Dilectual Analysis Of Cronbach's Alpha Values
Part I: Scale construction
After administering Draft I of scale to 25 individuals, it was seen that for some statements inter-item correlation with other statements was poor. These were modified or removed. E.g. “…unconcerned about speaking” was changed to “…concerned…”. This was because since a majority of average speakers do not consider speaking to be easy or do not admit to being ‘unconcerned’ about it. Thus the scores were not as low as should be expected from typical individuals. Similarly, based on the opinions of two experienced speech language pathologists as well as feedback obtained from participants, some minor modifications were made. E.g. “favourable” in the labels for one of the questions was changed to “positive” to make it easier to interpret the meaning.
Changes …show more content…
The Cronbach’s alpha values obtained from the pilot study using the final draft of the ISACS are given in Table 1.
(Insert Table 1 here)
Cronbach’s alpha values were acceptable[40] for three of the four subscales suggesting good internal consistency and reliability.[42,43] Cronbach’s alpha values for Section I were low for the PWF and SOP groups, a common occurrence for short scales with less than ten items.[44] Also, a high alpha value was not expected in this subscale as the items mainly elicit information about speech by describing the characteristics of a person’s speech. Thus a high score need not necessarily mean high impact for the Subscale I.
The response time varied between 15 to 20 minutes.
Part II: Translation
On a paired t-test, there was no significant difference in total ISACS scores (p=0.28) or subscale scores (p<0.05) obtained from English and Marathi versions on a paired t- test, confirming equivalence of the ISACS in both languages. This implied that both versions could be used interchangeably, depending on which of the two languages the participant was comfortable
After administering Draft I of scale to 25 individuals, it was seen that for some statements inter-item correlation with other statements was poor. These were modified or removed. E.g. “…unconcerned about speaking” was changed to “…concerned…”. This was because since a majority of average speakers do not consider speaking to be easy or do not admit to being ‘unconcerned’ about it. Thus the scores were not as low as should be expected from typical individuals. Similarly, based on the opinions of two experienced speech language pathologists as well as feedback obtained from participants, some minor modifications were made. E.g. “favourable” in the labels for one of the questions was changed to “positive” to make it easier to interpret the meaning.
Changes …show more content…
The Cronbach’s alpha values obtained from the pilot study using the final draft of the ISACS are given in Table 1.
(Insert Table 1 here)
Cronbach’s alpha values were acceptable[40] for three of the four subscales suggesting good internal consistency and reliability.[42,43] Cronbach’s alpha values for Section I were low for the PWF and SOP groups, a common occurrence for short scales with less than ten items.[44] Also, a high alpha value was not expected in this subscale as the items mainly elicit information about speech by describing the characteristics of a person’s speech. Thus a high score need not necessarily mean high impact for the Subscale I.
The response time varied between 15 to 20 minutes.
Part II: Translation
On a paired t-test, there was no significant difference in total ISACS scores (p=0.28) or subscale scores (p<0.05) obtained from English and Marathi versions on a paired t- test, confirming equivalence of the ISACS in both languages. This implied that both versions could be used interchangeably, depending on which of the two languages the participant was comfortable