Scale Construction - Science topic

I don't understand why you would want to include the items that were rejected by the experts, especially given the issues with CVI. Can you please explain why you are considering this option.

As long as you have complete data (i.e., responses on all four scales and evaluations of all three products), then your statistical approach will be identical. So, if you use the same requirements for both samples, then it should not matter which one you use.

Dear Ayse Seyma Turgut ,

Thank you for the interesting question. In my opinion, content validity and the instrument's reliability are proven by experts to be useful and important even if you modified the questions or not. Also, in the case of a different language.

Kind Regards,

Hi,

This is one of the most widely used books on scale development

DeVellis RF. Scale Development: Theory and Application. Los Angeles, CA: Sage Publications (2012).

Mohammad Jamil

The Rosenberg self-esteem scale (RSES), designed by sociologist Morris Rosenberg, is a commonly used self-esteem measure in social science research. It employs a scale of 0–30, with a score of less than 15 indicating an issue with low self-esteem. The RSES is constructed in the same manner as social-survey questions.

Faten Amer, Usually items not loading on their respective constructs are discarded. However, at times discarded items can be combined to make a new construct.

Robert has already given you advice regarding single items.

Can you share a little more information regarding the EFA, so that we could together see what you have done and how could we help you further? Information such as:

- Which extraction method are you using?

- How the number of factors was identified?

- and, which rotation method was employed?

There is a common misconception between PCA and EFA, and at times right parameters (above) not selected. Consider reading the following article, if not already done:

Costello, A. B., & Osborne, J. (2005). Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical assessment, research, and evaluation, 10(1), 7.

Yes you can use more than one loadcells with raspberry pi... check the raspberry pi GPIO

You may share the same HX711 for several loadcells but cannot measure readings simultaneously...

JASP is one of the latest software and it is free. If you need any assistance you can contact

Faten Amer , sample size it is not a problem at al in Bayesian Factor Analysis, see for example:

After the survey is completed..but it is better to increase the number of studied samples..so that the result will be bette

Yes Talha, you get free access to the Moca test at > https://www.mocatest.org

You will have to demonstrate your competence qualifications to get the required badge.

A useful example of a Functional backup is to ask the client to carry out a simple task AND report back to you through her/his nurse (or email /phone message, etc). Simple tasks like "Please send me your contact information tomorrow", or "Please draw a copy of the shape on this page, then give it to your nurse tomorrow morning", require INITIATIVE, a major executive function skill not tested by simple tests like MOCA where initiative is provided by the tester.

Note that the second task mentioned above requires LESS initiative than the first, since you leave behind a tangible cue for compliance.

Good fortune, Paul

xPsychologist Paul McGaffey, PhD(ABD)

Dear Dr. Jawad thank you very much indeed for the question.

I am taking information from Alyami, M., Henning, M., Krägeloh, C. U., & Alyami, H. (2020). Psychometric Evaluation of the Arabic Version of the Fear of COVID-19 Scale. International journal of mental health and addiction, 1–14. Advance online publication. https://doi.org/10.1007/s11469-020-00316-x

to answer your question.

I quote directly from the statistical section of this paper

"1. They assessed nternal consistency using Cronbach alpha coefficients (α), inter-item correlations as well as corrected item-total correlations.

2. According to DeVellis (2016); Nunnally and Bernstein (1994) a Cronbach’s α of .70 or higher indicates acceptable reliability

3. According to Ferketich (1991).” Inter-item correlations and corrected item-total correlations between .30 and .70 indicate medium to strong associations between items."

If all the components are having strong factor loadings this is an indication that the items are good.

For the second scale if you have 2 components you look at the rotated component matrix

.795 -.178

.843 -.202

If you have 2 components and one is positive and one is negative you go for the one which is positive and having higher value.

I hope this really helps. All the very best

Have you found an answer?

This would be helpful... Good luck

What about trying the satisfactory scales since they are related to the consumer behavior and atitudes towards a product and the consumer decision making (taking care of a product or not)?

