You are on page 1of 14

STATISTICS

CRITICAL JOURNAL REVIEW


“Hypothesis Test"

Lecturer: Freddy Tua Musa Panggabean, S.Pd., M.Pd. 


 

BY: GROUP 2

CINTA JOHANNA TESSALONIKA PASARIBU       4191131007


INDAH MUTIARA INSANI                                       4193332001
MHD. SHOLEH KURNIAWAN NST                        4193332002
PALAGUNA SIAHAAN             4193131017
THERESIA O AMBARITA                                       4193131003

BILINGUAL CHEMISTRY EDUCATION STUDY PROGRAM


FACULTY OF MATHEMATICS AND NATURAL SCIENCES
STATE UNIVERSITY OF MEDAN
2020
PREFACE

Praise to the Almighty God for His blessing and mercy, so that the authors can
accomplish this "Critical Journal Review" well and on time. This Critical Journal Review is
made to fulfill the Statistics assignment.
 The authors realizes that the Critical Journal Review is still far from the word perfect, as
there are still many shortcomings in its stacking. Nevertheless, the authors has tried to the fullest
extent possible with all the ability to complete this paper properly. Therefore, the criticism and
suggestion of constructive nature are expected of the readers to perfection this Critical Journal
Review.     
 At the end of the word, we thank you and hopefully this Critical Journal Review is
beneficial and can provide additional science and knowledge for the readers. 

Medan, October 2020

Group II
TABLE OF CONTENTS 
CHAPTER I
INTRODUCTION
Identity Of Journal
  Title of Journal : Statistical Hypothesis Testing: Overview and Application
 Name of Journal : Journal of Head and Face Pain
 Publisher : American Headache Society 
 Author : Turner, D. P., Deng, H., & Houle
 Year of publication : 2020
 Volume : 60
 ISSN : 0017 - 8784
CHAPTER II
SUMMARY OF THE ARTICLE

Statistical inference refers to those methods that allow the estimation of population
properties from observed samples. Very often, this process takes the form of formal hypothesis
testing. Although there are many ways a researcher could investigate a hypothesis, in medical
research, by far the most common is through the use of some form of statistical hypothesis
testing. Statistical hypothesis testing is a set of methods for statistical inference that has a
fascinating and contentious history.
A famous debate raged for decades between the early creators of these methods about the
proper application of the emerging technique that would eventually become the most popular
tool for statistical inference. The methods most commonly used today are a blend between the
«significance test» developed by Fisher and the «hypothesis test» developed by Neyman and
Pearson. Although modern application of statistical hypothesis testing has evolved over time,
perhaps tending toward the approaches advocated by Neyman and Pearson, a thorough
understanding of the principles of significance-based statistical hypothesis testing is crucial for
investigators, 0consumers of research, and even for the growing number of individuals who wish
to abandon the use of any hypothesis testing based on these principles.

DEFINING THE ISSUE

Most people who have read a headache research article have seen the signs that
significance-based hypothesis testing has been conducted. The use of P values, the term
«statistically significant,» and the array of statistical tests all convey that investigators are testing
a hypothesis to make an inference about some population. 12 This reasoning is at first
counterintuitive but becomes clearer when the idea of the null hypothesis is fully understood.
The Null Hypothesis.
This hypothesis, denoted H0, typically represents «no effect» or «no relationship»
between the variables under study. The null hypothesis is considered to be true until evidence is
found to reject it or falsify it. This method allows the estimation of the probability of observing a
test result at least as extreme as the test statistic, assuming that the null hypothesis is true. In
practice, a statistical test must be chosen that adequately describes the data under study.
Although the choice about which test to use depends on the nature of the data being analyzed,
each of these tests produces a test statistic.
Assuming that the null hypothesis is true, the average difference between the groups
should be small in relation to the variability of headache frequency within each group. Using a t-
test, a test statistic can be computed to estimate the probability, commonly known as a P value,
of observing results at least as extreme as the ones observed in the sample if the null hypothesis
were true. A key element of the NP approach is that it focuses on making an inferential decision
in regard to a hypothesis test with careful consideration of the anticipated effect under study and
the costs of making an inferential error. This approach simultaneously considers 2 competing
hypotheses, the null hypothesis and an alternative hypothesis. As before, the null hypothesis
posits that there are no differences or effects. However, the alternative hypothesis posits the
existence of an effect with a very specific effect size, or degree of difference.
STATISTICAL HYPOTHESIS TESTING IN HEADACHE RESEARCH

