Research Methods Flashcards
quantitative psychological types of research questions
- Difference: is one group of people DIFFERENT to another in some way?
- Association: is one construct RELATED to another
- Prediction: does one construct INFLUENCE others?
goal of psychological research
- Aim: make INFERENCES about a POPULATION.
- EG: a study investigating wellbeing in uni students.
The corresponding population = (ALL uni students)
population
everyone of interest to a research question
sample
a group of people taken from the population to participate in a study
why we take samples from a population
- not possible to recruit all people in a population, – SAMPLE is used
- samples used as INFERENCE about the population based on what happens with MEASUREMENT of sample
EG: Are psyc students smarter than the general population
- Test IQ for psychology students and general population
- recruit sample for psych students and compare to the ‘population’
- compare mean IQ of psychology students to general population’s IQ (100)
- based on results of this comparison, we will make an INFERENCE about the population of psychology students and whether or not its diff from general population in intelligence
EG: Do firefighters diff from the general population in their experience of anxiety?
- Need to know the typical level of anxiety for
- Firefighters
- People who are not firefighters
- Sample for firefighters population VS sample for non firefighters population
- compare MEAN levels of anxiety for both samples.
- based on comparison, INFER the result of sample mean comparison is likely to be same for the populations
construct
- INTANGIBLE, abstract attributes we THEORISE underlies OBSERVABLE BEHAVIOUR.
NOT DIRECTLY OBSERVABLE / measured - eg happiness, anxiety, intelligence
operational definition (numerical no.)
PROCESS of defining and measuring an UNOBSERVABLE construct indirectly (eg: IQ test for intelligence, questionnaire score for anxiety)
EG: intelligence as a construct
- physically intangible
- operational definition of contruct of intelligence = intelligence score
- IQ score allows intelligence to be measured INDIRECTLY
Where do research come from?
Literature search & Review (reading past researches and thinking what’s the NEXT STEP)
A question of: (TIPO)
- Theories
- Interest
- practical problems
- Observation
Research questions
broad ideas that typically ask about either
ASSOCIATION,
DIFFERENCE,
or CAUSATION.
Hypotheses
LOGICAL, SPECIFIC, TESTABLE, REFUTABLE and PREDICTIVE statements about what will happen in a psychological research study.
EG: state anxiety is negatively associated w mentalisation capacity
(hypotheses should NEVER predict that nothing will be observed) –framed qustions, not statements
Research Question & Hypothesis
Research Question –> Hypothesis
(informs)
Experimental Hypothesis (H1)
(alternative hypothesis)
a statement that PREDICTS an EFFECT (eg: difference / association)
Null hypothesis (H0)
predicts NO EFFECT (eg: no difference /no association)
Null hypothesis significance testing
–> process by which we can determine if our sample data provides for some sort of diff in terms of whatever being measured
- Propose null hypothesis that a population parameter (mean) has a particular VALUE
- ASSUME null hypothesis is TRUE
- Determine PROBABILITY of sample MEAN occurring IF the null hypothesis is true (eg is sample mean typical or extreme?)
- Involves statistical test based on a distribution of sample means, normal shape, can calculate the standard error - Probability of the sample mean LOW (TYPICAL) = REJECT the null hypothesis.
Probability is HIGH (EXTREME) = do not reject the null hypothesis.
The 2s Rule of Thumb
- In normal distribution, 95% of scores fall within approx 2 sd (s) of the mean
- those scores outside 2 sd (s) of the mean = extreme scores
- They are not expected as they occur infrequently in this distribution
Applying 2s Rule of Thumb - M=46.87
- S=4.84
- S=4.84 x 2 = 9.68
Lower limit = m - 2s = 37.19
Upper limit = m + 2s = 56.55 - More extreme than lower limit: 2.02%
- More extreme than upper limit: 4.04%
- Within 2s of distribution mean: 93.94%
Distribution of data
described according to :
central tendency (m) & variability (s)
Normal distribution
a bell-shaped curve, describing the spread of a characteristic throughout a population
- Most of the people are in the MIDDLE - peak of graph
- Reduce in frequency towards tails of graph
- SYMMETRICAL distribution
Typical scores
score that occur frequently
Extreme scores
- unusual to find extreme scores - ones that are VERY LOW or VERY HIGH
- They occur INFREQUENTLY in this distribution
- They indicate a DIFFERENCE to typical scores in terms of whatever is being measured
Distribution of sample means
made up of sample means from ALL of the RANDOM SAMPLES of a certain size (n) that could possibly be obtained from a population
- distribution of sample means = theoretical distribution governed by a mathematical theorem –> CENTRAL LIMIT THEOREM
Central limit theorem
- distribution of sample means = population mean
- provides us precise characteristics of the distribution of any distribution of sample means
- as sample size increase, the the distribution of sample means of size n, (randomly selected), APPROACHES a NORMAL distribution. (standard error –> 0)
- For large sample sizes (30 or more), the distribution of sample means will have a normal shape
Standard error
Standard deviation of the distribution of sample means
- s.d: s/square root of n
- As sample size INCREASES, standard error DECREASES —> 0
* Estimation of population mean becomes MORE PRECISE
- Larger the sample, the MORE RELIABLE estimate of population mean
How can we decide if individual’s scores are typical or extreme?