Thank you very much.

· Learn more Originally developed to rank rewards that nurses value and that encourage them to remain in their jobs, the McCloskey/Mueller Satisfaction Scale (MMSS) is being used extensively in research and practice to measure nurse job satisfaction.

Background:The eight-factor, 31-item McCloskey/Mueller Satisfaction Scale is one of the most widely used scales for measuring job satisfaction among nurses. However, this scale was developed in 1990 for the American nursing context, and its psychometric validity and utility for use with non-American nurse

Mueller and McCloskey Satisfaction Scale (MMSS) Original Citation – Mueller CW, McCloskey JC. Nurses' job satisfaction: a proposed measure. Nurs Res. 1990 Mar-Apr;39(2):113-7. View in PubMed Contact Information Sharon Sweeney Center for Nursing Classification & Clinical Effectiveness College of Nursing 407 NB University of Iowa Iowa City, Iowa 52242 sharon-sweeney@uiowa.edu Price & Availability – Contact Sharon Sweeney for application. Available for use with payment. Brief Description of Instrument – Measures nurses' job satisfaction in 8 domains: satisfaction with extrinsic rewards, scheduling, family/work balance, co-workers, interaction, professional opportunities, praise/recognition, and control/responsibility. Originally developed for use with hospital staff nurses. Scale Format – 5-point Likert (5=very satisfied, 3=neither satisfied nor dissatisfied, 1=very dissatisfied). 31 items. Administration Technique – Self-administered questionnaire.

Scoring and Interpretation – Sum items of subscale.

Test-retest Reliability – Global scale = 0.64, six month interval. For subscale information, please see original citation. 1 / 2

Mueller and McCloskey Satisfaction Scale (MMSS) Internal Consistency – Global Scale = 0.89. For subscale information, please see original citation. Construct Validity – Correlated with characteristics from the Job Characteristics Inventory. Correlations found in areas of autonomy, friendship opportunities, feedback, variety and task identity. Correlations with intent to stay on job also found. Please see original citation for details. Criterion-Related Validity – Correlated with Brayfield-Rothe General Satisfaction Scale, Hackman and Oldham's Job Diagnostic Survey. Correlations ranging from 0.53 to 0.75 were found. Please see original citation for details. Strengths – Reliable and valid measure. Limitations – Assess different job facets, but not how important that is to the individual nurse. Notes for Consideration – Can use as a global scale or as subscales as required. Published APN Studies using instrument – Koelbel P, Fuller S & Misener T (1991). An explanatory model of nurse practitioners job satisfaction. Journal of American Academy of Nurse Practitioners, 3(1), 17-24. View in PubMed Koelbel P, Fuller S & Misener T (1991b). Job satisfaction of nurse practitioners: An analysis using Herzberg’s theory. Nurse Practitioner, 16(4), 43, 46-52, 55-56. View in PubMed Wild P, Parsons V, Dietz E (2006). Nurse practitioner's characteristics and job satisfaction. Journal of the American Academy of Nurse Practitioners, 18(11), 544-549. View in PubMed Flannery J, Van Gaasbeek DE. Factors in job satisfaction of the psychiatric clinical nurse specialist. Nursingconnections. 1998 Winter;11(4):27-36. View in PubMed

Reference:

Muller McCloskey Job Satisfaction Scale MMSS is being scoring

Shweta Singh, here are some additional texts, though I'm not sure whether there are later editions for some of them, and some might be a bit too focused on health for your specific needs:

Aday LA, Cornelius L J: Designing and Conducting Health Surveys: A Comprehensive Guide, 3rd Ed, Jossey Bass, San Fransisco, 2006.

Kaplan RM, Saccuzzo DP: Psychological Testing: Principles, Applications, and Issues, 5th Ed, Wadsworth, Belmont, CA, 2001.

McDowell I, Newell C: Measuring Health: A Guide to Rating Scales and Questionnaires (2nd ed.), Oxford University Press, New York, 1996.

