Hypothesis Testing

 

Hypotheses are scientific falsifiable statements (Chung & Hyland, 2012) that are usually written in pairs, the null and the research hypotheses. The null, designated as H₀, is when there is no effect between the populations. The research hypothesis, designated as H₁, is when there’s an effect between the populations (Dancey & Reidy, 2017).

The research hypothesis can be one-directional or one tailed, as in there is a directional relationship between populations. It can be non-directional or two tailed hypothesis (Bruin, 2006). The major difference between experimental hypotheses and research hypotheses in correlational research is that experiments are always directional. Experimental hypotheses attempt to demonstrate causal and effect between independent and dependent variables. Whereas in correlational research, hypotheses observe relationships among variables, thus can be one or bi-directional, since they can also be descriptive (Dancey & Reidy, 2017). 

Research hypotheses are tested by trying to disprove their respective null hypotheses by providing quantitative evidence, through inferential statistics (SJSU, 2016). Inferential statistics is a statistical method used to make inferences about a population based on data taken from a random sample of a population (Minitab , 2017). A form of statistical inference that is used to determine the probability that the null hypothesis is correct, despite evidence that support the research hypothesis, is the null hypothesis significance test (NHST) . The result of the NHST is determined by the sample size and the binomial parameter, and expressed as a probability (p-value) in percentage or decimal. In psychology, the result of a study is accepted if the level of probability that the null hypothesis is correct is less than 5%, and expressed as P<0.5 , also known as level of statistical significance (Dancey & Reidy, 2017).

The rationale behind setting the level of statistical significance at P<0.5 has to do with what the scientific community perceives as acceptable level of error occurrence (Dancey & Reidy, 2017).  There are two types of error that can occur when taking NHST into account. The fist is called a type 1 error, which is when the research rejects the null hypothesis when it’s true. The second is called type 2 error, which is when the researcher accepts the null hypothesis when it is wrong. When P<0.5 the probability of type 1 error is less than 5%, and when P>0.5, the probability of type 2 error is less than 5%. Therefore, 5% is chosen as a balanced probability that tolerates the occurrence of both errors (Minitab, 2017). However, in the medical field, the tolerance for the occurrence of type 1 error is 1% (p<0.01). The tolerance for type 1 error is low because human life is at stake (Dahiru, 2008).

One major pitfall of NHST is that it’s not comparable and cumulative, whereas scientific research is (SJSU, 2016). Another is that psychological significance is determined by the level of effect, while NHST is mistaken for psychological significance, which is known as the permanent illusion (Cohen, 1994). Type 1 and type 2 errors are also pitfalls. However, all pitfalls can be accounted for with sound research design and research replication (Dancey & Reidy, 2017).

References

Bruin, J. (2006). Institute for digital research and education . Retrieved from University of Califirnia Los Angeles : https://stats.idre.ucla.edu/other/mult-pkg/faq/general/faq-what-are-the-differences-between-one-tailed-and-two-tailed-tests/

Chung, M. C., & Hyland, M. E. (2012). Evaluation of the idea that psychology is a science: what is science ? In M. C. Chung, & M. E. Hyland, History and Philosophy of Psychology (pp. 76 - 79). West Sussex: John Wiley & Sons Incorporated.

Cohen, J. (1994). The earth is round (p <.05). American Psychologist , 997- 1003.

Dahiru, T. (2008). P – value, a true test of statistical significance? A cautionary note. Ann Ib Postgraduate Med, 21-26.

Dancey, C., & Reidy, J. (2017). Hypothesis testing and statistical significance . In Statistics Without Maths for Psychology (7th ed.) (pp. 134-173). Harlow, UK: Pearson.

Minitab . (2017). What are inferential statistics ? Retrieved from Minitab: https://support.minitab.com/en-us/minitab-express/1/help-and-how-to/basic-statistics/inference/supporting-topics/basics/what-are-inferential-statistics/

Minitab. (2017). What are type I and type II errors? . Retrieved from Minitab: https://support.minitab.com/en-us/minitab-express/1/help-and-how-to/basic-statistics/inference/supporting-topics/basics/type-i-and-type-ii-error/

SJSU. (2016, May 8). Introduction to null hypothesis significance testing. Retrieved from San Jose State University: http://www.sjsu.edu/faculty/gerstman/StatPrimer/hyp-test.pdf