L.Question: how do you validate a new survey deigned to measure self-
Going to measure how good you feel about yourself, a way of
validating a survey in order to measure self-esteem.
Seeing scatter plots today, see correlations.
L.Q: What is the difference between a correlational study and an
Ans: a correlational study does not let you determine causation. We
don’t know if the changes are being caused by a different variable, an
experiment allows you to determine causation because you can control
for variables, isolate independent variable. There is only one thing that
changes when you measure behavior, other variables remain constant,
thus when you get change in variable than it was because of your
60 question survey
the main idea is to run 2 different self-esteem questionnaires, the
rosenburg is old. Sorensen: 50 ques, read statement and if u think
it is true, than u say t, or f. if healthy self esteem then most f, then
low score. The lower the better.
Rosenberg: 10 ques, strongly agree(3), agree(2), disagree (1),
stongly disagree. Gonna get score 0 – is strongly disagree. The
higher the better. Half Rosenberg positive question, half are
negative ques; avoid bias.
How to validate a questionnaire:
- good way vs. what we will do
1. Give the survey to everyone in the sample and get a score
2. Send everybody to the therapist and get a therapist to rate their
self esteem. 100$ a minimum. This way you get an expert to rate
people. They give the conductor a score out of 100
3. You calculate a correlation between the expert score and
experiment score. Your doing this to look for a strong positive correlation between the 2 (1 – (-1)), want score to be further
away from zero, positive would be 0.8 correlation.
Our Method: Cheap Method:
- Do each survey
- Calculate correlation between the scores from each survey
- Expecting a negative correlation: high score on rosenberg and
low in Sorenson.
- External validity of that correlation, comparing the two
- Internal validation: split half correlation. Is the whole survey
good, if good then any half of that survey should give you score
that is just as similar as the other one
- Split randomly, and choose whichever pile, each half has separate
score. Should get about the same score and should be highly
correlated. If 0.8 or more >>>> good
Background To Self Esteem:
- Maslow theory of higher of needs fits into self-esteem.
- LQ: How did chikmostihowy improve Maslow’s theory
Humanistic psychology: but they are very hard to be measured
such as what does it mean to be self-actualized.