Stats & Methodology

# Null Hypothesis | Statistics

Go to

## Null Hypothesis | Statistics

A null hypothesis is essential to scientific inquiry since it forms the basis for research inquiries. In addition to giving researchers a place to start, it enables them to develop alternative hypotheses that may be put to the test and assessed.

### The Role of the Null Hypothesis

Examining the relationship between variables or figuring out whether there are variations between groups is frequently important in scientific studies. According to the null hypothesis, oftentimes there is no discernible difference or connection between the variables under investigation. It denotes the absence of a relationship between the relevant components or an effect between them.

Researchers create the null hypothesis to serve as a reference point for comparison of their findings. Usually represented by the symbol H0, this hypothesis serves as a benchmark for determining the statistical significance of the study’s findings.

### Example

Let’s use a study on the effects of a novel intervention program on non-specific neck pain as an example to further clarify this idea. There may not be a discernible difference in neck pain between patients who receive the intervention and those who do not, according to the null hypothesis in this situation.

Thus, the null hypothesis can be written mathematically as H0: 1 – 2 = 0, where 1 is the average neck pain of those who get the intervention, and 2 is the average neck pain of those who do not.

After gathering information, researchers run statistical tests to see if the evidence confirms or refutes the null hypothesis. Researchers may reject the null hypothesis in favor of an alternate hypothesis if the data contradict the null hypothesis and show a significant difference or link.

### Don’t Do This

It is significant to highlight that the alternative hypothesis is not demonstrated by the null hypothesis’ rejection. Instead, it argues that the data might suggest the alternative hypothesis to be considered a more plausible explanation. Usually, the alternative hypothesis claims that there is a distinction or connection between the relevant variables.

In real life, researchers analyze the likelihood of obtaining the reported results under the null hypothesis by using various statistical tests, such as t-tests or chi-square tests. Researchers reject the null hypothesis and explore the alternative hypothesis if the probability falls below a preset significance level, commonly denoted as alpha (α), most often 0.05.

The act of testing hypotheses is essential to scientific inquiry because it enables researchers to make judgments based on empirical data. Researchers can increase knowledge and contribute to the comprehension of many phenomena by methodically analyzing and questioning the null hypothesis.

### Problems with Null Hypothesis Testing

One significant criticism is that it frequently ignores effect sizes and clinical significance in favor of concentrating only on statistical significance. Statistical significance does not reveal the size or significance of the effect that was seen; it just tells whether a finding is likely to have happened by chance. Testing huge data sets can results in significant results (rejecting the H0) for the tiniest differences.

Let’s go further on the previous example. You are interested in the VAS (visual analog scale) for pain post-treatment for two interventions for neck pain. You have about 1000 patients per group. Group A has an average of 2.2/10 post-treatment, and group B 2.4/10. Given that the groups are so large, chances are that this tiny difference results in a significant difference with null-hypothesis testing. However, 0.2/10 of a difference is hardly relevant. In terms of clinical significance, these two groups are equal.

Another problem is that the null hypothesis might be rejected or accepted, which can lead to a binary interpretation of the data. This dichotomous approach could oversimplify complicated occurrences and miss the finer details of the data.

Furthermore, until disproven, null hypothesis testing assumes that the null hypothesis is true. This may lead to a bias in favor of the null hypothesis and lead to the possible blindness to potentially significant effects.

Opponents contend that alternative methods such as effect size reporting, or Bayesian statistics, can offer a more thorough and instructive examination of study results, enabling a better grasp of the relevance and practical consequences of the findings.

### Summary

The null hypothesis, which states that there is no significant difference or association between the variables of interest, acts as the default assumption in a research investigation. In order to reject the null hypothesis in favor of an alternative one, the data should be incompatible with the null hypothesis, showing a significant difference. Scientists can improve hypotheses, investigate novel concepts, and deepen our understanding of the world through hypothesis testing. However, there exists a large group of critics against null hypothesis testing. It is not without its flaws.

### References

Neyman, J., & Pearson, E. S. (1933). On the problems of the most efficient tests of statistical hypotheses. Philosophical Transactions of the Royal Society of London, 231A, 289–338. https://doi.org/10.1098/rsta.1933.0009

GRAVES, SPENCER (1978). On the Neyman–Pearson Theory of Testing. The British Journal for the Philosophy of Science, 29(1), 1–24. doi:10.1093/bjps/29.1.1

Quintana, D.S., Williams, D.R. Bayesian alternatives for common null-hypothesis significance tests in psychiatry: a non-technical guide using JASP. BMC Psychiatry 18, 178 (2018). https://doi.org/10.1186/s12888-018-1761-4

Burnhamm and Anderson, P values are only an index to evidence: 20th- vs. 21st-century statistical science. Ecology (2014), https://doi.org/10.1890/13-1066.1

Like what you're learning?

#### Use the assessment app

• Over 300 orthopedic physical assessment tests
• Statistics, basic assessments, and screening tests included
• Direct links to PubMed references
• Concise test descriptions
• Video demonstration
• Easy search & favorites function
E-Book

Reviews