Thursday, July 23, 2015

Survey analysis

I completed an example of the church life survey that I prepared earlier [link]. Now comes the analysis.

I need to do more work to establish the domains of church capability, Christian engagement, church engagement, church inclusiveness (broadly these are the categories I had in mind), then they can be analysed. The response scale is in the form of a Likert scale, so its probably OK to assume that the scale is an equal interval scale for practical purposes (if not, and you believe that the ordered discrete variables in fact are discrete approximations of underlying variables with "normal", continuous distributions, you might want to have a look at Stas Kolenikov's Stata command -polychoric-: The command enables you to estimate polychoric correlation on your "Likert type variables" without assuming true interval scale measurements; this is probably theoretically neater).

However, this is maybe going too far for the typical church.

A simpler approach is to count up the number in each scale item for each question, then note the 'mode' (the item with the most scores), and the total number off mode. This alone should give an experienced minister the drift of congregational sentiment.

A little more sophisticated approach would look at the mode by age group and duration of membership. This could also throw up relationships between the Christian engagement, church engagement and church inclusiveness groups of questions. 

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.