Stats of Psychology Midterm

Nominal

Simplest & lowest level of measurement

• no specific order (cannot be ranked)

Common Examples

• Gender: Male, Female, Non-binary

• Blood Type: A, B, AB, O

• Place of Residence: City, state, or country code

• Eye Color: Blue, brown, green

Ordinal

Ranking

Show an order but intervals ain’t equal

Second most limited

Examples : Position in race/ Likert Scale/ Socioeconomic status

Interval

Differences between scale values are equal

Not real 0 (arbitrary)

Ex - Standard Test Scores/ Temperature/ Time/ Calendar

Ratio

Real 0

• Equal intervals

• Clear logical ranking

Ex. - Age/ Height/ Weight/ Income/ Distance/ Time intervals

Descriptive Statistics

Used to organize and describe the characteristics of a collection of data

Ex. Collecting students major, and age and finding what’s the most popular major

Two types - Central tendency and Variability

Central tendency

Single values that attempt to describe a data set by identifying entire group of scores

Mean

Most common type of average

sum of all values in a group divided by the number of values in that group

Median

Midpoint in the set of scores

Mode

value that occurs the most frequently

Variability

reflects how scores differ from one another

Standard deviation

represents the average amount of variability in a set of scores

abbreviated as s or SD

Null hypothesis

• No difference in relationship between variables

“There is no difference in the salary of factory workers based on gender.”

Non directional hypothesis

Hypothesis that shows difference but does not specify negative/positive direction

“There is a difference in exam scores between students who study for different amounts of time.”

Directional hypothesis

Hypothesis that describes exact relationship of variables, positive and negative

“Students who use active-learning methods will score higher on final exam averages than students who use traditional lecture methods."

Correlation Coefficient

Ranges between -1 and 1 and shows the relationship between variables

(-/+) sign dictates direction

negative -downward positive -upward

Correlation threshold

.2 ⬇ - very weak/no connection

.2 - .4 - weak

.4 - .6 - moderate

.6 - .8 -strong

.8 - 1 -very strong

Does correlation equal causation

No

Coefficient of determination

Percentage of variance of one variable accounted for by the variance of the other variable.

Ex -The variance of one variable can be explained by the other variable

•How to compute: Take the square of r

•Ex: the correlation between smoking and lung cancer is .8.

   r = .8, r2 = .64, 64%

•Smoking can explain 64% occurrence of lung cancer

Absolute value

strength of correlation.

the higher the stronger

Reliability

different testing measures that psychologists used

must be consistent, trustable, and have little variation

Observed score = True score + Error score

•Minimize the error score

•Measure consistent score

Test-retest

If a test is reliable/consistent over time

•How

•Run the test

•Rerun the test after some time

•Compute correlation between the two runs

• Results mean:

•0.9 and greater: Excellent reliability

•Between 0.9 and 0.8: Good reliability

•Between 0.8 and 0.7: Acceptable reliability

•Between 0.7 and 0.6: Questionable reliability

Problems/Cons

•Change in true score over time

•Test taking experience

•Carryover effect

Parallel form reliability

If two different forms of the same test are consistent

•How

•Run the two forms of the same test

•Compute correlation between the forms

•Pros:

•Eliminate carryover effect

•Eliminate test taking effect

•Eliminate the effect of change in true score

Cons

•Doubles the number of items needed

•Difficult to generate an equivalent test

Internal consistency reliability

If items of a test is consistent with one another

•How:

•Run the test

•Compute correlation between each item and others

Pros:

•It is used mostly on scales and questionnaires

•Eliminates the difficulties in parallel forms generating

•Easy to compute because SPSS can do that

•Cons:

•Difficult to use for observations

Inter-rater reliability

If two judges/rates agree on their judgments

•How:

•Let the rates make their judgments

•Compute the number of agreements over all possible agreement

•Or compute correlation

If not reliable, then what?

•Standardize instructions

•Increase the number of items

•Delete unclear items

•Eliminate the floor effect or ceiling effect

•Minimize the possible external effects

Validity

A property of assessments that shows that the tool measure/assess what it is intended to measure (What you’re testing for is what is being tested.)

Content Validity

•If the test items represent the universe of items

•A.k.a. face validity

•How:

•Generate the items of the test

•Ask an expert if it represents

the universe

Criterion validity

Compares the test with other measures which are already measured as valid.

•It is possible to set some criteria to measure concepts

•Types of criterion validity

•Concurrent - Measure the criterion at the same time

•Predictive -Measure the criterion later

Construct Validity

•If the operationalized variable measure what it’s supposed to.

• To make concepts/theories measurable

•How:

•Run the test

•Make some theoretical predictions

•Compute correlation between the test score and your predictions

Construct Validity - Convergent

If it correlates with the measures that are theoretically predicted to correlate

Construct Validity - Discriminant

If it is found unrelated with the measure that are theoretically predicted to be unrelated

If not valid, then what?

Need to rebuild your test

•Check your criterions

•Check your theoretical rationale

Relationship between reliability and validity

Can be reliability but not valid

Sample and Population

you test a sample to reflect on the population’s broader attitude.

Inferential Statistics

Inferring population parameter from sample statistics

•Test bilingual and monolingual sample on creativity

•Conduct some analysis

•Conclude whether it is real difference or chance

•Infer the findings to the population

Type I (and p value)

Rejecting null hypothesis, believing there is a relationship between variables when there is no(null) difference

p value we reported shows level of significance

•In psychology we usually use .05 and .01.

represented by the Greek letter alpha, or a

Type II error

Accepting a null hypothesis even when there is a difference (false null)

represented by the Greek letter groups. This is also called power, beta, or B

Null vs Research Hypothesis

•Null hypothesis states equality, research hypothesis states inequality

•Null hypothesis refers to the population, research hypothesis refers to a sample

•We directly test research hypothesis and indirectlytest null hypothesis

•We use Greek symbols for null hypothesis and Roman symbols for research hypothesis

Normal Distribution

visual representation of a distribution of scores that has three characteristics.

Mean, Mode and Median are the same

•Symmetrical

•Asymptotic - closer and closer to the horizontal axis but never touch