Volume 23, Issue 88 (5-2023)
refahj 2023, 23(88): 9-83 |
Back to browse issues page
Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
oreyzi H. (2023). Humpty Dumpty domination on correlational data analysis in social welfare researches. refahj. 23(88), 9-83. doi:10.32598/refahj.23.88.1908.5
URL: http://refahj.uswr.ac.ir/article-1-3991-en.html
URL: http://refahj.uswr.ac.ir/article-1-3991-en.html
Keywords: Private Language, Statistical error, Language misuse, Incorrect Language in hypothesis testing, Social welfare
Full-Text [PDF 989 kb]
(970 Downloads)
| Abstract (HTML) (1180 Views)
Full-Text: (695 Views)
Introduction
Identifying and correcting errors is essential to science, and it reminds the famous quote that science is self-correcting. While making mistakes was regarded as humane and was accepted in in philosophy of science, it especially gave rise to a glorious dictum which states “scientific hypothesis must have the chance to falsify”. In the current paper, the author pointed to errors in correlational studies, that may occur due to misusing of language. The author first mentioned the famous dialogue between Humpty dumpty and Alice in the Lewis Carrol seminal work, Alice in Wonderland. The private language that Wittgenstein introduced in his book “Tractaus” corresponded to Humpty Dumpty’s claim that “when I use a word it means just what I choose it to mean – neither more nor less”. Correlational studies suffer from a kind of private language that unequivocally seems to be incorrect from statistical point of view. It is formed by incorrect usage of language, in other words it is originated from some incorrect and illogical conclusions or from overgeneralization of relations between two variables separately which could not stand true in three variables or more variables in correlational studies. Loose designing of studies may lead to inappropriate conclusions especially when experimental or longitudinal research necessitates the forming of such conclusions. Besides, the author considered errors in statistical analysis which in some cases may lead to faulty analytic choices. Prevalence of errors in correlational studies intensified by misunderstanding of structural equation modelling, plus some kind of private language which may be lacking in terms of scientific logic. Erroneous results principally produce junk materials that entirely affect our knowledge. If scientists cannot produce reliable results, why should practitioners trust the scientific enterprise? The overwhelming majority of wrong- correlational studies with errors enter the body of research and without critic they welcomed to journals unfortunately. Also because of its accurate explanation many mathematical equations introduce core concept of essay.
Critical Review of References
Critical Review of articles essentially in observational research conclude thirteen texts in social welfare area. First some papers contain exploratory factor analysis claim hypothesis testing, while we must mention that only confirmatory analysis deserve to have hypothesis. If the items being loaded under one factor were not be able to distinguish the items under another factor, it is wrong to call them with different labels. The construct must be meaningful. It is very common to take place, and we cannot report the result of EFA. This error put in danger the validity of construct. Sometimes the variables with different concept are identical from psychometric point of view, for example their correlation coefficient is above 0.90. This error also put in danger the construct validity of research. Sometimes arguments in observational research contain logical fallacies, and hypothesis testing based on these arguments are incoherent. Some researchers overestimate the statistical software as being so powerful that can produce hypothesis. The tendency to apply stepwise regression analysis without hypothesis and giving the role of combination of variables to software is criticized. It is recommended to apply hierarchical regression analysis with certain theoretical foundation. The overestimation of software performance is preferred by researchers because of its simplicity, while the result is meaningless. In some research studies while the theoretical foundation required to do moderation analysis, researchers may move toward mediation analysis. One example is buffer hypothesis. Some researchers by wrongful translation of moderator conclude to apply mediator analysis. Sometimes researchers only follow relationships between x
y , x M and M y relationships and may miss the last equation in Barron Kenny (1986). It is argued that fit indexes are not enough to conclude mediation or some other casual relationships, because equivalent models confirmed simultaneously. The longitudinal or experimental design or strong theoretical basis must be present to reinforce and prefer one model over other equivalent models. Hierarchical linear models in many correlational studies is very necessary, while it is neglected inappropriately. The wrong mental model about different types of computer software is discussed, especially in structures equation modelling (SEM) such as LISREI, AMOS and EQS. Sometimes it is assumed that different types of software can make hypothesis without justifying literature and confirm them. It is like John Searl Chinese Man that produce instructions to translate without understanding them. Most research studies apply SEM without awareness, and this usually results in errors. Examples of common mistakes in correlational studies were reported to illuminate discussion. Almost all of these arguments were produced because of weak language for arguments and misunderstanding of statistical methods in correlational studies. Some papers put artificial statements in illusory findings of literature to justify their hypothesis. Moreover, based on the suggestions provided in the current study at least three paper must be retracted.