Murphy KR, Davidshofer CO: Psychological Testing: Principles and Applications, 4th Ed, Prentice Hall, Upper Saddle River, NJ, 1998.

Nunnally JC, Bernstein IH: Psychometric Theory, 3rd Ed, McGraw Hill, New York, 1994.

Streiner DL, Norman GR: Health Measurement Scales: A Practical Guide to their Development and Use, 2nd Ed, Oxford University Press, New York, 1995.

موضوع جميل جدا

أحتاج إلى مثل هذا ربما نستفيد من دراستكم أيها الدكتور العزيز

Dear Mariam,

Please, how many items did you use to capture attitude? Did you also capture the two dimension of attitude - experiential and instrumental? What semantic differential scales did you use 5 point or 7 point? For example, if you use 5 point scale for semantic differential, say useless/useful, useless: 1,2,3,4,5: useful. Please, recode this as -2,-1,0,1,2. Then you generate the attitudinal score by summing all the items. So positive average attitudinal score represents positive or favorable attitude towards the behaviour whereas average negative attitudinal score indicates negative or unfavorable attitude towards the behaviour.

This how you compute behavioral belief: First recode the point scale as I have indicated previously. Afterwards, just multiple the belief strength by the corresponding outcome evaluation for each items. Then, sum up all to get behavioural belief index.

This computation is based Fishbein and Ajzen 2010 (The reasoned action approach book).

I hope this explanation helps with your analysis.

All the best

Emmanuel

Hello David,

As the 0-6 scale your describe is a simple, linear transformation of the 0-3 scale (by doubling the scores), there is no statistical or measurement advantage to either. Whether respondents react similarly to the different response scales is an empirical question.

Good luck with your work.

The sun was just an example. I would make the statement and then using likert scales have 1 as never, going up the scale to 5 always. Sorry, I should have put that.

Maria Acevedo Velazquez, just quickly in case it helps, I think that obtaining item-total correlations (through the reliability section of SPSS) is not going to give you the kind of answers that you want.

I think you also need to be clear about what is mean by discrimination. In some contexts, it means that scores on one construct are different from scores on another construct so that researchers are assured they're not just measuring the same construct in two different ways. In your context (from what I can pick up from your question), you want to be able to discriminate participants one from the other. That's a different kettle of fish.

I hope my pointing out those different things (both of which are dealt with in the Boateng et al. article that you mention) helps you set your sights a bit more clearly.

Vivien So... One of the very important aspects of scale development is the sample size. You need to have to adequate sample size to perform EFA analysis, otherwise, problems like skewness and kurtosis will appear. Further, in your study (as it is exploratory in nature), you can initially retain items with low commonalities (<.40), but check whether items these are cross-loaded on some other factors. If so, delete them. These are the following articles that you may find useful:

Dear Harmanjit Singh,

I tried to pick the most relevant literature that you are seeking.

https://research.brighton.ac.uk/en/publications/do-firms-still-need-to-be-social-firm-generated-content-in-social

https://pure.strath.ac.uk/ws/portalfiles/portal/69138087/Osei_Frimpong_McLean_RESER_2017_Examining_the_moderating_effects_of_firm_generated_content_on_online_social_brand_engagement.pdf

https://www.tandfonline.com/doi/abs/10.1080/10904018.2017.1330656?scroll=top&needAccess=true&journalCode=hijl20

Hello Maria,

If you use the R plug-in for spss, then yes, there are a lot of things (including IRT) that you can do (see this link for information: https://developer.ibm.com/technologies/analytics/tutorials/ba-call-r-spss/).

On the other hand, the free, stand-alone program, FACTOR, handles the polychoric correlations quite nicely and offers many features not readily available in spss (such as parallel analysis) for factor analytic studies:

Good luck with your work.

Thank you so much Dr. . Sincerely appreciate your prompt response.

Warm regards,

Vibha

Generally speaking, the EFA does not have to be followed by CFA, since the two analysis seeks different targets. EFA is mainly used for scale reduction and CFA is mainly used when the aim is to validate a measurement model in the structural equation context.