The use of the term «significant» along with P values and statistical tests appears in
nearly every published article that includes testing of a hypothesis. However, there have been
recent advances in modeling data that utilize alternative approaches to conducting statistical
inferences as opposed to simply testing null hypotheses using significance-based hypothesis
testing. One particular paper focused on applying these methods to test a hypothesis.

COMMON PROBLEMS IN STATISTICAL HYPOTHESIS TESTING

While the above descriptions cover the basic elements of statistical hypothesis testing,
many additional details are often involved in the application of these methods. Below are several
frequently encountered problems in statistical hypothesis testing. To Specify a Testable
Hypothesis. A testable hypothesis could be, «we hypothesize that there is a direct correlation
between headache frequency and disability.» This statement implies that the authors expect r >
and will test this correlation against the null hypothesis of r =.
This would certainly be even clearer if the null hypothesis were explicitly articulated
rather than merely implied, but this is seldom observed in current medical publication practice.
Crafting a well-conceived and clearly defined hypothesis testing framework can help authors to
properly consider many important study design-related issues before collecting data, especially
in the NP testing framework. Further, clearly specifying a testable hypothesis allows
straightforward interpretation of results and can help avoid post hoc revisions of the core
hypothesis that can lead to inflated chances of inferential errors. Or Unknown Statistical Power.
In NHST, statistical power is the probability that the null hypothesis will be rejected,
conditional on a defined effect size. Though this is a critical consideration, researchers do not
always give adequate attention to statistical power. When statistical power is low or unknown,
interpreting results, especially those that are statistically nonsignificant, is challenging because it
is not possible to distinguish whether there is a lack of an effect or whether there is a lack of
statistical power to detect an effect. In fact, when a hypothesis test is conducted with a specific
sample size, the existence of a specific effect size is tested, whether or not the researcher knows
this effect size.
The 2 groups are very similar, with 9.0 vs 9.01 days/month, but because of the large
sample size, the P value from a t-test is sufficiently low to reject the null hypothesis, P =. Small
Sample Sizes Make it Difficult to Reject the Null Hypothesis. The 2 groups are quite different,
with 8.9 vs 2.3 days/month, but because of the small sample size, the P value from a t-test is not
sufficiently low to reject the null hypothesis, P =. With small sample sizes, it is difficult to
interpret null hypothesis testing without a careful consideration of the statistical power.
Though often assumed, a P value is not the probability that the null hypothesis is true. 28
The P value really indicates the probability of observing an effect as extreme or more extreme
than the actually observed effect if the null hypothesis were true. 03 indicates that there is only a
3% chance of observing results as extreme or more extreme than the actually observed findings
if the null hypothesis of no association were in fact true. Of «Significance».
Because simply rejecting the null hypothesis does not necessarily imply that the findings
are important, the interpretation of NHST must be made in conjunction with the size of the group
differences or size of the effect under study. Stated differently, a statistically significant finding
is a statement about the improbability of the null hypothesis, but it is not a statement about the
clinical or practical significance of an effect. She designs a placebo-controlled trial and
establishes the null hypothesis as new treatment = placebo and sets the statistical significance
threshold as P <. 001, so we would conclude that this effect is statistically significant, and we
would reject the null hypothesis.
However, this «significant» finding has very little practical or clinical significance
because any treatment that does not destroy the entire lice colony does not actually rid a person
of the condition. Merely finding statistical significance does not necessarily imply practical or
clinical significance of any kind. To Consider Uncertainty in Estimates. 05 indicates that the
observed differences are not consistent with the null hypothesis, it does not tell us the size of the
effect, the direction of the effect, or the uncertainty around any observed effect size.
This wide interval is inconsistent with an OR = 1.0 but suggests considerable uncertainty
in the estimate. A much larger sample size for the same estimate will reduce the uncertainty, but
this uncertainty is not well expressed using only a P value. Thus, it is important to fully report
the sample sizes, descriptive statistics, and relevant uncertainty estimates for all results.
CHAPTER III
EXCESS OF THE ARTICLE

