Question

This exercise must be completed on the four subscales below, and you should, therefore, produce four reliability analyses. o Subscale 1 (Fearof statistics): items 1, 3, 4, 5, 12, 16, 20, 21 o Subscale 2 (Peerevaluation): items 2, 9, 19, 22, 23 o Subscale 3 (Fearof computers): items 6, 7, 10, 13, 14, 15, 18 o Subscale 4 (Fear of mathematics): items 8, 11, 17

Answer 1

Reliability Analysis on SPSS

To conduct each reliability analysis on these data you need to follow the Analyze⇒Scale⇒Reliability Analysis … menu path to display the dialog box . Select any items from the list that you want to analyze (to begin with let’s do the items from the fear of computers subscale) on the left hand side of the dialog box and transfer them to the box labelled Items.

Selecting the List item labels checkbox will list all of the variable labels for each variable (which can be useful for checking to which items your variables relate). There are several reliability analyses you can run, but the default option is Cronbach’s alpha, which is the one we want.

If you click on "Scale if item deleted" you can access the dialog box.

In the statistics dialog box you can select several things, but the one most important for questionnaire reliability is: Scale if

item deleted. This option provides a value of Cronbach’s alpha for each item on your scale. It tells us what the value of alpha would be if that item were deleted. If our questionnaire is reliable then we would not expect any one item to greatly affect the overall reliability. In other words, no item should cause a substantial decrease in alpha. If it does then we have serious cause for concern and you should consider dropping that item from the questionnaire. As 0.8 is seen as a good value for alpha, we would hope that all values of alpha if item deleted should be around 0.8 or higher.

Use the simple set of options to run a basic reliability analysis. Click on "Correlations" return to the main dialog box and then click "OK" to run the analysis.

It shows the results of this basic reliability analysis for the fear of computing subscale. The values in the column labelled Corrected Item-Total Correlation are the correlations between each item and the total score from the questionnaire. In a reliable scale all items should correlate with the total. So, we’re looking for items that don’t correlate with the overall score from the scale: if any of these values are less than about .3 (depends slightly on your sample size—with bigger samples smaller correlation coefficients are acceptable) then we’ve got problems because it means that a particular item does not correlate very well with the scale overall. Items with low correlations may have to be dropped. For these data, all data have item-total correlations above .3, which is encouraging. The values in the column labelled Alpha if Item is Deleted are the values of the overall alpha if that item isn’t included in the calculation. As such, they reflect the change in Cronbach’s alpha that would be seen if a particular item were deleted. The overall alpha is .823, and so all values in this column should be around that same value. We’re looking for values of alpha greater than the overall alpha because if the deletion of an item increases Cronbach’s alpha then this means that the deletion of that item improves reliability. None of the items here would substantially affect reliability if they were deleted. The worst offender is question 10: deleting this question would increase the alpha from .823 to .824. Nevertheless this increase is not dramatic and both values reflect a reasonable degree of reliability.

Finally, and perhaps most important, the value of Alpha at the very bottom is The Cronbach’s alpha: the overall reliability of the scale. To re-iterate we’re looking for values in the magnitude of .7 to .8 (or there about) bearing in mind what we’ve already noted about effects from the number of items. In this case alpha is slightly above .8, and is certainly in the region indicated by Kline, so this probably indicates good reliability. As a final point, it’s worth noting that if items do need to be removed at this stage then you should re-run your factor analysis as well to make sure that the deletion of the item has not affected the factor structure.

OUTPUT

R E L I A B I L I T Y A N A L Y S I S - S C A L E (A L P H A)

Correlation Matrix

Q06 Q07 Q10 Q13 Q14 Q15 Q18

Q06 1.0000

Q07 .5136 1.0000

Q10 .3222 .2837 1.0000

Q13 .4664 .4421 .3020 1.0000

Q14 .4022 .4407 .2547 .4498 1.0000

Q15 .3599 .3914 .2952 .3422 .3801 1.0000

Q18 .5133 .5009 .2925 .5329 .4983 .3429 1.0000

N of Cases = 2571.0

Item-total Statistics

Scale Scale Corrected

Mean Variance Item- Squared Alpha

if Item if Item Total Multiple if Item

Deleted Deleted Correlation Correlation Deleted

Q06 15.8650 17.6141 .6187 .3981 .7906

Q07 15.1684 17.7370 .6190 .3949 .7905

Q10 15.8114 20.7360 .3999 .1665 .8239

Q13 15.6429 18.8086 .6067 .3844 .7937

Q14 15.2159 18.7188 .5768 .3504 .7980

Q15 15.3259 19.3217 .4913 .2497 .8119

Q18 15.5235 17.8324 .6474 .4475 .7855

Reliability Coefficients 7 items

Alpha = .8234 Standardized item alpha = .8214