Now regarding the part that there is no theory available and that your research has not a proposed research model yet. However, I would expect that your questionnaire has a set of pre-defined (factors/constructs) that the (items/questions) are trying to measure.

Therefore, If you are developing a new scale then you definitely should apply EFA and ensure that the new (items/questions) are properly loaded on the targeted (factors/constructs). Later, once your new scale validated with EFA of identified (factors/constructs), you can run CFA, to validate the measurement model.

Hello Jessica,

Factor solutions can include variables that show salient affiliation with more than one factor (some personality measures are notorious for this type of structure). However, the concerns associated with cross-loading are usually: (a) the structure is more complex than Thurstone's idealized concept of "simple structure"; (b) it may make the task of characterizing what a factor represents more challenging; and (c) perhaps the variable isn't as well measured or defined as it could have been.

The answer to your question depends on your specific research aims as concerns this set of variables (and how these do or don't reflect the concerns listed above). If your goal is to derive the "cleanest" possible structure, then throwing out variables/items may be the way to go. Do recognize that the possible concern here is that you end up defining very restricted factors that may not fully represent the target constructs. As well, depending on your sample, it's possible that the resultant structure to choose may be overfitted to the sample and not generalize as well to other data sets.

In any event, if you elect to retain cross-loading items, then:

1. Yes, they are included in any representation of a factor (e.g., an estimated factor score, or just a summated score), or for score reliability estimates;

2. You present a factor structure matrix/table that shows all variables that you deem salient with each factor (factor pattern and factor inter-correlation matrices as well, if you used an oblique rotation);

3. If all loadings on a factor are negative, then you may reverse the sign and characterize the resultant variate as a reversed polarity construct estimate. (You may do the same with mixed sign loadings, as long as you reverse each variable's sign accordingly.)

4. The facetious answer is, if you get down to two items or fewer, you've likely gone too far! The more serious answer is, there is no way to predict in advance how many "keepers" there are from a preliminary set of variables that were constructed, identified, or adapted to tap one or more constructs. That's why people engage in EFA or CFA in the first place; to help identify what structures are supported by data and what structures are not.

The final note I would make is that Cronbach's alpha for an "overall" scale score might not be the best indicator, especially when you have identified multiple factors for that batch of variables. For individual factor scales, sure.

Good luck with your work.

All of the responses above and others out there on the internet seem not backed by any scientific references. For that reason, this response aims to equip readers with proper knowledge from a book of a guru in Statistics, Joseph F. Hair, Jr.

First, it must be noted that the term cross-loading stemmed from the idea that one variable has moderate-size loadings on several factors, all of which are significant, which makes the interpretation job more arduous.

A loading is considered significant (over a certain threshold) depending on the sample size needed for significance [1], which can be seen as follow:

Factor loading - Sample size needed for significance

-----------------------------

.30 - 350

.35 - 250

.40 - 200

.45 - 150

.50 - 120

.55 - 100

.60 - 85

.65 - 70

.70 - 60

.75 - 50

-----------------------------

When a variable is found to have more than one significant loading (depending on the sample size) it is termed a cross-loading, which makes it troublesome to label all the factors which are sharing the same variable and thus hard to make those factors be distinct and represent separate concepts. The ultimate goal is to reduce the number of significant loadings on each row of the factor matrix (i.e. make each variable associate with only one factor). The solution is to try different rotation methods to eliminate any cross-loadings and thus define a simpler structure. If the cross-loadings persist, it becomes a candidate for deletion. Another approach is to examine each variable's communality to assess whether the variables meet acceptable levels of explanation. All variables with communalities less than .50 are viewed insufficient.