- by comparing one individual score with a distribution of other individuals’ score
- using sample mean and distribution sample means
alpha level (level of significance)
- If sample mean has less than 5% (Alpha level) probability unlikely that H0 is true, if it is more than 5% probability likely that H0 is true
- defines which sample means in a distribution of sample means are typical, and which are extreme, if the null hypothesis is true
- When comparison distribution is perfectly normal, the critical limits set by the 5% Alpha Level are precisely +/-1.96 standard errors from the mean of the distribution
- If our sample mean is inside these limits, the probability is greater than 5% and therefore high. DO not reject the null hypothesis
- If our sample mean is outside these limits, the probability is lower than 5% and therefore low. Reject the null hypothesis
Single Sample z-test
- involves calculating z score for sample mean
- expresses how many standard errors our sample mean is AWAY from H0
- a z-score of z=1.5 indicate:
our sample mean is 1.5 standard errors ABOVE the mean of the distribution of sample means - if z score of z=(-)1.5, indicate
our sample mean is 1.5 standard errors BELOW the mean of the distribution of sample means
Formula: z = (M-μ)/σ
z = z-score for sample mean
M = our sample mean
μ - population mean
σ - standard error of the mean
Note that the population standard deviation is known - if z-score is < 1.96, do not reject H0
- if z-score is > 1.96, reject H0
EG: Are psychology students smarter
σ = 2.19, more extreme than alpha level of 5% (1.96),
hence, probability = LOW, so REJECT
Single sample t-test
- Used when we dk the population standard deviation
- Almost all aspects are SAME as z-test
- We still use NHST to assign a value that indicates no effect, we still apply an alpha level of 5%, determine the probability of our mean occurring, result determines whether the null hypothesis is rejected or not
- BUT.. alpha level of 5% is NOT fixed
- Degrees of freedom (df) - one less than our sample size for a single sample z-test (n-1)
- The critical limit varies along with df
- If t-score NOT extreme for the degrees of freedom, = do not reject the null hypothesis
- If the t-score is extreme for the degrees of freedom, reject the null hypothesis
independent group design
appropriate research design for use with an INDEPENDENT sample t-test.
- assesses if difference between the two sample means is diff to zero or one sample is < or >than the other
control group
Participants in the control group DO NOT receive the treatment / intervention that is being assessed. The control group provides a benchmark to compare the results of the treatment group to.
JASP
output gives descriptive, t-score, p value, effect size and more
repeated measures research design
(is there a change across time?)
T1 - pre intervention
Treatment
T2 - post intervention
What would be the alternative hypothesis for a repeated measures research design using a repeated measures t-test?
The samples will come from DIFFERENT POPULATIONS before and after an event.
Repeated measures t-test
(paired samples, related samples)
- To assess whether there is a significant difference in participants scores before and after an event.
Correlational research design
– > investigate RELATIONSHIP BETWEEN TWO VARIABLES (eg: age and IQ)
Correlation checklist:
- examine rs between x and y
- each participant provide 2 pieces of data (age and IQ)
- no control
- examine linear / symmetrical
- positive linear ass.; as x increase, y increase
- negative linear ass.; as x decrease, y decrease.
- correlation CANNOT determination about cause and effect; no IV, DV
- Could be a 3rd variable - those who exercise more may engage in another activity that improves their happiness
Each participant must provide 2 pieces of data (for each variable eg provide time spent exercising and happiness questionnaire score)
Gather evidence about association
Test of correlation: Pearson correlation coefficient (r)
–> Pearson’s r is. a measure of correlation in a SAMPLE
Pearson’s R measures the strength of linear rs between x and y
- VALUE of r lies between -1 and 1
(closer to 1/-1 means larger effect/stronger, closer to 0 means smaller effect/weaker) - SIZE of r specifies how close the data is to the straight line (ie. how strong is the linear association)
- SIGN of r specifies direction of association
Pearson’s (r) in JASP
Pearson’s r procedure in JASP tells us:
- STRENGTH of correlation
- DIRECTION of correlation
- determines if we can INFER from sample to population correlations using standard null hypothesis significant testing (NHST)
if there association observed in a sample (r) is ALSO present in the population
* if r is large enough, so that it’s EXTREME in a distribution of sample correlation coefficient, then we can infer that there IS an association between two variables in a population.