Instrumental Measures of Physical Variables

* Air Flow

* Fundamental Frequency

* Electromyography

* Evoked Potentials

Observer Measurements of Behavioral Variables

* Clinician Perception (Used when making judgements)

* Transcription and analysis of language samples

* Formal Speech and language tests

Measurement

"Assignment of numbers or labels to objects or events to define their properties according to a set of rules. "

"Process of abstraction about an object or event because we measure its various attributes or features."

... when doing this- must have an ambiguous set of rules

Levels of Measurement

* Nominal Scales (name): EX: verbal/nonverbal, delayed, normal

* Ordinal Scale (< or >): EX: mild, moderate, severe

* Interval (equivalent units): EX: could give a test and get a standard score

* Ratio Scales (absolute zero): EX: MLU

Quality of Measurement

Key Factors that need to be Controlled

* Test Enviornment

* Instrument Calibration

* Instructions to participants

* Observer bias (non interactional effect and interactional effect)

Reliability

Synonyms: Dependability, Consistency, Predictability, Stability, Reproducibility.

Definition: "general trustworthiness of obtained data."

Interactional effect

Controlling Observer Bias

* Use multiple observers

* Minimize participant/observer contact

* Blinding procedures: Process of collecting data and performing other tasks without knowledge of research hypothesis.

* Double blinding: neither subjects or research team know who's assigned to control treatment.

Reliability

*Definiton: Degree to which we can depend on a measurement

- If we measure the same set of objects or subjects repeatedly with the same or comparable test instrument, will we get the same or similar results?

Measurement Precision

*Definition: "A precise measure can be expected to remain reasonably stable if the measurement procedure is repeated with the same participant)

Reliability of measurement

*Definition: ratio of true score variance to observed score variance. X=T+E

X=observed score

T= true score

E= random error

Any measurement has two components:

1. True score

2. Measurement error

* Reliability tells us about measurement error

More measurement error = greater variability in scores

Error measurement should be low with a reliable test

Types of Error influencing measurement reliability

1. Systematic

2. Unsystematic

Systematic Error

*Definition: Error that occurs with every repeated measurement

Unsystematic Error

Ways Error is Introduced

1. Characteristics of examinee

2. Behavior of examiner-scorer

3. Aspects of test content

4. Time factors

5. Situation factors

Sources of Error

1. poorly standardized instructions

2. scoring errors

3. subjectivity of measurement (rating scales are highly subjective)

4. Limited sampling of behavior

5. Fluctuation in participant's attention

6. guessing

7. instability

8. test environment

Stability

Test-Retest reliability

Assessed by measuring same individuals at two points in time

Equivalence Measures

Alternate form method (aka parallel form method)

*avoids potential carry-over effects

*principle limitation

Internal Consistency