Discussion
Some Hindus bring an elephant to be exhibited in a darkroom. It is familiar to us narrated by Iranian poet Rumi. The condition of correlational studies in Iran is worse, the Hindus definition of elephants despite limitation of their perception was something like shotgun- approach while most papers concerning correlational statistical methods resemble much ado about nothing. Researchers use a variety of rationalizations to make sense of their insignificant research studies that have no value neither for practitioners nor for other researchers. Motivation to gather data in these research studies appears to be different. These researchers think they are contributing to science as a great and noble enterprise and satisfy their ambitions. Most correlational studies in social welfare investigated in this paper were utterly redundant. This prevalent opinion that academics would strive for excellence motivate them to write papers. The methodology of correlational studies in their mind was very easy, they conceptualize hypothesis like humpty Dumpty and produce statements that reflect in arbitrary diagrams and good fitness indexes contribute to confirm their pseudo-hypothesis, in spite of their illusory knowledge. This process is excellent by nonsense. These papers became mass product, the majority of which added nothing to scientific advancement and it should be noted that correlational study only uncover a relationship, it cannot provide a conclusive reason for why there is a relationship. Associations observed in correlational studies is foundation for generating hypothesis that can be tested in experimental or longitudinal studies. The experimental studies then can predict direction of effect in complicated configuration and confirmed by fit indexes. Equivalent models refute all researchers who claim their discourse justifications are sufficient, while they are not. The modern knowledge of correlational study is complex, but includes some aspects of sound simplicity. The pitfall of this gap deceive novice research. responsivity of peer review journals is significant. They must judge by rigor and accurate statistical and research method knowledge. The author of the current paper refers to a complex form of ordinary least square regression that is used to analyze variance in the outcome variables when the predictor variables at varying hierarchical levels namely Hierarchical Linear Modeling, was not apply in research studies that need this type of statistical analysis because of its complexity. Mass production of structural equation modelling in spite of inappropriateness in many papers seems to be simple (but it is not). Although structural equation modelling is a powerful, multivariate technique, misunderstanding of its scientific foundations especially of its heritage from the first generation (path analysis) and the second generation (factor analysis) may produce nonsense research studies being justified by incoherent line of speech.
.
Ethical considerations
Author Consideration
The author spent eight months to review all papers and their reference cited.
Funding
There was no financial sponsor for this critical review.
Conflicts of interest
This review does not overlap with other published works by the author.
Following the ethics of research
The equity principle of ethics was taken into account as building block of critical review .
Identifying and correcting errors is essential to science, and it reminds the famous quote that science is self-correcting. While making mistakes was regarded as humane and was accepted in in philosophy of science, it especially gave rise to a glorious dictum which states “scientific hypothesis must have the chance to falsify”. In the current paper, the author pointed to errors in correlational studies, that may occur due to misusing of language. The author first mentioned the famous dialogue between Humpty dumpty and Alice in the Lewis Carrol seminal work, Alice in Wonderland. The private language that Wittgenstein introduced in his book “Tractaus” corresponded to Humpty Dumpty’s claim that “when I use a word it means just what I choose it to mean – neither more nor less”. Correlational studies suffer from a kind of private language that unequivocally seems to be incorrect from statistical point of view. It is formed by incorrect usage of language, in other words it is originated from some incorrect and illogical conclusions or from overgeneralization of relations between two variables separately which could not stand true in three variables or more variables in correlational studies. Loose designing of studies may lead to inappropriate conclusions especially when experimental or longitudinal research necessitates the forming of such conclusions. Besides, the author considered errors in statistical analysis which in some cases may lead to faulty analytic choices. Prevalence of errors in correlational studies intensified by misunderstanding of structural equation modelling, plus some kind of private language which may be lacking in terms of scientific logic. Erroneous results principally produce junk materials that entirely affect our knowledge. If scientists cannot produce reliable results, why should practitioners trust the scientific enterprise? The overwhelming majority of wrong- correlational studies with errors enter the body of research and without critic they welcomed to journals unfortunately. Also because of its accurate explanation many mathematical equations introduce core concept of essay.
