I've already made the point that these cutoff points for alpha are folklore, and they are most definitely not what either Cronbach or Nunally recommended. 

Your conclusion that the items are valid cannot be supported by alpha either. It measures internal consistency, not validity. 

See https://www.researchgate.net/post/Cronbachs_alpha_below_07

Basically, I said there

The urban legend that an alpha of 0.7 is acceptable for a measurement scale is just that: an urban legend. It is chronicled in a very amusing paper by Lance (Lance CE, Butts MM, Michels LC. The Sources of Four Commonly Reported Cutoff Criteria: What Did They Really Say? Organizational Research Methods. 2006 Apr 1;9(2):202–20. )

Nunally's 1978 paper actually said "what a satisfactory level of reliability is depends on how a measure is being used. In the early stages of research . . . one saves time and energy by working with instruments that have only modest reliability, for which purpose reliabilities of .70 or higher will suffice. . . . In contrast to the standards in basic research, in many applied settings a reliability of .80 is not nearly high enough. "

Thank you Dr Carlos for your detailed explanation. I did encounter some journals and books employed 0.5 as a minimal/ acceptable requirement to acknowledge that particular instrument is reliable - so, if i use below references as basis where 0.5 is acceptable cronbach, is it fine?

see, 

Bowling A: Research Methods in Health. Investigating health and health

services. Second edition. Buckingham, Open University Press; 2002.

Schmitt M: Uses and abuse of coefficient alpha. Psychol Assessment 1996,

8:350-353.

L. D. Streiner and G. R. Norman, Health Measurement Scales:

A Practical Guide to Their Development and Use, Oxford

University Press, New York, NY, USA, 2008.

D. G. Altman, Practical Statistics For Medical Research, Chapman

& Hall, London, UK, 1991.

J. C. Nunally, Psychometric Theory, McGraw-Hill, New York,

NY, USA, 1978.

Davis, F.B., 1964. Educational Measurements and Their Interpretation, Wadsworth, Belmont,California.

Chakrapani, C. 2004. Statistics in market research. Arnold Publisher, London

Kehoe, J. 1995. Basic Item Analysis for Multiple-Choice Tests. Practical Assessment, Research & Evaluation,4(10).

It's depend of field of research. In psychology, for example, acceptable level is .70 but sometimes if You have explanation for higher level of an error (e.g. wide range of the subject) it could be .55 minimum.

A similar question was posed recently for test-retest reliability measures. The responses to that question may be useful. (https://www.researchgate.net/post/What_is_the_acceptable_value_for_test_retest_reliability_of_a_psychological_test?_tpcectx=qa_overview_following&_trid=5429d2ced039b11c228b45ae_)

As said above, methodologists stress the problems with alpha (e.g., Bruce Thompson, http://directory.cehd.tamu.edu/view.epl?nid=bruce-thompson), but many in applied fields of psychology continue to report it. I think it is unwise to have some value of alpha to aim for. This could encourage people simply to repeat questions which will increase alpha (sometimes this could be good, to check consistency in responding, but then it is for a different purpose) or to only measure a narrow conception of the construct.

I would be weary of having even of range of alpha levels that are viewed as "acceptable" or "good enough" across all situations. Further, surely some humans in the same situations will differ in what they find "good enough".

ps. A good Thompson chapter is Bruce Thompson (2003). Understanding

reliability and coefficient alpha, really. In Thompson, B. (Ed.). Score

Reliability. Sage Publications.

In this discussion only a bottom threshold is discussed. When constructing a scale defined by the items included, I look for an alpha between .75 and .85. Items then have an underlying factor in common, but are still different enough but not too different. What I find interesting is that many researchers argue that an alpha above .90 is very good. In my opinion in such a case you have too many items asking the exact same thing. An alpha between .65 and .90 is acceptable or satisfactory (but not satisfying!). My field is the social sciences.

Thank you for all the insightful remarks, my field is also considered as social sciences, more precisely, environmental economics. However, i personally like the argument by Schmitt (1996) and Thompson (2003)on the misunderstanding and clarification on reliability and consistency. The reason why i posed this issue is due to one of my alpha's value is relatively low i.e., 0.53 (albeit it is low consistency, but i believe the subscales/ items are valid and it should not be attenuated in this sense) and the rest are 0.73 and 0.84. 

I've already made the point that these cutoff points for alpha are folklore, and they are most definitely not what either Cronbach or Nunally recommended. 

Your conclusion that the items are valid cannot be supported by alpha either. It measures internal consistency, not validity. 

See https://www.researchgate.net/post/Cronbachs_alpha_below_07

Basically, I said there

The urban legend that an alpha of 0.7 is acceptable for a measurement scale is just that: an urban legend. It is chronicled in a very amusing paper by Lance (Lance CE, Butts MM, Michels LC. The Sources of Four Commonly Reported Cutoff Criteria: What Did They Really Say? Organizational Research Methods. 2006 Apr 1;9(2):202–20. )

Nunally's 1978 paper actually said "what a satisfactory level of reliability is depends on how a measure is being used. In the early stages of research . . . one saves time and energy by working with instruments that have only modest reliability, for which purpose reliabilities of .70 or higher will suffice. . . . In contrast to the standards in basic research, in many applied settings a reliability of .80 is not nearly high enough. "

I would like to second the comments of theo D.

AND his rules of thumbs comments also apply to the measure I recommend below.

Modern articles should prefer the reporting of what is commonly referred to as maximised or composite reliability.

Werts, C. E.; Linn, R. L., and Joreskog, K. G. Intraclass reliability estimates: Testing structural assumptions. Educational and Psychological Measurement. 1974; 34:25-33.

This estimate of reliability rightfully takes the correct weighted contribution into the reliability calculation. That is .... it follows the congeneric model that Joreskog talks about.

The max reliability is the upper bound measure of reliability.

Raykov, Tenko. and Shrout, Patrick E. Reliability of Scales With General Structure: Point and Interval Estimation Using a Structural Equation Modeling Approach. Structural Equation Modeling. 2002; 9(2):195-212.

Cronbach's alpha calculation equally weights variable contribution in the statistic calculation.

FOR elementary background see:

Hattie, John. Methodology Review: Assessing Unidimensionality of Tests and Items. Applied Psychological Measurement. 1985; 9(2):139-164.

Journal reviews that still push for Cronbach's alpha reporting in my opinion... is antiquated.

my 2 cents worth.

brad

On this issue, I suggest to have a look at this paper:

Internal consistency, retest reliability, and their implications for personality scale validity.

McCrae et al. (2011). Personality and Social Psychology Review, Vol 15(1), Feb, 2011. pp. 28-50.

Have a look at this paper:

http://link.springer.com/article/10.1007/s11336-008-9101-0/fulltext.html

Psychometrika.  10.1007/s11336-008-9101-0. Theory and Methods.

On the Use, the Misuse, and the Very Limited Usefulness of Cronbach’s Alpha

Klaas Sijtsma.