RESPECIFY THE MODEL IF NEEDED

What if we should not eliminate the variable base on rigid statistics because of the true meaning that a variable is carrying? Problems include (1) a variable has no significant loadings, (2) even with a significant loading, a variable's communality is deemed too low, (3) a variable has a cross-loading. In these cases, researchers can take any combination of the following remedies:

+ Ignore those problematic variables and interpret the solution as is but the researcher must note that the variables in question are poorly presented in the factor solution

+ Consider possible deletion: depending on the variable's overall contribution to the research as well as its communality index. If the variable is of minor importance to the study's objective and also has unacceptable communality value, then delete it and derive new factors solutions with those variables omitted.

+ Employ alternative rotation method: could be oblique method if only orthogonal had been used.

+ Decrease/increase the number of factors retained: to see whether a smaller/larger factor structure will solve the problem.

+ Modify the type of factor model used (component versus common factor): to assess whether varying the type of variance considered affects the factor structure.

Note:

No matter which options are chosen, the ultimate objective is to obtain a factor structure with both empirical and conceptual support. As we can see, many tricks can be used to improve upon the structure, but the ultimate responsibility rests with the researcher and the conceptual foundation underlying the analysis. Indeed, some empirical researches chose to preserve the cross-loadings to support their story-telling that a certain variable has indeed double effects on various factors [2]. So, ultimately, it's your call whether or not to remove a variable base on your empirical and conceptual knowledge/experience.

Reference:

[1] Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2009). Multivariate Data Analysis 7th Edition Pearson Prentice Hall

[2] Le, T. C., & Cheong, F. (2010). Perceptions of risk and risk management in Vietnamese Catfish farming: An empirical study. Aquaculture Economics & Management, 14(4), 282-314. https://doi.org/10.1080/13657305.2010.526019

I order to test if the formative indicators result in substantially high R-Squared (also known as redundancy analysis), you can define one single indicator of the tentative outcome of all the formative indicators. If you are taking the composite of the multiple dimensions, you must understand that the dimensions should have substantial correlation amongst them, and secondly, the development of a composite should justify theoretically.

Manzu, I would capture some of the advice above with regard to your small sample size, but, much more pessimistically, I would say that you are largely wasting your time and effort to conduct any kind of factor analysis with as few participants as you have. Your results are almost inevitably going to be wonky.

Just as one example, one rule of thumb for exploratory factor analysis recommends 10 times the number of participants as there are items. Although there are other ways of determining an appropriate number of participants, you really have approximately only one participant for each item.

I'm sure you don't have anywhere near enough participants for conducting confirmatory factor analysis, either.

If you want to try to do something valid with your data, I'd look carefully at the interitem correlations and wording of your items. Perhaps they would give you a sense of which items might be redundant.

I don't know why you claim that corrected item-total correlations are able to indicate "discrimination among respondents", but I certainly think that those correlations would be meaningless if you have seven domains and are using the total score across them all as a basis for the total score in the corrected item-total correlations.

For what it's worth, I'd not use coefficient alpha to help sort things out. In my view, that metric is far less useful than is generally believed. There is substantial and authoritative information about that - information that many researchers seem either to be unaware of or to ignore.

You can re-score each variable, x, as its z-score, where z = [x - (mean of x)]/(sd of x). That will give every z-scored variable a mean of zero and a std. dev. of 1.0. With equal dispersions, all variables will have the same weight in an additive composite scale calculated from the z-scores. This approach is covered in most standard textbooks on forming composite scales, but I don't have access just now to my usual resources so can't give you a citation; sorry.

Z-scores treat the underlying variables as interval/ratio in measurement, whereas Likert items are actually just ordinal. Doing EFA on those items makes the same assumption about level of measurement, so you may not be bothered by that aspect of the z-score approach. But if that is a concern, you can accomplish approximately equal weights by just re-scaling your variables to give them all the same range. You could re-code your 4-point items as follows: 1=1, 2=2.33, 3=3.66, 4=5. That would (approximately) preserve the equal-appearing intervals of your original 4-point items, while giving all items the same range, therefore approximately the same dispersion, therefore approximately equal weight in a composite, additive scale.

Shafiq Ur Rehman ,

COVID-19 has triggered wartime effects. Thus, I'd recommend measuring it using a single-item scale. See examples below.

