3020 Measurements in Psychology

Standardising scores into a z score is a ___ transformation

linear

Transforming z scores into a percentile rank is a ___ transformation

non-linear

If someone scores between ±/- 1 SD of the mean on a test, what percentage and fraction of the population are they the same as?

68% or middle two-thirds

If someone scores between ±/- 2 SD of the mean on a test, what percentage of the population are they the same as?

95%

Distinguish interval and ratio scales (giving examples of each)

Interval = interval between levels is meaningful but there is no absolute zero e.g., temp. C, time of day. Ratio = ratio between things is meaningful because there IS an absolute zero e.g., distance in cm, timing in seconds

“An interval scale is the only scale where twice the score is actually meaningful” - true or false?

False, this is only try of ratio scales

Are most psychological scales ordinal or interval? Give an example

Most are ordinal over interval. An example is IQ where an IQ of 140 is NOT twice that of 70, but treated more as ordinal

How is cognitive impairment defined in terms of IQ?

An IQ 2 SDs below the mean so an IQ less than 70

“Normal distributions are convenient because we can do more powerful statistical tests and make scales comparable” true or false?

True, and this is why we may do a non-linear transformation to make scales normally distributed

Describe a non-linear transformation and when it is used compared to a linear transformation

Non-linear transformation = when data is not normally distributed so you stretch out bits of scale more than others to make it normal. Linear transformation = raw scores are converted into standard scores but distribution shape isn’t changed e.g., z scores

“Z-scores are an example of a non-linear transformation” - true or false? Why?

False, z-scores are an example of a linear transformation because they standardise raw scores but do NOT change the shape of the distribution

If someone gets a raw score of 30 which equates toa percentile rank of 60, what does that mean compared to the rest of the cohort?

It means that 60% of the cohort scored 30 or less

With a positive z score, you look at the ___ portion, with a negative you look at the ____ portion

larger, smaller

Converting a z score into a percentile rank is an example of what kind of transformation?

Non-linear as it does change the distribution

“Correlation coefficients show us whether a relationship is statistically significant” - true or false?

False, they only show the strength and direction

The smaller the degree of scatter of scores, the ___ the correlation efficient

greater

“The slope of a correlation demonstrates the magnitude of the relationship” - true or false?

False, the spread of scores demonstrates the magnitude

“Linear transformations can change the correlation between two variables” - true of false?

False

Are significant outliers a problem in correlations?

Yes, they can skew understanding of relationship between two variables

What type of test is used to measure significance of a correlation?

T-test comparing significance to zero

What values are deemed a Large, Medium, and Small correlation?

L = .50, M = .30, S = .10

Distinguish Pearon’s r from Spearman’s Rho

Pearson’s = parametric, variables must be normally distributed, on an interval scale, sensitive

Spearman’s = non-parametric, variables don’t need to be normally distributed and can be ordinal, interval, or ratio, less sensitive

Why is Spearman’s Rho less sensitive than Pearson’s r?

Because it uses rank order rather than raw scores 

What are the three fundamentals of measurement?

Standardisation = ensuring different measurements are in equivalent units before comparing them. Reliability = the extent to which a measurement tool gives consistent measurements each use. Validity = extent to which a measure actually measures what you’re intending