10, 2023

How to find hypothesis testing and how to know which test to use?

 How to find hypothesis testing and how to know which test to use?

We generally start with the null hypothesis.

Null hypothesis means that there is no difference in the reality. 

For the sake of example suppose you have two medicines: Drug A, and Drug B. 

Now, we want to know which drug is better! 

This is our research question. To know this we have to frame our hypothesis: 

Now its consensus among scientists that its easier to reject the null hypothesis than to prove the alternate hypothesis. The reason of this is Suppose you have to prove that every cow is black in this world. Now you can disprove this simply on observing one white cow! this is rejection of null hypothesis. 

But if will test every cow for whether its black or white 

PSM / COMMUNITY MEDICINE by Dr Abhishek Jaiswal is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
Based on a work at learnpsm@blogspot.com.
Permissions beyond the scope of this license may be available at jaiswal.fph@gmail.com.

Friday, August 12, 2022

Behaviour Codes in Cancer

Morphology: 

The morphology code records the type of cell that has become neoplastic and its biologic activity; in other words, it records the kind of tumor that has developed and how it behaves. There are three parts to a complete morphology code:

4 digits cell type (histology)

1 digit behavior

1 digit grade, differentiation or phenotype

In ICD-O morphology codes, a common root codes the cell type of a given tumor, while an additional digit codes the behavior. The grade, differentiation, or phenotype code provides supplementary information about the tumor.

Behavior: 

The behavior of a tumor is the way it acts within the body. Pathologists use a variety of observations to determine the behavior of a tumor.

A tumor can grow in place without the potential for spread (/0, benign); it can be malignant but still growing in place (/2, noninvasive or in situ); it can invade surrounding tissues (/3, malignant, primary site); or even disseminate from its point of origin and begin to grow at another site (/6, metastatic).

5th Digit Behavior Code for Neoplasms

Code	Neoplasm
/0	Benign
/1	Uncertain whether benign or malignant
	Borderline malignancy
	Low malignant potential
	Uncertain malignant potential
/2	Carcinoma in situ
	Intraepithelial
	Noninfiltrating
	Noninvasive
/3	Malignant, primary site
/6*	Malignant, metastatic site
	Malignant, secondary site
/9*	Malignant, uncertain whether primary or metastatic site

* Not used by cancer registries (used by some pathologists in some parts of the world)

PSM / COMMUNITY MEDICINE by Dr Abhishek Jaiswal is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
Based on a work at learnpsm@blogspot.com.
Permissions beyond the scope of this license may be available at jaiswal.fph@gmail.com.

Cancer related organizations

The Surveillance, Epidemiology and End Results (SEER) Program is a premier source for cancer statistics in the United States. The SEER Program collects information on incidence, prevalence and survival from specific geographic areas representing 28 percent of the US population and reports on all these data plus cancer mortality data for the entire country. 

PSM / COMMUNITY MEDICINE by Dr Abhishek Jaiswal is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
Based on a work at learnpsm@blogspot.com.
Permissions beyond the scope of this license may be available at jaiswal.fph@gmail.com.

