8 B measurements and errors in measurements

variable

is any measurable characteristic that can take on different values.

study variable

is the variable of interest in a specific research

Scale of measurement

(nominal, ordinal, interval, ratio)

Type

(qualitative vs. quantitative)

Behavior

(continuous vs. discrete)

interview

questionnaire

observation

review of record

methods of using study variables

In-person interview

Telephone or online interview

types of interview

interview

advantage

• Clarification of unclear responses

• Better response rate

• Observing nonverbal cues

limitations

• Time consuming

• Potential influence (response bias)

questionnaire

Participants answer written questions without an interviewer

questionnaire

advantages

• Efficient for large samples

• Anonymity can improve accuracy

• More standardized response

limitation

• Misinterpretation of questions

• Lower response rate than interviews

• No opportunity to probe or clarify

OBSERVATION

Recording behaviors or characteristics directly.

Direct observations

Observation using equipment

types of observation

observation

advantage

• Real - time, objective data

• Useful for behaviors or clinical
  measurements

limitation

• Expensive

• Observer effect (Hawthorne phenomenon)

review of records

Using existing documents (clinical charts, registries).

review of records

advantage

• Quick and inexpensive

• Historical data available

limitations

• Quality depends on source records

• Incomplete or inaccurate documentation

measurement error

is any deviation between the measured value and the true value.

study participant

observer

instrument

data processing

sources of errors

Study Participant Errors

• Subjective responses influenced by memory gaps, embarrassment, or social desirabilly

• Biologic variability (e.g., fluctuating blood pressure)

Observer Errors

• Prior knowledge or expectations influencing measurement

• Non-neutral behavior

• Failure to follow standard protocols

• Incorrect transcription or data encoding

• Different clinicians applying criteria inconsistently (diagnostic variability)

Instrument Errors

Poorly designed questionnaires

Unclear instructions

Calibration issues in equipment

Mechanical failure

Problems combining subscale items into a score (index problems)

Data Processing Errors

Mistakes during data entry or coding

Misuse of statistical software

Inaccurate transcription of results

validity and reliability

To ensure data quality, researchers assess ___ and ___

VALIDITY (ACCURACY)

- "Are we measuring what we are supposed to measure?"

Face validity

looks valid on surface

Content validity

all relevant domains included

Construct validity

aligns with theoretical concepts

Criterion validity

comparison with a gold standard. It is the basis for sensitivity and specificity

RELIABILITY (CONSISTENCY)

"Are measurements repeatable?"

SENSITIVITY

- Proportion of people with the disease who test positive.
A highly sensitive test → good for screening → few missed cases.

SPECIFICITY

• Proportion of people without the disease who test negative.

• A highly specific test → good for confirmation

• → few false alarms.

False Positive (FP)

Non-diseased labeled as diseased

False Negative (FN)

Diseased labeled as disease-free

Positive Predictive Value (PPV)

Probability that a positive test truly indicates disease

Negative Predictive Value (NPV)

Probability that a negative test truly indicates absence

cut off vlues

disease prevalence

factors affecting sensitivity and specificity

Lower cut-off

• higher sensitivity, lower specificity

Higher cut-off

• higher specificity, lower sensitivity

Higher prevalence

→ more true positives → improved sensitivity

Higher prevalence

→ more individuals classified as diseased → possibly lower specificity