Hi! Factor analysis help to identify the underlying dimensions. Sometimes the measurement items varies based on various contexts. Besides if the measurement is not well established conducting factor analysis can produce clear dimensions that can be used for the particular research model. Thanks

The Kuwait University Anxiety Scale is one of them.

If you want a short unidimensional measure of mental toughness that has demonstrated promising invariance across sports and general samples let me know.

https://www.frontiersin.org/articles/10.3389/fpsyg.2019.01933/full

Happy to provide items (NO COSTS) and assist.

Thanks a lot, Gayan. Do you also happen to know any scales for mapping the self determination theory?

Hi Björn Hommel ;

you can use sirt package for this purpose. DETECT and Noharm tests can be conducted with sirt package. You can use expl.detect() function for DETECT method and noharm.sirt() function for NOHARM method.

Hope helps.

AFK

If you get positive correlation, respondents with higher time spent are disagree more( 5: disagree )

Positive correlation means high value with high value ( coding value)

Thanks Matt!

Communalism seems a good proxy for what I'm after.

We're also considering what you mentioned: developing a scale/construct to address this. The proactive end of the scale would probably match closely what Zimbardo describes as "small everyday heroes", that is people who (1) care for others and (2) do step up even if it comes with a risk/cost.

Best,

Michal

Dear @Kenneth Philips if you have collected your data you have very few options:

-Keep all your data and make all item scores as comparable by making them even with a similar method to what I described.

-Split the scale into two if it makes sense.

-Redo the data collection phase.

As you know already, it is a must your advisor agree to your solution. "Making even procedure" is easily trackable through desriptives such as mean and median, so you have to write what you have done in your thesis.

attached file may help....

Hello Marek,

Qualtrics (www.qualtrics.com) software offers side-by-side questions with a large range of options. It is, however, commercial software.

Good luck with your work!

It isn't clear why you consider rating from poor to good to be ordinal, and consider rating disagree to agree to be interval. They are both the same. Most people would treat them as continuous and use correlation and other methods based on correlation (e.g., regression, SEM). Some people, however, are concerned that intervals might be unequal and prefer to use methods designed for categorical variables (logistic regression, log-linear analysis).

Agree with Peter Samuels ...

Interesting..

Hi everyone. I am sorry to interrupt this discussion but I was wondering if i could get some expert advice on how to analyse my data.

I am trying to run the statistical analysis for a cross cultural study comprising of two country samples. I have 5 variables, one mediator(perceived value) and 3 moderators (subjective wellbeing, healthy lifestyle, ethical identity), IV (conspicuous consumption), DV(customer loyalty) all measured on a 7 point likert scale. I was wondering if 3 moderators is too many and what would be the statistical tests I should perform for my study.

Try this App

https://play.google.com/store/apps/details?id=inferentialstatistics.myapplication

We had a similar problem in the context of educational assessment where very often people work with "judgement scales" of both types, i.e. two-sided bounded (like your [1,5]) as well as one-sided bounded (like your [1,inf] or [0, inf]), and they have to merge those distinct scales in some way. Fortunately, there are not so many ways to do such mergings, if you want to keep things simple and choose the right mathematical framework which suggests you the appropriate transformation. Just picking the one or another transformation to fit your data e.g. to a normal (or some other) distribution is IMHO a brute statistical approach which doesn't do justice - plausibly - to the inherent structure of your data. For instance, in our case, it was plausible to assume a sub-additive operation on scores, something of the form a + b - a*b, and derive a structure which turns out to be a semi-module. From here, everything else could be derived - without guesswork. The 1-1 transformation from a bounded to a unbounded interval we used had the relatively simple form: 2-log ( score / (1 - score) ), where 2-log is the binary log function, and scores are numbers between 0 and 1. The resulting numbers are between -inf and +inf, but of course we can adapt it such that the resulting numbers start from 0 onwards. Because the transformation is 1-1, it has an inverse which goes the other way back. Works great! Probably there are statistical techniques for doing tests with your data, but that is none of our concern in our context.