Thursday, August 4, 2022

Public Health days

World Braille Day	4 January
World Leprosy Day	Last Sunday in January
World Cancer Day	4 February
International Day of Zero Tolerance for Female Genital Mutilation	6 February
National Deworming Day	10 February
International Epilepsy Day	second Monday of February
International Day of Women and Girls in Science	11 February
World Unani Day	11 February
Sexual Reproductive Health Awareness Day	12 February
International Childhood Cancer Day	15 February
World Day of Social Justice	20 February
Rare Disease Day	28/29 February
International Condom Day	13 February
Zero Discrimination Day	01Mar
World Birth Defect Day	03Mar
World Hearing Day	03Mar
World Obesity Day	04Mar
No Smoking Day	Second Wednesday of March
World Glaucoma Day	12Mar
World Glaucoma Week	March 713 March 1218
International Women’s Day	08Mar
World kidney day	14Mar Second Thursday of March
Salt awareness week	11
Measles Immunization Day	March 16
World Kidney Day	Second Thursday in March
International Day of Happiness	20Mar
World Oral Health Day	20Mar
International Day for the Elimination of Racial Discrimination	21Mar
World Down Syndrome Day	21Mar
World Water Day	22Mar
World Tuberculosis Day	24Mar
International Day of Solidarity with Detained and Missing Staff Members	25Mar
World Autism Awareness Day	02Apr
International Day of Sport for Development and Peace	06Apr
World Health Day	07Apr
World health worker week	4
World Chagas Disease Day	14Apr
World haemophilia day	17Apr
World Liver Day	April 19
World Creativity and Innovation Day	21Apr
Earth Day	April 22
World Meningitis Day	24Apr
World Malaria Day	25Apr
World Immunization Week	The last week of April
World Day for Safety and Health at Work	28Apr
World Asthma Day	First Tuesday in the month of May
World Hand Hygiene Day	05May
International Day of the Midwife	05May
UN Global Road Safety Week	6 May
World Red Cross Day	8 May
World Thalassaemia Day	8 May
Mother's Day	Second Sunday of May
International Nurses Day	12May
International Day of Families	15May
National Dengue Day	May 16
International Day of action for women’s health	18May
World Family Doctor Day	19May
International Day against Homophobia, Transphobia and Biphobia	17May
World Hypertension Day	17May
International Day to End Obstetric Fistula	23May
World Multiple Sclerosis Day	May 25 (Last Wednesday of May)
Menstrual Hygiene Day	28May
International Day of Action for Women’s Health / International Women’s Health Day	28 May
World No Tobacco Day	31 May
World Environment Day	June 5
World Brain Tumour Day	June 8
World Blood Donor Day	14Jun
World Elder Abuse Awareness Day	15Jun
Autistic Pride Day	June 18
International Day for the Elimination of Sexual Violence in Conflict	19Jun
International Day of Yoga	21Jun
International Day Against Drug Abuse and Illicit Trafficking	26Jun
National Doctors Day	July 1
World Population Day	11Jul
World Brain Day	22Jul
World Drowning Prevention Day	25 July
World Hepatitis Day	28Jul
ORS Day	July 29
World Breastfeeding Week	1 to 7 August
World’s Indigenous People Day	09Aug
International Youth Day	12Aug
World Humanitarian Day	19Aug
World Mosquito Day	August 20
African Traditional Medicine Day	31Aug
National Eye Donation Fortnight	25th August  8th September
National Nutrition week	September 1 to 7
Spinal Cord Injury Day	September 5
World Physical Therapy Day	08Sep
World Suicide Prevention Day	10Sep
World Sepsis Day	13Sep
World Marrow Donor Day	September 16
World Patient Safety Day	September 17
World Alzheimer’s Day	21Sep
World Pharmacists Day	25Sep
World Lung Day	25Sep
World Rabies Day	28Sep
World Heart Day	29Sep
World Day of Deaf	Last Sunday of September
World Contraception Day	26Sep
Breast Cancer Awareness Month	October
International Day for the Elderly	1 October
World Vegetarian Day	October 1
International Day of Non-Violence	2 October
National Anti-Drug Addiction Day	October 2
World Sight Day	October 9
World Mental Health Day	10Oct
International Day of the Girl Child	11Oct
World Cerebral Palsy Day	First Wednesday of October
International Day for Disaster Reduction	13 October
World Thrombosis Day	13 October
Global Handwashing Day	October 15
World Sight Day	Second Thursday of October
World Hospice and Palliative Care Day	The second Saturday of October
World Arthritis Day	October 12
International Day of Rural Women	15 October
World Food Day	16 October
World Trauma Day	October 17
World Statistics Day	20Oct
World Osteoporosis Day	20 October
World Iodine Deficiency Day	October 21
United Nations Day	24 October
World Polio Day	24 October
World Obesity Day	October 26
World Psoriasis Day	29 October
World Stroke Day	29 October
World Thrift Day	October 30
World Cities Day	31 October
One Health Day	3 Nov
World Immunisation Day	November 10
World Pneumonia Day	12 November
World Antibiotic Awareness Week	18-24 Nov
World Diabetes Day	14 November
International Day for Tolerance	16 November
National Epilepsy Day	November 17
World COPD Day	19 November
World Toilet Day	19 November
World Day of Remembrance for Road Traffic Victims	The third Sunday of November
World Day of Research for Health	18 November
New Born Care Week	November 15-21
Universal Children’s Day	20 Nov
International Day for the Elimination of Violence against Women	25 Nov
World AIDS Day	1 December
National Pollution Prevention Day	December 2
International Day of Persons with Disabilities	3 December
International Volunteer Day for Economic and Social Development	5 December
FCHV Day	05Dec
World Patient Safety Day	9 December
International Anti-Corruption Day	9 December
Human Rights Day	10 December
International Universal Health Coverage Day	12 December
International Human Solidarity Day	20 Dec

Subscribe to: Posts (Atom)