To define a factor in a factor analysis you need at least 3 items. The reason is because 2 items define only a single correlation.

But you can still get an alpha reliability from only 2 items, which makes sense if the two items are "highly" correlated.

Dear Suresh,

There is no hard and fast rule for the minimum number of items to measure a single dimension of a scale. As the number of items increase, the percent of variance explained by the model increases. That means much of the variation in the model is explained by the items. Therefore if you want, say 70% of the variance to be explained by the model, and if 5 of the items can enable you do that, it is right to have five items.

In general, it depends on that specific topic of interest.

Dear Suresh, 

Number of items depend on the attitude scale model which you propose to adopt. As such there is no minimum or maximum. You have to formulate on your own based on personal assessment.

Best regards

To define a factor in a factor analysis you need at least 3 items. The reason is because 2 items define only a single correlation.

But you can still get an alpha reliability from only 2 items, which makes sense if the two items are "highly" correlated.

Thank you Morgan, Mani and Tolu for your valuable inputs.

with regards

Dear Suresh though there are no hard and fast rules, atleast 20 items are required to enable the respondents to involve and express their responses through different dimensions as the researchers required to analyse in detailed manner. Researchers may have negative and positive  items to eliminate error responses. There is a sufficient ground for the respondents to express their response. If the items are very minimal 3-5, then the statistical  treatment will be difficult

I have started a thread on ResearchGate that collects resources on introductory-level scaling issues:

https://www.researchgate.net/post/How_can_one_help_people_create_simple_scales_from_Likert-scored_items

Wow!

Impressive information. Thank you all.

Perfect

I know I'm a little late to the party, but here's some useful information, in case someone still needs it.

A similar question was posted earlier and received some very good answers. So, instead of copy-pasting the info, please make use of this excellent resource:

https://www.researchgate.net/post/Does_a_factor_consist_of_2_items_in_Exploratory_Factor_Analysis_Is_it_acceptable

I have formulated a scale for

ORGANIZATIONAL JUSTICE, which has three sub-factors namely, procedural, interactional and distributive justice having to have items two, three and three respectively. Is the number of items valid?