Reliability Improvement Analysis
Reliability Improvement
Reliability of a scale is improved by pruning of weak items. Deleting a weak item from a measurement scale may sometimes lead to jeopardizing scale’s theoretical construct. When this happens, the item needs to be redesigned and revaluated. Wording of items in a measurement scale plays an important role in determining its discriminating power. Weak items in the scale are identified by computing Corrected Item-Total Correlation (CITC) (Ferketich, 1991) and Item-Deleted Cronbach’s Alpha (IDCA) (Santos, 1999) as explained hereunder.
CITC means correlation of an item with the scale’s total after removing the item. It is computed by using SPSS. CITC enables identification of weak items. Higher than 0.8 value of CITC indicates …show more content…
The process of improvement was continued till until all the possible improvements were realized. Comparison was made between the older and the newer versions of SRD scale. Results and Discussions
Results obtained by scheduling and executing the above-mentioned assessment and improvement activities are reported and discussed in the order of their execution.
Item Review for NORS Version of SRD
Item review of NORS version of SRD scale did not offer any cue for improvement. Feedback from respondents also did not throw new light on item wordings except for suggesting minor editorial correction of replacing “is conflicting” with “conflicts” in the last three statements. Face validity of this version of SRD scale, therefore, could not be doubted.
Labelled as SRD-1, the updated NORS version of SRD scale was taken up for enhancement in this study. Labelled as 1a, 1b, 1c, 1d, 1e, 1f, 1g, SRD-1 comprised the following items.
(1a) My role demands me to do what is against my judgment;
(1b) My role provides me opportunities to use my expertise;
(1c) I would like to do things for the organization which are quite different from what I am doing in my …show more content…
Reliability Assessment of SRD-1
SRD was measured on 222 respondent described above, using SRD-1 scale. Reliability assessment of the scale was done after inverting the scores for reverse scoring items (1b and 1d).
(Place Table-1 about here)
Table-1 furnishes the result of SRD-1 reliability assessment. The scale has unacceptable reliability with CA = 0.323. CITC values for items 1a, 1e, 1f & 1g are in the acceptable range. CITC values for items 1b and 1d are negative and that for item 1c is lower than 0.3 indicating weakness of items Ib, Ic and Id. IDCA for items 1a, 1c, 1e, 1f, 1g are lower than scale CA indicating strength of these items. IDCA for items 1b and 1d, on the other hand are higher than scale CA, reconfirming the weakness of 1b and 1d.
Validity Assessment of SRD-1
There can be no acceptable validity with unacceptable reliability. EFA was done for SRD-1, however, to study its factor structure and variance explained. Two factors were obtained (instead of one) with Eigen values of 3.390 and 1.117, explaining the variance of 35.062% and 29.313% respectively. Cumulative variance explained by the two factors was 64.374%. SRD-1 has obviously failed in its validity assessment.
Enhancement of
Reliability of a scale is improved by pruning of weak items. Deleting a weak item from a measurement scale may sometimes lead to jeopardizing scale’s theoretical construct. When this happens, the item needs to be redesigned and revaluated. Wording of items in a measurement scale plays an important role in determining its discriminating power. Weak items in the scale are identified by computing Corrected Item-Total Correlation (CITC) (Ferketich, 1991) and Item-Deleted Cronbach’s Alpha (IDCA) (Santos, 1999) as explained hereunder.
CITC means correlation of an item with the scale’s total after removing the item. It is computed by using SPSS. CITC enables identification of weak items. Higher than 0.8 value of CITC indicates …show more content…
The process of improvement was continued till until all the possible improvements were realized. Comparison was made between the older and the newer versions of SRD scale. Results and Discussions
Results obtained by scheduling and executing the above-mentioned assessment and improvement activities are reported and discussed in the order of their execution.
Item Review for NORS Version of SRD
Item review of NORS version of SRD scale did not offer any cue for improvement. Feedback from respondents also did not throw new light on item wordings except for suggesting minor editorial correction of replacing “is conflicting” with “conflicts” in the last three statements. Face validity of this version of SRD scale, therefore, could not be doubted.
Labelled as SRD-1, the updated NORS version of SRD scale was taken up for enhancement in this study. Labelled as 1a, 1b, 1c, 1d, 1e, 1f, 1g, SRD-1 comprised the following items.
(1a) My role demands me to do what is against my judgment;
(1b) My role provides me opportunities to use my expertise;
(1c) I would like to do things for the organization which are quite different from what I am doing in my …show more content…
Reliability Assessment of SRD-1
SRD was measured on 222 respondent described above, using SRD-1 scale. Reliability assessment of the scale was done after inverting the scores for reverse scoring items (1b and 1d).
(Place Table-1 about here)
Table-1 furnishes the result of SRD-1 reliability assessment. The scale has unacceptable reliability with CA = 0.323. CITC values for items 1a, 1e, 1f & 1g are in the acceptable range. CITC values for items 1b and 1d are negative and that for item 1c is lower than 0.3 indicating weakness of items Ib, Ic and Id. IDCA for items 1a, 1c, 1e, 1f, 1g are lower than scale CA indicating strength of these items. IDCA for items 1b and 1d, on the other hand are higher than scale CA, reconfirming the weakness of 1b and 1d.
Validity Assessment of SRD-1
There can be no acceptable validity with unacceptable reliability. EFA was done for SRD-1, however, to study its factor structure and variance explained. Two factors were obtained (instead of one) with Eigen values of 3.390 and 1.117, explaining the variance of 35.062% and 29.313% respectively. Cumulative variance explained by the two factors was 64.374%. SRD-1 has obviously failed in its validity assessment.
Enhancement of