Critical Review of References
Critical Review of articles essentially in observational research conclude thirteen texts in social welfare area. First some papers contain exploratory factor analysis claim hypothesis testing, while we must mention that only confirmatory analysis deserve to have hypothesis. If the items being loaded under one factor were not be able to distinguish the items under another factor, it is wrong to call them with different labels. The construct must be meaningful. It is very common to take place, and we cannot report the result of EFA. This error put in danger the validity of construct. Sometimes the variables with different concept are identical from psychometric point of view, for example their correlation coefficient is above 0.90. This error also put in danger the construct validity of research. Sometimes arguments in observational research contain logical fallacies, and hypothesis testing based on these arguments are incoherent. Some researchers overestimate the statistical software as being so powerful that can produce hypothesis. The tendency to apply stepwise regression analysis without hypothesis and giving the role of combination of variables to software is criticized. It is recommended to apply hierarchical regression analysis with certain theoretical foundation. The overestimation of software performance is preferred by researchers because of its simplicity, while the result is meaningless. In some research studies while the theoretical foundation required to do moderation analysis, researchers may move toward mediation analysis. One example is buffer hypothesis. Some researchers by wrongful translation of moderator conclude to apply mediator analysis. Sometimes researchers only follow relationships between x
Discussion
Some Hindus bring an elephant to be exhibited in a darkroom. It is familiar to us narrated by Iranian poet Rumi. The condition of correlational studies in Iran is worse, the Hindus definition of elephants despite limitation of their perception was something like shotgun- approach while most papers concerning correlational statistical methods resemble much ado about nothing. Researchers use a variety of rationalizations to make sense of their insignificant research studies that have no value neither for practitioners nor for other researchers. Motivation to gather data in these research studies appears to be different. These researchers think they are contributing to science as a great and noble enterprise and satisfy their ambitions. Most correlational studies in social welfare investigated in this paper were utterly redundant. This prevalent opinion that academics would strive for excellence motivate them to write papers. The methodology of correlational studies in their mind was very easy, they conceptualize hypothesis like humpty Dumpty and produce statements that reflect in arbitrary diagrams and good fitness indexes contribute to confirm their pseudo-hypothesis, in spite of their illusory knowledge. This process is excellent by nonsense. These papers became mass product, the majority of which added nothing to scientific advancement and it should be noted that correlational study only uncover a relationship, it cannot provide a conclusive reason for why there is a relationship. Associations observed in correlational studies is foundation for generating hypothesis that can be tested in experimental or longitudinal studies. The experimental studies then can predict direction of effect in complicated configuration and confirmed by fit indexes. Equivalent models refute all researchers who claim their discourse justifications are sufficient, while they are not. The modern knowledge of correlational study is complex, but includes some aspects of sound simplicity. The pitfall of this gap deceive novice research. responsivity of peer review journals is significant. They must judge by rigor and accurate statistical and research method knowledge. The author of the current paper refers to a complex form of ordinary least square regression that is used to analyze variance in the outcome variables when the predictor variables at varying hierarchical levels namely Hierarchical Linear Modeling, was not apply in research studies that need this type of statistical analysis because of its complexity. Mass production of structural equation modelling in spite of inappropriateness in many papers seems to be simple (but it is not). Although structural equation modelling is a powerful, multivariate technique, misunderstanding of its scientific foundations especially of its heritage from the first generation (path analysis) and the second generation (factor analysis) may produce nonsense research studies being justified by incoherent line of speech.
.
Ethical considerations
Author Consideration
The author spent eight months to review all papers and their reference cited.
Funding
There was no financial sponsor for this critical review.
Conflicts of interest
This review does not overlap with other published works by the author.
Following the ethics of research
The equity principle of ethics was taken into account as building block of critical review .
Type of Study: review |
Received: 2021/11/7 | Accepted: 2023/05/9 | Published: 2023/05/9
Received: 2021/11/7 | Accepted: 2023/05/9 | Published: 2023/05/9
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |
Contact Information
Social Welfare Quarterly
Social Welfare Quarterly, Social Welfare Research Department, University of Social Welfare and Rehabilitation, Daneshjoo Blvd, Evin, Tehran, Iran, 19857-13834
Journal Tel: 02171732851
Website: http://refahj.uswr.ac.ir
Email: refahj@uswr.ac.ir
