If you get positive correlation, respondents with higher time spent are disagree more( 5: disagree )
Positive correlation means high value with high value ( coding value)
Let’s start with the scale of measurement and your analysis. Your items are ordinal, but the chi-squared analysis that you are using treats these data as nominal. So, it is as if you were asking whether eye-color is related to hair-color; though perhaps a silly example, this is certainly a question that can reasonably be addressed with chi-squared. Thus, the contingency table application of chi-squared is not, unto itself, equipped to address your directionality question. Moreover, you are relinquishing the information vested in the ordinality of your items by treating them as nominal. This may result in several issues – missing the nature of relationships, loss of power, etc.
My argument is not that the chi-squared is “wrong,” rather that it is impoverished in respect to your data. In any case, if you wish to use the chi-squared, you can ascertain a notion of the “direction” of the effect by considering the difference between observed and expected values, and no, the reverse coding will make no difference. But do realize that those expected values represent expectations from probabilities calculated on the null assumption of independence of nominal variables.
The requirement for creating a scale from a set of items is that they all be scored in the same direction, which means that you have nothing but positive correlations. So, if all you items were asked in a "reverse" direction, that would make no difference.
- Badges
- Science topic
- Similar topics
- Social Science
- Media
- Social Media