Here are three ways to answer your question

From me

Whether or not to use subscales with low Cronbach's α in statistical analysis depends on a number of factors, including the specific research question, the sample size, and the availability of alternative measures.

Factors to consider:

• Research question: If your research question is specifically about one of the subscales, then you may need to use that subscale even if the Cronbach's α is low. However, if you are more interested in the overall construct that the questionnaire measures, then you may be able to use the total score instead.
• Sample size: A larger sample size can compensate for lower reliability. If you have a large sample size, you may be able to use subscales with lower Cronbach's α without too much loss of power.
• Availability of alternative measures: If there are alternative measures of the same construct that have higher reliability, then you may want to consider using those measures instead.

Recommendations:

• If the subscales have very low Cronbach's α (e.g., below 0.50), then it is generally not recommended to use them in statistical analysis. This is because the results may be unreliable.
• If the subscales have moderately low Cronbach's α (e.g., between 0.50 and 0.60), then you may be able to use them in statistical analysis, but you should be aware of the limitations of the data. You should also be careful not to overinterpret the results.
• If the subscales have acceptable Cronbach's α (e.g., above 0.60), then you can use them in statistical analysis without too much concern.

If you are unsure whether or not to use subscales with low Cronbach's α in your statistical analysis, it is a good idea to consult with a statistician or research methodologist.

References:

• Nunnally, J. C. (1978). Psychometric theory (2nd ed.). McGraw-Hill.
• George, D., & Mallery, P. (2003). SPSS for Windows step by step: A simple guide and reference (4th ed.). Allyn & Bacon.
• Field, A. (2013). Discovering statistics using SPSS (4th ed.). Sage Publications.

Example:

Suppose you are interested in the relationship between anxiety and academic performance. You develop a questionnaire to measure anxiety, which has two subscales: social anxiety and test anxiety. The Cronbach's α for the social anxiety subscale is 0.55, and the Cronbach's α for the test anxiety subscale is 0.60. The Cronbach's α for the total anxiety score is 0.70.

In this case, you may want to consider using the total anxiety score in your statistical analysis, as it has the highest reliability. However, if you are specifically interested in the relationship between social anxiety and academic performance, then you may need to use the social anxiety subscale, even though its reliability is lower.

If you do decide to use the social anxiety subscale, you should be aware of the limitations of the data. You should also be careful not to overinterpret the results. For example, if you find a statistically significant relationship between social anxiety and academic performance, you should not conclude that social anxiety causes poor academic performance. It is possible that there are other factors that explain the relationship.

I hope this information is helpful.

From my friend 1

It is not recommended to use subscales with low Cronbach's alpha in statistical analysis. Cronbach's alpha is a measure of internal consistency, which means how well the items in a subscale measure the same underlying construct. A low Cronbach's alpha indicates that the items in the subscale are not measuring the same thing very well, which can lead to unreliable results.

If the total score of the questionnaire has a desirable Cronbach's alpha, then you can use the total score in your statistical analysis. However, it is important to note that the total score may not be a good measure of the underlying construct that you are interested in. For example, if the questionnaire is measuring multiple constructs, then the total score may be a mixture of scores from different constructs.

If you are interested in analyzing the subscales of the questionnaire, then you should consider using a statistical method that accounts for the low Cronbach's alpha, such as structural equation modeling (SEM). SEM allows you to test the measurement model of the questionnaire and to estimate factor scores for each subscale.

Here are some references on the use of subscales with low Cronbach's alpha in statistical analysis:

• Kline, R. B. (2015). Principles and practices of structural equation modeling (4th ed.). Guilford Publications.
• McDonald, R. P. (1999). Test theory: A unified treatment. Lawrence Erlbaum Associates.
• Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). McGraw-Hill.

In your specific case, you are planning to use a repeated measures MANCOVA. This statistical method requires that the dependent variables have good internal consistency. Therefore, I would recommend that you use the total score of the questionnaire in your MANCOVA analysis, rather than the subscales.

However, you should also consider using SEM to analyze the subscales of the questionnaire. This would allow you to test the measurement model of the questionnaire and to estimate factor scores for each subscale. The factor scores could then be used in your MANCOVA analysis as dependent variables.

From my friend 2

Whether to use subscales with low Cronbach's α or use only total score with desirable α in statistical analysis depends on a number of factors, including:

• The purpose of your study. Are you interested in the overall construct measured by the questionnaire, or are you interested in the specific subscales? If you are interested in the specific subscales, then you may need to use them even if their Cronbach's α is low.
• The sample size. If you have a large sample size, then you can use subscales with lower Cronbach's α without losing too much power. However, if you have a small sample size, then you may want to avoid using subscales with low Cronbach's α, as this could lead to unreliable results.
• The other statistical tests you are using. Some statistical tests, such as MANCOVA, are more sensitive to low reliability than others. If you are using a statistical test that is sensitive to low reliability, then you may want to avoid using subscales with low Cronbach's α.

In your specific case, you have a questionnaire with total score and scores for subscales. The subscales have few items, and the Cronbach's α is not desirable for the subscales, but it is desirable for the total score. You are planning to use a repeated measure MANCOVA.

