diametrically opposite to each other. For example, if an individual scores high on the

traditional self-identity index, can he/she also score high on the modem self-identity

index? Can farmers have both positive scores on both traditional and modern self-

identity? Are these categories mutually exclusive?

 Reliability accesses how consistently the instrument will produce the same results

(Carmines & Zeller, 1979). I am confident of the reliability of my scales and indices

because of the process I used to develop them. These instruments have high internal

consistency and are replicable. Internal consistency is the degree to which items in a

scale or index actually measures the same construct as intended to measure. Cronbach's

alpha (Sullivan, 2001) is a measure of internal consistency based on the correlation

between all possible split-halves of the items. It also provides the correlation between

each item and the total score of the scale or index and provides an overall coefficient of

scale reliability. I deleted items from the scale and indices that did not correlate well

with the other items, thus improving the reliability and raising the Cronbach's alpha

value.

 Unlike scales, index inter-item reliability is not determined a priori. These

instruments may exhibit lower reliability than scales because they assume that all items in

the instrument are related without testing if this is true a priori (M.E. Swisher, personal

communications, March 2006). I calculated internal consistency with Cronbach's alpha

and total inter-item correlation after data collection to determine if the items in the

indices were related to each other and to improve reliability of the indices. I removed

items from some of the indices to improve the reliability of instruments (Table 3-1).

I also improved the replicability of this study by developing clear operational definitions

of the variables and constructs.