measures the attribute -the less variation an instrument produces in repeated measurements of an attribute , the higher its reliability RELIABILITY -An instrument is reliable if its measures accurately reflect true scores. A reliable measure is one that maximizes the true score component and minimizes the error component of an obtained score. RELIABILITY Many psychosocial scales contain two or more subscales , each of which tap distinct, but related, concepts (e.g measure of independent functioning might include subscales for motor activities , communication and socializing.) The reliability of each subscale is typically assessed and,if subscale scores are summed for a total score, the scales overall reliability would be assessed. 3 ASPECTS OF RELIABILITY 1. STABILITY 2. INTERNAL CONSISTENCY 3. EQUIVALENCE 1. STABILITY -The stability of a measure is the extent to which the same scores are obtained when the instrument is used with the same people on separate occasions. 1. STABILITY TEST-RETEST RELIABILITY PROCEDURES -the researchers administers the same measure to a sample of people on two occasions, and then compares the scores. 1. STABILITY TEST-RETEST RELIABILITY PROCEDURES Ex. We are interested in the stability of a self-report scale that measured self-esteem in adolescents. Because self-esteem is a fairly stable attribute that would not change markedly from one day to the next, we would expect a reliable measure of it to yield consistent scores on two separate test. 1. STABILITY TEST-RETEST RELIABILITY PROCEDURES Ex. (cont.) As a check on the instruments stability, we arrange to administer the scale 3 weeks apart to a sample of teenagers. Table 1 Data for Test-retest Reliability of Self-Esteem Scale SUBJECT NUMBER TIME 1 TIME 2 1 55 57 2 49 46 3 78 74 4 37 35 5 44 46 6 50 56 7 58 55 8 62 66 9 48 50 10 67 63 r=.95 1. STABILITY TEST-RETEST RELIABILITY PROCEDURES Ex. (cont.) On the whole, differences on the two tests are not large. RELIABILITY COEFFICIENT=a numeric index of a measures reliability to objectively determine exactly how small the differences are. 1. STABILITY RELIABILITY COEFFICIENT -a numeric index of a measures reliability to objectively determine exactly how small the differences are. -designated as r range from .00-1.00* -the higher the value, the more reliable (stable) is the measuring instrument. In the example shown in Table 1, the reliability coefficient is .95 which is quite high. 1. STABILITY TEST-RETEST RELIABILITY PROCEDURES Disadvantage: -Many traits of interest change over time, independently of the instruments stability (attitude, mood, knowledge can be modified by experiences between 2 measurements) -Therefore, stability indexes are most appropriate for relatively enduring characteristics-personality and abilities. Example of test-retest reliability -Needham, Abderhalden, Dassen, Haug, Fischer (2004) developed and tested a short form of the PERCEPTION OF AGGRESSION SCALE, which is designed to measure psychiatric nurses attitudes towards, and perceptions of patient aggression. To select items with the strongest stability, the researchers administered the full scales to three groups of psychiatric nurses twice: to one group after 4 days, to the second after 14 days, and to the third after 70 days. The 12 items selected for the short form has a test-retest reliability of .77. 2. INTERNAL CONSISTENCY -An instrument may be said to have Internal consistency reliability to the extent that all its subparts measure the same characteristic. -This approach assesses an important source of measurement error in multi item measures: the sampling of items. 2. INTERNAL CONSISTENCY How to assess? A. SPLIT HALF TECHNIQUE -oldest -The items comprising a scale are split into two groups (usually odd vs even items) and scored, and then the scores on the two half-tests are used to compute reliability coefficient. If the two half-test are really measuring the same attribute, the reliability coefficient will be high. 2. INTERNAL CONSISTENCY How to assess? B. CRONBACHS ALPHA (or COEFFICIENT ALPHA) This method gives an estimate of the split-half correlation for all possible ways of dividing the measure into two halves, not just odd versus even items. 3. EQUIVALENCE -used primarily with observational instruments -determines the consistency or equivalence of the instrument by different observers or raters. 3. EQUIVALENCE Risk for observer error -the degree of error can be assessed through INTERRATER (OR INTEROBSERVER) RELIABILITY. 3. EQUIVALENCE INTERRATER (OR INTEROBSERVER) RELIABILITY -which is estimated by having 2 or more trained observers make simultaneous, independent observations. 3. EQUIVALENCE INTERRATER (OR INTEROBSERVER) RELIABILITY -(cont.) the resulting data can then be used to calculate an index of equivalence or agreement. -A reliability coefficient can be computed to demonstrate the strength of the relationships between the observers ratings. 3. EQUIVALENCE INTERRATER (OR INTEROBSERVER) RELIABILITY -When two independent observers score some phenomenon congruently, the scores are likely to be accurate and reliable. EXAMPLE: -Kovach (2002) observed the behaviors of older persons with dementia over 30 minute observation sessions. Interrater reliability, calculated as percentage agreement between 2 raters, was .74 for the variable activity, .92 for noxiousness, and .84 for agitation. INTERPRETATION OF RELIABILITY COEFFIECIENTS -Reliability coefficients are an important indicator of an instrument's quality. -A measure with low reliability prevents an adequate testing of research hypotheses. -If data fail to confirm a hypothesis, one possibility is that the measuring tool was unreliable-not necessarily that that the expected relationship do not exist. INTERPRETATION OF RELIABILITY COEFFIECIENTS -Knowledge about an instruments reliability thus is critical in interpreting research results, especially if research hypotheses are not supported. -Reliability estimates vary according to the procedures used to obtain them. Estimates of reliability computed by different procedures for the same instrument are not identical. INTERPRETATION OF RELIABILITY COEFFIECIENTS -reliability of an instrument is related to sample heterogeneity. The more homogenous the sample (i.e. the more similar the scores), the lower the reliability coefficient will be. This is because instruments are designed to measure differences, and if sample members are similar to one another, it is more difficult for the instrument to discriminate reliably among those who possess varying degrees of the attribute. INTERPRETATION OF RELIABILITY COEFFIECIENTS -longer instruments (i.e. those with more items) tend to have higher reliability than shorter ones. VALIDITY -