Week 3

Week 3: Measurement and prediction 2

Hello,  this week shall we enjoy with the 4 keywords of the measurement and prediction. Let’s look closely into each word in terms of the definition and how to use them in social research.

👉Reliability * An instrument is reliable if it gives the same measurement every time it is used.
* Reliability is the degree to which an assessment tool produces stable and consistent results.

Establishing Reliability 

1. Internal consistency or Cronbach's alpha

2. Test- re-test; intraclass correlation

3. Intra-  rater reliability; intraclass correlation

4. Inter-  rater reliability; intraclass correlation

👉 Validity defined as an instrument is valid if ti measures what it purpose to measure.

Establishing Reliability

1. Face validity ; no formal criteria
2. Content Validity; Linn's criteria for proportion of experts where endosment ids required

3. Factorial validity or Factor analysis

4. Criterion
* Condurent ----- correlation with a measure of the same phenomenon (contemporaneous)
* Predictive ------correlation with another measure at a later time

5. Convergence validity(divergent)---- correlation or lack of correlation with measure of other phenomena predicted to correlate (or not) with the new scales

Discriminant Reliability  is ability to discriminate between case 

NB 

- The construct validity of no scale had ever been established

- All the tests described have rarely been done for one scale

- Psychometric work often takes many years such as 10, 15, or 20 years


It is said that the social research contained reliability alone is not sufficient. Indeed, it needs to be valid as well.

I would recommend to read more about this topic in this link https://chfasoa.uni.edu/reliabilityandvalidity.htm

👉Sensitivity and Specificity

Sensitivity is defined as the proportion of people with the disease who will have a positive result.

Specificity is the proportion of people without the disease who will have a negative result.

https://newonlinecourses.science.psu.edu/stat507/node/71/

👉Receiver Operating Characteristic or (ROC)

is the method can be used to the cutoff point is the creation of the Receiver Operator Characteristic (ROC) curve, which creates a plotted graph of the relationship between the true positive rate (Sensitivity) and the false positive rate (1-Specificity).

The instruments should be highly sensitivity and high specificity. The latter will result in a low false positive rate. As a result, the ROC curve is aligned with the upper left corner. In addition, the development of the ROC curve helps to compare the effectiveness of diagnostic tests by comparing the under curve of each type of curved surface represents more than the higher performance.

Read more in this like can help you more understanding about this topic
https://www.medcalc.org/manual/roc-curves.php

The mission completed for this week!! it was really fantastic assignment which allow me searching via several resources.....I would say I am just a small spot in the ocean of knowledge!!


Best wish
Prapasri Poopayang