I can imagine this happening.  Students sometimes believe that ther knowledge or ability is higher than it really is.  Their self-efficacy sometimes leads them into believing that they know as much as the teacher, or that they know enough and do not need to learn more.  This would reduce their motivation for academic achievement.

Self efficacy means that students are confident about their ability.  Students may become to confident that they think they no longer have anything to learn...or at least that their current teacher cannot teach them anything.  Such students would likely not study as hard and thus have poorer academic achievement.  

When self-efficacy lags, effort will decline

Perception of self-efficacy will differ from subject  to subject. When student's self-efficacy drops within a subject, they give lesser importance and  less devotion of time and effort or may even switch to another 

Dear Firdaus, Michael, Krishnan, and Stefano  (grazie)

1. Many variables in psychology and sociology are quadratic and not linear. We tend to favor linearity because of our need for parsimony, but quadratics are not difficult to understand and make great sense in reality. For example, too little and too much anxiety are both bad.

2. I can also see many situations where self-efficacy is contaminated or inflated by narcissism, being coddled as a child to the point where one believes he or she can do anything, or being part of the 21st century where via tweets and Facebook, etc everyone can become "famous" overnight without any measurable achievement..... All of these  may inflate reported self-esteem on a measure without the requisite achievement, social/ intellectual abilities, and motivation to justify the reported self-esteem.

Good luck to you.

Peter

Hi Firdaus

If there is no significant relationship, and you have a very small negative correlation, I would suggest investigating further using a person-oriented approach. This might reveal groups of students who had a positive correlation between self-efficacy and achievement, and others who had negative or no association. Then you can check these groups of students for cultural or gender differences for example, to help explain the divergent associations, if these indeed do exist in the data.

Personally I'd do this exploration before discussing the substantive implications of a negative and insignificant relationship existing, as this might be misleading for thinking about what is actually happening in your sample (e.g. the direction of correlations might be multi-modal in effect therefore the average correlation value that you have is misleading).

Also then you could have a fuller discussion about which proportion of your sample fits with existing literature and which does not - this is especially interesting as I gather your data are from Indonesia therefore might not fit with common Western findings. 

Another way is to see if gender or social background mediated the correlation, or do a multi-group analysis.

Best,

Jenny

As some of my colleagues have aid, merely self-efficacy don´t works alone: the type and desnity of tasks, the past experience (positive  or negative), the consequences oh doing good  or bad te exercices, the cultural context...... That implies a qualitative multifactoiral approach that lets know the PROCESSES or, complementary, quantitative approach with representative samples that lets to know the variance attributed to self-efficacy. Really it is complex.

Could i please get a copy of your friends research work? It'll help me a lot on mine