Question

One measure of reliability we discussed briefly is item-total correlation. Item-total correlation is specifically the correlation between a single item's value, and the summation of all remaining items for each respondent (imagine a survey instrument with 10 items, all on a likert scale.) So, in this example there is an item-total correlation for every item. Does scale usage affect item-total correlation? Why or why not? Think back to what you know about how correlation can be measured (algebraically).

Answer 1

The example is about a survey instrument with 10 items,all on a likert scale.Now the question is all on which point Likert scale: a 5-point or 3 point or 2-oint(0 and 1) Likert Scale??

As by definition item-total correlation is nothing but a measure of correlation between two continuous variables.So, sometimes instead of 5-point likert scale if we use 2 or3 point then there is a chance of increment/decrement in the item total correlation.Researchers always do the trick to change of scaling of the likert responses to get the cutoff of cronbach's alpha. As item total correlation completely depends on the item values.So i think Scale usage affect item-total correlation.

(Possible Reason: A typical five-level Likert item, for example, could be: Strongly disagree Disagree Neither agree nor disagree Agree Strongly agree The above 5-options can be changed into 3-point Likert Scale 1.Bad(Strongly disagree, disagree) 2.Neutral(Neither agree nor disagree) 3.Good(Agree,Strongly agree) The item correlation can be different due to change of scaling(for above two separate cases.

Concept of Item total Correlation: ‘discrimination’ referring to how well a question differentiates between participants who know the material and those that do not know the material. Participants who know the material taught to them should get high scores on questions and high overall assessment scores. Participants who did not master the material should get low scores on questions and lower overall assessment scores.))