In grad school, I am becoming re-acquainted with the Assessment in Speech-Language Pathology..., 5th ed. (ASLP-5) by Shipley & McAfee. The first closed-book/note test upon me, I need to connect the dots from Ch. 1 - 3, so here it goes:
Ch. 1 Foundations of Assessment
Overview - a good assessment = one that is thorough, uses variety of assessment modalites, valid, reliable, and is tailored to the individual client.
This entails, 1.obtaining
historical info on client, family/caregiver(s) & nature of disorder;
2.
interview with client &/or family/caregiver(s), or both;
3.
evaluating the structural and fxn'l integrity of oralfacial mechanism;
4. taking adequate
speech-language sample(s) and evaluating it/them in light of use/abilities of the various areas of communication - in the case of dysphagia, assessing client's chewing/swallowing abilities;
5.
screening clients hearing abilities (or obtaining eval info on this);
6.
evaluating assessment info to determine impressions, diagnosis or conclusions, prognosis, and recommendations;
7.
sharing clinical findings thru and interview with client, formal (written) report or informally, e.g., phone contact w/physician.
Keep in mind - some disorders have extensive history (hx), others do not; Client(s) (Cx) has/have different primary communicative problems. Some cases require extensive interviewing or more detailed report writing, whereas others do not.
* No matter the lengths required for this effort, some consideration for each of the 7 areas listed above is necessary...
NORMED-REF tests
Should be normally distributed along the bell curve
Should be standardized (scorable to std. & %ile)
Should be normed to a population from which a rep sample has been taken
Should be
valid:
- Face: tests what it claims to
- Content: rep of the content domain it sets out to test, e.g., articulation
- Construct: measures pre-determined theoretical content, e.g., PLS-5 tests skills that show, based on empirical observation, skills that the typical child will master by the various ages...
Criterion: established by external criterion -
-- Concurrent: in comparison to a widely accepted standard, e.g., GFTA-2 and tests created thereafter, or Stanford-Binet IQ test...
-- Predictive: test should be able to show what Cx is able to do down the road, e.g., GRE purports to show how students will perform in grad school...
Reliability = results are replicable
- Test-retest: refers to test's stability over time
- Split-half: refers to tests internal consistency
-- scores from one 1/2, e.g., odd #s are compared against those of the other 1/2, e.g., even #s of test.
- Rater: refers to level of agmt among indiv rating test
-- intra - establishes results consistent by one indiv over time
-- inter - establishes results consistent by multiple indiv over time
Alternate form (aka, parallel form): refers to test's coefficient with a similar test, or do they both produce similar results when admin'd to similar groups of ppl.
Standardization(aka, formal): std != to norm-ref - it simply means a score is attached (but not necessarily reviewed for validity/reliability).
* What to look for in testing manuals:
- Purpose(s) of test
- Age range it's designed and stdzed for
- Test constrxn & development
- Admin & scoring procedures
- Normative sample group and statistical info derived from it
- Test VALIDITY/RELIABILITY
CHRONOLIGICAL AGE OF CX: how to establish.
* I noticed a variance among the ASLP-5 and some automated (Pearson, Superduperinc) forms out there. I'll need to clarify what constitutes a month, how it all applies (borrow previous/later one), if Feb is it 28 days, or 29 + for leap year, and banker's year (30 days for every month), 31 days, or the actual amount of the month borrowed from...
BASAL: what's the starting point for the test?
CEILING: what's the test's ending point?
ACCOMMODATIONS/MODIFICATION OF STDZED TESTING:
** Stdzed admin.: Follows protocol to the letter
- Accommodations: minor adjustments to testing situation that don't comprise stdzed procedure, e.g., large print version, etc.
- Modifications: changes to testing protocol, e.g., re-wording test, allowing extra time, etc. -
Understanding Stdzd Scores:
- Z- score: tells how many std deviations (SDs) a stdzed score (SS) is from mean
- Stanine = 9-unit scale where a score of 5 = ave. performance. Each score, exc. 1 & 9, is equally distributed across the curve
- Confidence interval: rep. degree of certainty on the part of test developer - the higher the better, e.g., 95%. Thus, a range or reliable scores, not just a single one.
- Age/Grade equivalency: least useful of the SS and most misleading = the average raw scores on a test...
Health Insurance Portability and Accountability Act (HIPAA)
Fed law designed to improve health care system by
Protecting privacy & confidentiality (among other things)