I agree with Danial Goldman that it is up to your university to make the decisions, but I can say that at my university, this would certainly be adequate for a masters thesis but not necessarily for a PhD.

One possible way to resolve this would be at the level of validation, since that typically requires collecting other measures to demonstrate that the proposed measures has the predicted relationship to other concepts. So, a detailed pretest of the measure might be sufficient.

You can develop your questionnaire, based on these assumptions, and then use the results to possibly falsify those assumptions, which would be great, as that's the goal of science anyway. But you need to word the questionnaire in a way that does not produce a bias towards or away from your assumption.

Is our questionnaire going to be a large scale survey or an interview? If it's an interview especially, you're going to want the questions to be open ended so as not the lead the interviewee. So you wouldn't want a question like "do you agree that when your work is of poor quality, your supervisor acts abusively?" And instead perhaps ask "how does your supervisor respond to the quality of your work?"

Make sure you get an independent "back translation" into English to make sure that the content does indeed match the original.

More items, less answer choices, more rating training.

Hope this helps,

Matt

If the assumptions underlying both reliability measures were met, i.e. essential tau-equivalence for alpha and parallel measures for retest reliability, both measures would be identical. However, the assumptions are seldom tested, and violating the assumptions leads to different values. One reason for a low rest-reliability is the time-lag: With increasing time interval between the measurement time-points the correlation will decrease because of occasion-specific influences. This correlation is often treated as a reliability coefficient, which is generally not correct because the assumptions for this coefficient were not met.

HTH, Karin

I have seen this work, the authors have similar studies, it helped me quite. Thank you very much for your support.

Best wishes..

Matthew Kerry Thank you so much. That was extremely helpful.

Leon Vahlkamp, you are also suggested to collect data to examine factor structure of the translated scale (especially if the scale is not unidimensional) using exploratory and confirmatory factor analysis.

Most studies use either the Bennett & Robinson deviance scale that assesses deviance directed at organizations or people, or the Spector et al. Counterproductive Work Behavior Checklist (CWB-C) that in addition to CWB-Organization and CWB-People, assesses five dimensions of Abuse, Production Deviance, Sabotage, Theft, and Withdrawal:

You can find the Bennett & Robinson scale in their 2000 paper, and you can find the CWB-C scale here: http://shell.cas.usf.edu/~pspector/scalepage.html

Bennett, R. J., & Robinson, S. L. (2000). Development of a measure of workplace deviance. Journal of Applied Psychology, 85(3), 349-360. doi:http://dx.doi.org/10.1037/0021-9010.85.3.349

Spector, P. E., Fox, S., Penney, L. M., Bruursema, K., Goh, A., & Kessler, S. (2006). The dimensionality of counterproductivity: Are all counterproductive behaviors created equal? Journal of Vocational Behavior, 68(3), 446-460. doi:http://dx.doi.org/10.1016/j.jvb.2005.10.005

Hi Leopold,

in measurement questions like this one I think the first issue to attend to is whether you conceive of the measurand (the thing being measured) as a latent trait/variable or not. Measurands that we believe are latent traits are (real but un-observable) things that exist inside people, like intelligence or extroversion. In that case you assume that the latent variable has an underlying continuous structure... this implies a couple of things: 1) that you attempt to scale the responses so that individuals are reporting about themselves ("I always greet the older people first"), 2) as you choose items (indicators of the latent variable) you attempt to scale the response options - even when they are likert scaled - in such a way that you can add them up so that higher scores indicate more of the thing being measured. So this would imply choosing a number of items that all point at 'manners' and that are all scaled positively. Lower scores will indicate the absence of the thing being measured.

On the other hand you may conceive of 'manners' as a constructed entity and would be creating therefore an index variable. Index variables do not need to have indicators that are all on the same scale and responses to items can be combined in any way (for example you might combine items so that they maximize the prediction of some outcome).

