CHAPTER 7: IMPORTANCE OF
RELIABILITY 1
Evaluation of an Individual’s Test Score
Point estimate: specific value, best guess/estimate of an individual’s
standing on a psy attribute.
Confidence interval: confidence interval reflects a range of values
that is often interpreted as a range in which the true score is likely to
fall. (because observed scores are just estimates of true scores)
Point estimates
1. Observed score, at a point in time the observed score is the person’s
point estimate (single best estimate of the psych. Attribute)
Best estimate of true score
2. Adjusted true score estimate, considers the measurement error.
If a person is measured in two occasions, we can use the observed
score of the first time to estimate the results of the second occasion
> adjusted score that reflects regression to the mean.
Best estimate for a predicted score on a test
Regression to the mean: likelihood that in the send testing the
individual’s score is more likely to be closer to the mean (based on CTT and
measurement error’s ability to inflate or deflate a score)
If first score was above mean, in the second time > lower score
If first score was below mean, in the second time > higher score
Adjusted true score estimate: reflects the discrepancy in individual’s
observed scores that is likely to arise between two testing occasions. The
discrepancy can be influenced by
1. Reliability of test scores
2. Size of difference between the individual’s original score and the
mean of test scores
3. Direction of that difference
Xest= X + Rxx( Xo−X )
The lowest the reliability > the bigger the difference between the
estimated true score and the observed true score.
o Which indicates that the difference is bigger when the test is more
influence by measurement error.
The extremity of the observed scores influences the difference
between the estimated true score and the observed true scores.
o More extreme scores (high or low) > larger difference
True score Confidence Intervals
Link between reliability & CI precision, through SEm
CI =Xo ± ( z ) ( SEm )
% of chance that true score fall in this interval which means
RELIABILITY 1
Evaluation of an Individual’s Test Score
Point estimate: specific value, best guess/estimate of an individual’s
standing on a psy attribute.
Confidence interval: confidence interval reflects a range of values
that is often interpreted as a range in which the true score is likely to
fall. (because observed scores are just estimates of true scores)
Point estimates
1. Observed score, at a point in time the observed score is the person’s
point estimate (single best estimate of the psych. Attribute)
Best estimate of true score
2. Adjusted true score estimate, considers the measurement error.
If a person is measured in two occasions, we can use the observed
score of the first time to estimate the results of the second occasion
> adjusted score that reflects regression to the mean.
Best estimate for a predicted score on a test
Regression to the mean: likelihood that in the send testing the
individual’s score is more likely to be closer to the mean (based on CTT and
measurement error’s ability to inflate or deflate a score)
If first score was above mean, in the second time > lower score
If first score was below mean, in the second time > higher score
Adjusted true score estimate: reflects the discrepancy in individual’s
observed scores that is likely to arise between two testing occasions. The
discrepancy can be influenced by
1. Reliability of test scores
2. Size of difference between the individual’s original score and the
mean of test scores
3. Direction of that difference
Xest= X + Rxx( Xo−X )
The lowest the reliability > the bigger the difference between the
estimated true score and the observed true score.
o Which indicates that the difference is bigger when the test is more
influence by measurement error.
The extremity of the observed scores influences the difference
between the estimated true score and the observed true scores.
o More extreme scores (high or low) > larger difference
True score Confidence Intervals
Link between reliability & CI precision, through SEm
CI =Xo ± ( z ) ( SEm )
% of chance that true score fall in this interval which means
