Stata automatically reversed my data

[Noralie vt Hul]

Hi all,

Can anybody help me out how I can prevent STATA from reversing the data of aut3? Students filled in a survey with a five points lickert scale and I already reversed the data of the negatively asked questions myself (before entering the data in STATA).

. alpha aut1 aut2 aut3 aut4 aut5 aut6 aut7 aut8

Test scale = mean(unstandardized items)
Reversed item: aut3

Average interitem covariance: .2150992
Number of items in the scale: 8
Scale reliability coefficient: 0.7011

. alpha aut1 aut2 aut3 aut4 aut5 aut6 aut7 aut8, item

Test scale = mean(unstandardized items)

average
item-test item-rest interitem
Item Obs Sign correlation correlation covariance alpha

aut1 104 + 0.6266 0.4687 .2044125 0.6554
aut2 104 + 0.5809 0.4318 .2171085 0.6653
aut3 104 - 0.4409 0.2481 .2414382 0.7019
aut4 104 + 0.7026 0.5664 .1891559 0.6331
aut5 104 + 0.6260 0.4738 .2058395 0.6550
aut6 104 + 0.5828 0.3697 .2083645 0.6798
aut7 104 + 0.4360 0.2311 .2423806 0.7069
aut8 104 + 0.5751 0.3840 .2120941 0.6742

Test scale .2150992 0.7011


. summarize aut3

Variable Obs Mean Std. Dev. Min Max

aut3 104 2.365385 .9353644 1 5

[Joseph Coveney]

Have you tried the asis option?

alpha aut1 aut2 aut3 aut4 aut5 aut6 aut7 aut8, item asis

[Noralie vt Hul]

Great thanks! Didn't know about this option. If I use the code you suggested, I get the following results. I already expected a low alpha with question 3 present, at least now I can argue why I wanna leave this question out.

Test scale = mean(unstandardized items)

average
item-test item-rest interitem
Item Obs Sign correlation correlation covariance alpha

aut1 104 + 0.6074 0.4222 .1298899 0.5237
aut2 104 + 0.6191 0.4593 .1308857 0.5185
aut3 104 + -0.0209 -0.2481 .2414382 0.7019
aut4 104 + 0.6680 0.5005 .118972 0.4986
aut5 104 + 0.5899 0.4072 .1337174 0.5297
aut6 104 + 0.6242 0.3943 .123084 0.5273
aut7 104 + 0.3842 0.1465 .1705786 0.6063
aut8 104 + 0.6075 0.3992 .1278984 0.5276

Test scale .147058 0.5947


.