In this journal, there are several advantages, namely in this journal the researcher
discusses the use of statistics in drawing conclusions on headache problems with the t-test. The
explanation given by the researcher was also not at length but directly explained the data
processing or research implementation from beginning to end. This journal describes various
methods of calculation for quantitative data processing. The researcher also provides sources
from experts, applies the suggested methods, uses many references to support the arguments he
gives. The researchers also show us some of the formulas they use and give us a table of their
research. The researchers managed to get the goal of what he wanted in writing this journal. The
linkages presented also make it easier for readers to understand the contents of the journal. With
the completeness and thoroughness of this discussion, this journal has good expertise in
providing an understanding of the use of statistics and its application to produce appropriate
conclusions in research.
CHAPTER IV
WEAKNESS OF ARTICLES / RESEARCH RESULTS

From this journal, there are several weaknesses. There are no pictures related to the topic
to make it easier for readers to understand the journal. The author does not provide pictorial
illustrations of the steps in examples of types of approaches, so readers can only understand if
working on the data directly either on the null hypothesis approach or on common problems in
statistical hypothesis testing. The author does not provide examples on the Null Hypothesis
Significance Testing to make it easier for readers to understand the types of approaches to make
decisions about the null hypothesis.
CHAPTER V
IMPLICATIONS FOR DEVELOPMENT IN INDONESIA

Hypothesis testing is a decision-making method based on data analysis, both from


controlled experiments and from observations (uncontrolled). In statistical terms a result can be
said to be statistically significant if the event is almost impossible to cause by chance, within the
predetermined probability limits.
Hypotheses are conjectures that may or may not be right. He will be rejected if it is false
or false, and will be accepted if the factors correct him. Rejection and acceptance of hypotheses,
therefore, are very dependent on the results of the investigation of other factors.
To determine the error of the hypothesis results on the population value. Determine the
initial hypothesis / assumption that can be accepted or not from a study.
CHAPTER VI
CONCLUSIONS AND RECOMMENDATIONS

NHST is an amalgam method of statistical inference that arose out of the methods
proposed by Fisher and by Neyman and Pearson. Due to these limitations and the widespread
misuse of NHST, a growing chorus of critics has called for fundamental changes or abandoning
NHST entirely. Thus, a complete understanding of NHST requires also understanding the
arguments of those who wish to abolish its use. Confusion is generated by selectively utilizing
elements of both approaches simultaneously, and critics have argued that the combined approach
lacks coherence in underlying philosophy.
Additional critics dislike the approach on different grounds. These critics, who note that
the P <. 05 threshold has led to many false positives in the literature, favor reducing the
statistical significance threshold to P <. 005.32 There are other critics who favor completely
abandoning the use of null hypothesis testing.
It seems unlikely that in the near future NHST will be dislodged from its place as the
most frequent tool for statistical inference.
REFERENCE

Turner, D. P., Deng, H., &amp; Houle, T. T. (2020). Statistical Hypothesis Testing: Overview
and 
Application. Headache: The Journal of Head and Face Pain, 60(2), 302-308.
doi:10.1111/head.13706.
ATTACHMENT 

You might also like