Case 2: Measuring the fear of falling of elderly
people
Research questions
In 2004, a random sample of 300 persons, aged 70 and older, was drawn from the
population registry of the city Heerlen. Next to the items of the Falls Efficacy Scale-
International (FES-I) scale, the questionnaire contained questions on the
respondent’s gender, age, number of fall incidents in the previous year and the
respondent’s general concern to fall. These measurements comprise the T1
measurement. Four weeks later the questionnaire was sent again to the same
respondents, this time only containing the FES-I items. This is denoted as the T2
measurement.
The variables T1fesi1_imp up to T1fesi16_imp contain the scores on items of the
FES-I for the first measurement (T1) and T2fesi1_imp up to T2fesi16_imp for the
second measurement (T2). If persons filled out at least 13 of the 16 items of the
scale, their missing values have been imputed through stochastic regression
imputation. After these imputations, on T1 208 respondents, and on T2 194
respondents had scores on all items of the scale.
The main research question:
Is the FES-I a reliable and valid measurement instrument for measuring fear of
falling among elderly people?
In addressing this question, the following issues should be examined (issues a to d need
only be investigated for the T1 measurements):
a. Perform an internal consistency analysis of the FES-I. Examine whether items should
be removed from the scale. If necessary, remove items one by one. Motivate removal of
an item not only on statistical grounds, but also based on the contents of the item. How
do you evaluate the internal consistency of the final scale (that is, after possibly deleting
some “bad” items)?
people
Research questions
In 2004, a random sample of 300 persons, aged 70 and older, was drawn from the
population registry of the city Heerlen. Next to the items of the Falls Efficacy Scale-
International (FES-I) scale, the questionnaire contained questions on the
respondent’s gender, age, number of fall incidents in the previous year and the
respondent’s general concern to fall. These measurements comprise the T1
measurement. Four weeks later the questionnaire was sent again to the same
respondents, this time only containing the FES-I items. This is denoted as the T2
measurement.
The variables T1fesi1_imp up to T1fesi16_imp contain the scores on items of the
FES-I for the first measurement (T1) and T2fesi1_imp up to T2fesi16_imp for the
second measurement (T2). If persons filled out at least 13 of the 16 items of the
scale, their missing values have been imputed through stochastic regression
imputation. After these imputations, on T1 208 respondents, and on T2 194
respondents had scores on all items of the scale.
The main research question:
Is the FES-I a reliable and valid measurement instrument for measuring fear of
falling among elderly people?
In addressing this question, the following issues should be examined (issues a to d need
only be investigated for the T1 measurements):
a. Perform an internal consistency analysis of the FES-I. Examine whether items should
be removed from the scale. If necessary, remove items one by one. Motivate removal of
an item not only on statistical grounds, but also based on the contents of the item. How
do you evaluate the internal consistency of the final scale (that is, after possibly deleting
some “bad” items)?
