Achenbach System of Empirically Based Assessment (ASEBA) T-scores
Source:vignettes/articles/aseba.Rmd
aseba.RmdIntroduction
The Achenbach System of Empirically Based Assessment, ASEBA (Achenbach, 2017; Achenbach, 2019), is a widely used, evidence-based tool designed to assess behavioral, emotional, social, and adaptive functioning in individuals aged 1½ to 90+ years. Developed by Thomas Achenbach in response to the limitations of early psychiatric classifications like the DSM-I, it employs a multi-informant approach to capture behavioral patterns across different settings (e.g., home, school) and perspectives (e.g., parents, teachers, self-reports).
T-score
T-scores are standardized scores that indicate how an individual’s performance compares to an age and sex matched normative sample. These scores are derived by converting raw scores into a standardized format with a mean of 50 and a standard deviation of 10. To establish these norms, ASEBA collected data from large, representative samples across various societies, taking into account factors such as age, sex, and cultural background. This extensive normative data allows for the comparison of an individual’s scores to those of a relevant peer group, facilitating the identification of behavioral, emotional, social, and thought problems (Achenbach, 2019).
ABCD ASEBA forms
The ABCD Study® uses six ASEBA forms:
- Child Behavior Checklist (CBCL)
- Adult Behavior Checklist (ABCL)
- Youth Self-Report (YSR)
- Adult Self-Report (ASR)
- Brief Problem Monitor (BPM; Youth)
- Brief Problem Monitor (BPM; Teacher)
Note: For the 6.0 release, no T-scores are computed for the ASR form (raw scores are available in the tabulated data). This may change in future releases.
Usage
To calculate the T-scores for the ASEBA forms, one must obtain the
norming tables. Since these tables are a proprietary product licensed by
ASEBA, they are not included in the ABCDscores package.
Therefore, we will demonstrate how to compute the T-scores using a
simulated norming table. To use the code in a real-world setting, you
would need to obtain the norming tables from ASEBA.
The norming table
The ASEBA norm table needs to be organized as a dataframe with the following columns:
-
sex: character, either"1"for male or"2"for female. -
age_min: numeric, the minimum age of the norm group. -
age_max: numeric, the maximum age of the norm group. -
scale_r: numeric, the raw score. -
scale_t: numeric, the T-score.
Here is an example for a simulated norming table:
data_norm <- tibble::tibble(
sex = c("1", "1", "1", "1", "1"),
age_min = 18,
age_max = 35,
scale_r = 0:4,
scale_t = 20:24
)
data_norm
#> # A tibble: 5 × 5
#> sex age_min age_max scale_r scale_t
#> <chr> <dbl> <dbl> <int> <int>
#> 1 1 18 35 0 20
#> 2 1 18 35 1 21
#> 3 1 18 35 2 22
#> 4 1 18 35 3 23
#> 5 1 18 35 4 24Compute T-scores
To compute the T-scores, we can use the
compute_<score_name>() functions. The input and
output and some important arguments of these functions are very similar
to other score functions in the ABCDscores package. Please
refer to the Get started page for an
overview of how to use these functions.
The T-score functions have the following additional arguments:
-
data_norm: tibble, the norming table (as the one generated above). -
col_age: character, specifies the column name in the input data that contains the age of the participants. This is required to match the age of the participant to the age range in the norming table. For example, for scoremh_p_abcl__afs__frnd_tscore, the default value ofcol_ageis"mh_p_abcl__cg2__age_001". Please refer to the reference pages to see the default value ofcol_agefor a given score. -
col_sex: character, specifies the column name in the input data that contains the participants’ age. Similarly, this information is required to look up the respective T-score. For example, for scoremh_p_abcl__afs__frnd_tscore, the default value ofcol_sexis"mh_p_abcl__cg2_gi". Please refer to the reference pages to see the default value ofcol_sexfor a given score.