I recommend the very good article "Three Cs in measurement models: causal indicators, composite indicators, and covariates"by Bollen & Bauldry (Psychol Methods. 2011 Sep;16(3):265-84. doi: 10.1037/a0024448.)

Hope that is helpful to you in this very interesting conversation!

Appraisal theory / contextualised factor (sometimes called affective decision ecology) might be relevant. But you will need to bring these perspectives into your discourse.

Affective

Attitude

Emotion

and some other related concepts are used in art works. Movie industry uses this and observes what elicits response...e.g. music.

In regards to IRT, I would suggest you to start non-parametric IRT (KernsmoothIRT) in R to evaluate/describe the use of response options for all items. May want to drop some if they do not work well with overall scores. If response options look OK, move to a parametric graded response model (ltm package in R) for item parameter estimation, identification of patterns of responding, and constructing of a standard error of measurement that accounts for where on the continuum of scale performs best/worst. It might be meaningful to start with the theory behind the scale development to help.

Three-Component Model file attached. In the same file, you also have the Organizational Questionnaire.  Hope it will help you. 

The purpose of scale construction is to design a questionnaire that provides a quantitative measurement of an abstract theoretical variable.

•

Scale construction can be seen as a specialized area of survey design. Not all surveys are scales, however. Before trying to apply the procedures presented here to the construction of your questionnaire you should decide whether it really is a scale. The most obvious key is that a scale uses a moderately large number of items to measure a single construct. The items within a scale are typically

interchangeable , meaning that the response to one item has exactly the same theoretical meaning as the response to any other item within the scale. Each

item is designed to be a different way to “ask” about the same theoretical variable. Some questions may written so that more positive responses indicate less of the variable in question , but the expectation is that the magnitude of the correlations between items (whether positive or negative) should be relatively high throughout a scale. Additionally, after a participant completes a scale the responses are aggregated in some way, such as by taking the average, to produce the actual measurement. The specific response provided to any individual question is usually not of interest.

• Good scales possess both validity and reliability

. A scale has validity if it properly represents the theoretical construct it is meant to measure. A scale has reliability if repeated measurements under the same circumstances tend to produce the same results. These two concepts are very important to scale construction.

• Sometimes a single questionnaire contains items from several different scales mixed together. This is perfectly legitimate. In this case your items will not all be interchangeable - items making up different subscales will be slightly different. The items should, however, be interchangeable within each subscale.

These two standards can be useful as a reference: 

1. Risk management: Risk assessment techniques BS EN 31010:2010

2. Multi-criteria analysis: a manual http://eprints.lse.ac.uk/12761/1/Multi-criteria_Analysis.pdf

I used semi-quantitative analysis for my PhD using rating matrices. 

If you have categorical variables, then the analysis of covariance would be suitable. But you mention Likert scale, so we assume you have rating 1-7; in which case the ANCOVA may not be most optimal

There is a difference between working with individual items that used Likert-style scoring, versus combining those items into a scale that would have interval-level measurement properties.

Because there have been so many questions about this issue, I have compiled a set of resources on it.:

I don't recommend cutting your variables up into categories for two reasons. First, you lose information. Second, trying to examine interaction effects when both your variables include multiple categories can be a real headache.

Hence, you should follow the advice of others and test multiplicative effects between your independent variables in the regressions equation.

There is a lot of debate whether parametric tests or non-parametric tests can be used. But in summary if you are prepared to assume your data represents interval data then parametric tests are usually not inappropriate.

Some useful references in this article: http://eprints.uwe.ac.uk/30814/

With regards the neutral response, these references may be of interest:

Armstrong, R. L. (1987). The midpoint on a five-point Likert-type scale. Perceptual and Motor Skills, 64(2), 359-362.

Bishop, P. A., & Herron, R. L. (2015). Use and misuse of the Likert item responses and other ordinal measures. International journal of exercise science, 8(3), 297.

P.s. my sample size is 104 respondents

Very helpful..thanks a lot, you