research methods advanced week 1= intro

intro

hypothetico-deductive method

thoery ,hypothesis ,operationalization of concepts, selection of participants, survey studies/ experimental designs, data collection, analysuis and findings

survey

measures naturally occurring variables as no manipulation

experimental

manipulate variables to isolates effects

systematic variation

variation that an be explained by the model (the statistic) (EFFECT) (h1)

unsystematic variation

variance not explained by the model (the statistic) (ERROR)(h0)

what is the test statistic

variance explained by the model divided by the variance not explainted by the model

our test statistic provides a ratio of good fit to bad fit- HIGHER STATISTIC BETTER THE FIT

hypothesis generation and testing

we pose hypothesis, analyses our data, we calculate probability of getting result if null hypothesis is true, we then reject or fail this hypothesis.

type 1 error

false positives- hasty rejection where you conclude there is an effect when there is not

type 2 error

false negatives- hasty rejection of the alternative hypothesis where you conclude there isn’t an effect when there is one

alpha and beta effects

alpha (0.05) probability of type 1 error

beta (0.20) probability of making type 2 error

effect sizes

dependent on test used, but used to conlude if there is an effect when there isnt. they can be significant but meaningless. it is the MAGNITUDE of the statistical effect found and compare studies

effect sizes table

power analysis

attempt to control type 2 errors by telling us the statisticl=al powers associated with test. this can be done with a priori (before study) and a post hoc (after data collection and inferential statistics)

the mean

the average of a set of numbers, gives indication of central tendencies of a data set

the variance

sum of all the squared differences divided by the number of participants minus 1. indication of the spread of data

standardized deviation

sum of all the squared differences divided by the number of scores munus 1, then square rooted. same as variance but square rooted down making it easier to interpret

parametric statistics

makes assumptions about the data being normally distributed, homogeneity of variance, used in ratio and interval data and used for group differences in equally sized groups

non-parametric statistics

makes no assumptions about the data , violation of normality assumption, used for ordinal data and small group sizes

research integrity- 4 principles

honesty, accountability, professional courtacy and fairness and good stewardship

replication crisis

many findings in past are false due to low statisitcal power and bias. around only 36% of studies could be replicated and replications of studies found smaller effect size4s.

why dont studies replicate

publication bias, faliure to control for bias, p-hacking. poor quality control, low statisitcal power, HARKING