A single item measured on a Likert scale is ranked or ordinal data. Correlation tests with another variable measured on an interval/ratio scale uses the Spearman rho statistical test. If you are comparing two groups on a single item Likert scale, the appropriate statistic is the Mann Whitney U test.

Determination of basic statistical tests is fairly straightforward once you have answered two questions: (a) what is the basic comparison that you are making, comparison of two variables or comparison of two (or more) groups?  (b) What is the scale of measurement for each variable you are testing?

Pearson r is for interval/ration x interval/ratio correlations. Spearman rho for ordinal correlations. Phi coefficient for nominal correlations.

Group comparisons, of course, depend on the scale of measurement for the dependent variable. T-test and ANOVA are for interval/ration group comparisons. Mann Whitney U test for two group ordinal comparisons. Chi square is for nominal group comparisons.

The risks of using ANOVA and other parametric statistical techniques to analyze Likert (or other ordinary-scaled data) have been greatly exaggerated. For a thorough discussion of this issue, see:

Norman, G. (2010). Likert scales, levels of measurement and the "laws" of statistics. Advances in Health Science Education. DOI 10.1007/s10459-010-9222-y.

You can study my papers  : 

*The effects of work-related stress on work ability index (WAI) among Iranian workers

* Multiple dimensions of work–related risk factors and their relationship to work ability among industrial workers in Iran

* Role  of  Psychosocial and Physical Work-Related Factors on the Health – Related Quality of Life of Iranian Industrial Workers

Thank you so much. One thing I could not make out that so many writers have used pearson correlation and also multiple regression with such type of article!

Tota, welcome to the world of science!

Donald, although I agree that parametric tests are robust and there are many myths about them, I do not agree that the article you suggested discusses the topic thoroughly.

I agree with Peter.Very nicely explained But what about factor analysis as a first step followed by regression to find out the linkage It is better to give table /data and ask to get more clarity

After statistical analysis of the data, some advanced methods such as order logistic, multiple logistic can be applied to find the relationship between dependent variable and independent variables.

can you give examples on order logistic or multiple logistic to understand the same 

As a new researcher, the absolutely fundamental thing is to understand the distribution of each variable at the simplest level. Start with the basic frequency of values for each variable. Do values on the variable have a distribution or do they seem to collect at one end or the other? Then, moving toward a scale, check inter-item correlations and correlations with total score. That may identify you need to do a factor analysis. Presuming that others have used the same scale, be sure to understand similarities and differences in your basic distributions and correlations compared to prior work before moving onto the higher-level statistical analysis methods mentioned above. That comparison can provide illuminating information in terms of similarities and differences given your sample and the samples of earlier researchers. The basic underlying principle is to be sure you understand what you CAN do given the data you worked so hard to obtain before you measure what you seem to have done (Tukey, Exploratory Data Analysis). Following basic exploratory data analysis principles before moving forward minimizes garbage in, garbage  out (GIGO).

Thanks