I performed an exploratory factor analysis on a questionnaire for democracy. This gave three factors. The items in the second and third factors all load negatively on the factors. The case of the second factor is understandable for me because the items in this factor are all worded negatively. The problem is with the third factor that includes two items worded negatively and nine items worded positively. Those worded negatively are related to justice in the classroom (E.g., there is no transparency regarding the exam grade), while those worded positively are related to engagement/dialogue in the classroom (E.g., the teacher listens to students' discussions). It was hard for me to identify this factor. Could I call it 'absence of dialogue'? Another issue, what about the fact that it includes items from theoretically two constructs (justice and absence of dialogue)? What do we do in this case?
Dear Wajeeh,
It is your job to interpret the meaning of the factor based on the items which load upon it, either positively or negatively. This is a qualitative process. EFA cannot help you to do this. It may be that your method is inappropriate or your criteria for choosing how many factors to extract or your criteria for including items in a factor needs to be changed.
I would also recommend carrying out a principal component analysis with just the items in one factor to double check that they do indeed form a single construct without the presence of other information in the analysis.
Dear Peter,
Thank you very much for your two suggestions. Can you elaborate more regarding the first one; i.e. the method. In the 'extraction' option, I used the 'eigenvalues greater than one' , and the 'method' was 'principal components'. In the 'options' option, I used 'absolute value below .1' (or below .3). These are the conditions which I chose for extraction.
all the negative items needs to be reversed at the time of tabulation of data or entering into data sheet of softwares. please do this I hope your problem will be solved.
Regards
Dear Wajeeh
Those questions which are negative in nature, call reverse questions. During data entry if the responded have ticked 1or 2 or 4 or 5, you should enter 5 instead of 1, 4 instead of 2, 2 instead of 4, or 1 instead of 5, respectively. If your mean is less than mid-point =3 and significant, it means that your respondents are not agree with questions raised, and if the mean is high than mid-point and significant, means that respondents are disagree with questions raised.
I hope that it will help you. Best of luck
In EFA the correlation among a group of observed variables are identified and transforms into a small number of related factor(s). In brief, EFA
capture the groups of observed variables which are consistently moving together. In this process the consistent
movements of observed variables are identified through factor extraction and factor rotation.
Thus, a negative worded question will move in opposite direction (negative correlation). Thus, re-code your negative worded questionnaires.
Attached link may help further.
Article An Easy Approach to Exploratory Factor Analysis: Marketing P...
- Badges
- Science topic