For data preparation, please also refer to the Get started page. After creating a dataset
containing the needed columns, downloading it, and reading it into a
data frame, the data should be ready for the T-score computation. For
example, assuming a data frame containing all needed columns called
data and the norming table called data_norm,
we can execute
compute_mh_p_abcl__afs__frnd_tscore(data, data_norm)to compute the T-score mh_p_abcl__afs__frnd_tscore for
the ABCL.
Examples
As we cannot provide the actual norming tables as part of this
package, we cannot run the code above. Here we demonstrate how to
compute T-scores using a simulated norming table and a simulated data
table using the lower level function ss_tscore().
All specific compute...() T-score functions use the
lower level function ss_tscore(). The only difference
between scale-specific T-score functions are default values, such as
col_age and col_sex.
Create a simulated data table:
data <- tibble::tibble(
var1 = c(0, 1, NA, 1, 2, 3),
var2 = c(1, 2, 1, 2, 5, 3),
age = c(18, 20, 25, 99, 35, 30),
sex = c("1", "1", "1", "1", "1", "2")
)
data
#> # A tibble: 6 × 4
#> var1 var2 age sex
#> <dbl> <dbl> <dbl> <chr>
#> 1 0 1 18 1
#> 2 1 2 20 1
#> 3 NA 1 25 1
#> 4 1 2 99 1
#> 5 2 5 35 1
#> 6 3 3 30 2Compute T-scores using the simulated norm and data tables:
ss_tscore(
data = data,
data_norm = data_norm,
col_age = "age", # specify the age column
col_sex = "sex", # specify the sex column
vars = c("var1", "var2") # specify individual items summarized in the score
)
#> # A tibble: 6 × 5
#> var1 var2 age sex tscore
#> <dbl> <dbl> <dbl> <chr> <int>
#> 1 0 1 18 1 21
#> 2 1 2 20 1 23
#> 3 NA 1 25 1 21
#> 4 1 2 99 1 NA
#> 5 2 5 35 1 NA
#> 6 3 3 30 2 NAIf ss_tscore() cannot find a match in the norming table,
it will return NA for the T-score. For example, in the
above example, the scores for the following rows are
NA:
- row 4: because the age is 99, which is not in the age range [18, 35].
- row 5: because the raw score is 5, which makes the raw score sum 7, which is not in the raw score range [0, 4].
- row 6: because the sex is “2”, which is not included in the simulated norming table.
Customize T-score computation
We can use the max_na and exclude arguments
to modify certain aspects of the T-score computation.
-
The
max_naargument specifies the maximum number of missing values that are allowed to still compute the T-score. For example, we can setmax_na = 0to allow no missing values in the summarized items (the default forss_tscore()isNULL, i.e., no limit, but the specific functions for the ASEBA T-scores impose a limit). If we rerun the code above with this change,NAis returned for row 3 becausevar1isNAin that row.ss_tscore( data = data, data_norm = data_norm, col_age = "age", col_sex = "sex", vars = c("var1", "var2"), max_na = 0 # specify maximum number of missing values ) #> # A tibble: 6 × 5 #> var1 var2 age sex tscore #> <dbl> <dbl> <dbl> <chr> <int> #> 1 0 1 18 1 21 #> 2 1 2 20 1 23 #> 3 NA 1 25 1 NA #> 4 1 2 99 1 NA #> 5 2 5 35 1 NA #> 6 3 3 30 2 NA -
The
excludeargument excludes certain values, i.e, converts them toNAbefore the score computation. For example, we can setexclude = "0"to exclude any"0"values from the computation. If we rerun the code above with this change,NAis returned for row 1 becausevar1is"0"in that row.ss_tscore( data = data, data_norm = data_norm, col_age = "age", col_sex = "sex", vars = c("var1", "var2"), max_na = 0, exclude = c("0") # specify values to exclude from the computation ) #> # A tibble: 6 × 5 #> var1 var2 age sex tscore #> <dbl> <dbl> <dbl> <chr> <int> #> 1 0 1 18 1 NA #> 2 1 2 20 1 23 #> 3 NA 1 25 1 NA #> 4 1 2 99 1 NA #> 5 2 5 35 1 NA #> 6 3 3 30 2 NA