The mean score is computed by adding the item responses for a particular
subscale, and then dividing by the number of completed items. The mean score
can range from 1 through 4, and higher scores indicate a greater potential to
foster. So, for example, a score of 3 indicates that, on average, the worker or
applicant answered "agree" (or "disagree", in the case of a reverse scored
items) to the statements measuring that particular dimension.
Mean scores give a general impression of a person's potential to foster using
the intuitively understandable disagreement/agreement scale used to rate the
individual items. They're also used to derive the normative scores discussed
below. In addition, mean scores, not normative scores, should be used to
summarize scores for groups of applicants or for research purposes.
Percentile ranks
The percentile rank indicates the percentage of people in the normative sample
who are at or below the mean score. A higher percentile rank suggests greater
fostering potential, relative to those in the normative sample. For example,
someone with a mean score corresponding to a percentile rank of 75 scored
higher than 75% of the people in the normative sample; someone with a mean
score corresponding to a percentile rank of 25 scored higher than only 25% of
those in the normative sample.
In making decisions about strengths and areas for development and support we
suggest you consider someone with a percentile rank greater than 75 as having a
special strength in a particular area, and someone with a percentile rank below
25 as needing further development and support in that area. However, note that
the probability of misclassification is high for applicants with scores close
to the margins of each category (e.g., someone with a percentile rank from 23
to 27 or from 73 to 77). When this happens be very cautious about classifying
someone in one category or the other, or don't classify him or her at all.
Percentile ranks simply tell us whether someone is higher or lower than someone
else, but not by how much. Small differences in the potential to foster can
lead to large differences in percentile ranks and visa versa. Typically a
change over time or a difference between people with percentile ranks of 5 and
10 is much greater than a difference of 50 and 55. So, percentile ranks
shouldn't be used to determine the amount of change over time, differences
among people, or differences among subscales for a particular person.
Do not confuse percentile ranks with percentage correct, a type of raw score
most often reported for tests. Percentage correct indicates the percentage of
items on a test that were answered correctly. Percentile ranks refer to the
percentage of people in a normative sample who are at or below a given raw
score.
T-scores
Higher T-scores represent greater fostering potential, relative to those in the
normative sample. T-scores below 50 indicate less than average potential, and
T-scores greater than 50 indicate greater than average potential. Someone with
a T-score below 50 scored below the mean of the normative sample, someone with
a T-score of 50 scored at the mean, and someone with a T-score above 50 scored
above the mean.
All measuring tools have some measurement error. For example, think about
weighing yourself on a bathroom scale several times in one day and each time
getting slightly different results, all of which are pretty close to your true
weight that day. The score report presents standard error notations
to illustrate the range in which each true T-score probably falls, with the
obtained score in the middle.
The error notations can be used to examine differences among people (e.g., a
worker and a foster mother applicant), or differences among subscales for a
person. When two error ranges overlap it suggests that the true scores aren't
really that different, but when they don't overlap it suggests a genuine
difference. This is how you can tell if there are genuine differences between
subscale scores for a person, or differences between, say, a worker and a
foster mother on one of the CFAI subscales.