Repeated measure MANCOVA is a statistical test that is sensitive to low reliability. Therefore, it is not recommended to use the subscales in your repeated measure MANCOVA unless you have a very large sample size.

Instead, you should use the total score in your repeated measure MANCOVA. The total score has a higher Cronbach's α, so it is more reliable.

References:

• Field, A. (2013). Discovering statistics using IBM SPSS Statistics (4th ed.). Sage.
• Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). McGraw-Hill.
• Shrout, P. E., & Shrout, M. F. (2011). Reliability. Wiley.

Additional considerations:

• If you are interested in the specific subscales, you could consider using a different statistical test that is less sensitive to low reliability, such as exploratory factor analysis (EFA).
• You could also consider revising the subscales to improve their reliability. For example, you could remove items that have low correlations with the other items in the subscale, or you could add new items.
• Finally, you could acknowledge the low reliability of the subscales in your study limitations.

Good luck

Alireza Azimpour , I am afraid you have been spammed. The "answers" posted by E.A. Gawad are simply pasted from ChatGPT. You can verify this by pasting a chunk of the text into https://gptzero.me.

Of course Alireza doesn't know the answer, or care if the information he posts here is wrong, or misleading. He's just trying to game his score on ResearchGate. And of course, anyone who is checking up on him will spot the fact instantly that he's just a fraud. But in the meantime, we get junk all over RG posted by these frauds.

Hello Alireza Azimpour. The best answer I have for you is this excerpt from the preface to Robert P. Abelson’s book, "Statistics as Principled Argument".

---start of excerpt---

For many students, statistics is an island, separated from other aspects of the research enterprise. Statistics is viewed as an unpleasant obligation, to be dismissed as rapidly as possible so that they can get on with the rest of their lives. Furthermore, it is very hard to deal with uncertainty, whether in life or in the little world of statistical inference. Many students try to avoid ambiguity by seizing upon tangible calculations, with stacks of computer output to add weight to their numbers. Students become rule-bound, thinking of statistical practice as a medical or religious regimen. They ask questions such as, "Am I allowed to analyze my data with this method?" in the querulous manner of a patient or parishioner anxious to avoid sickness or sin, and they seem to want a prescriptive answer, such as, "Run an analysis of variance according to the directions on the computer package, get lots of sleep, and call me in the morning."

For years, I always responded to students who asked, "Can I do this?" by saying something like, "You can do anything you want, but if you use method M you’ll be open to criticism Z. You can argue your case effectively, however, if you use procedure P and are lucky enough to get result R. If you don’t get Result R, then I’m afraid you’ll have to settle for a weaker claim."

Eventually, I began to appreciate an underlying implication of the way I found myself responding: namely, that the presentation of the inferences drawn from statistical analysis importantly involves rhetoric. When you do research, critics may quarrel with the interpretation of your results, and you had better be prepared with convincing counterarguments. (These critics may never in reality materialize, but the anticipation of criticism is fundamental to good research and data analysis. In fact, imagined encounters with antagonistic sharpsters should inform the design of your research in the first place.) There are analogous features between the claims of a statistical analyst and a case presented by a lawyer--the case may be persuasive or flimsy (even fishy), the style of inference may be loose or tight, prior conventions and rules of evidence may be invoked or flouted, and so on. (p. xii)

---end of excerpt---

Abelson, R. P. (2012). Statistics as principled argument. Psychology Press.

Thank you very much Alireza Azimpour, Ronán Michael Conroy, and Bruce Weaver

Ali Reza Azimpour, Best regards, three carefully reviewed answers with useful references and actually drafted with artificial intelligence to avoid phrasing problems and linguistic defects. You can benefit from them directly or consult experts in modeling and simulation who are familiar with advanced statistical processes to discuss the answers. Please publish what will benefit you here because I need to increase my knowledge daily. Perhaps I missed something or perhaps my colleagues and I were wrong, as no one possesses the complete truth about creatures.

I will look at the positive aspect of Dr Ronán Michael Conroy's response... which is his keenness for the published information to be documented and not misleading. This is an important thing. He could have discussed the answers objectively as he specializes in statistics or presented them to specialists in modeling and simulation and is familiar with statistical processes. Please present the responses here. To benefit from it, as I mentioned before, this is something I love. As for manipulating its results on ResearchGate, this is a matter for young people, not those of late age like me. Note: The published photo (it is from approximately 20 twenty years ago) is mine, placed by my young colleagues who created this page, which I visit very frequently, perhaps two or three times, when I am on vacations that extend for days throughout the year, in order to use it in scientific tourism through scientific books and publications, and trying to help people. Scientifically for free.

My greetings to the three of you for your keenness on knowledge, even if your specializations and viewpoints differ.

Happy times everyone

E.A. Gawad I'm sorry. Your "friend" who is actually a second AI text generator? You are trying to pass off computer-generated text as your own contribution. Do you really have the competence to spot the errors in the text you posted. Where, for example, does the reference to Nunally fit? And why the absolutely vague references to basic books on SPSS? Because you don't know what you are talking about.

I repeat: you are a fraud.

Thank you very much

Instead of arguing, you can help the researcher by proposing a practical answer that will help him solve his problem

Save everyone's time

Good luck