RM Lecture 1 - Intro to Research Design and Data

control condition

helps rule out alternative explanations for your results

between subject design

separate participants in group A vs group B, who face different conditions

between subject designs pros and cons

Each participant only contributes one data point, avoids participant/experimenter effects. But takes longer and is less powerful than a within-subject design and introduces variation due to individual differences

within subjects design

same participants participate in both conditions

within subjects design pros and cons

each participant contributes multiple data points, accounts for individual differences, cost and time effective. But, can introduce order effects/fatigue effects

matched pairs design

participants matched across conditions to try to account for individual differences (ie gender, experience). often difficult to match people accurately but would be good choice for twin studies

extraneous variables

not controlled in the experiment, could have effect on DV

confounding variable

extraneous variables that vary systematically with the IV to influence the DV, eg one condition much longer and more boring than the other causing fatigue effect

categorical data

labels, nominal and ordinal, sometimes called discrete data

nominal data

category labels with no hierarchical order (eg gender, nationality)

ordinal data

nominal data that has an inherent order (eg stage in illness)

numerical/continuous data

data on a scale, interval and ratio

interval data

scalar/continuous data that has no meaningful zero, intervals are equidistant (eg temp in F)

ratio data

scalar continuous data with an absolute zero, intervals are equidistant (eg time)

descriptive statistics

numbers that summarize your data, eg. mean, median, presented in text/tables/graphs, shows the pattern of your data

z-score

represents a datapoint’s relationship to the mean of a group of values, useful to compare scores between participants or across conditions

nominal data can be presented as…

frequencies

continuous data should be presented as…

means and standard deviation

normal distribution

bell-shaped curve, bell curve has majority of the data around the average/central point, most tests will req that data is normally distributed

confidence interval

used as a measure of the spread of data, instead of range or SD, we are 95% confident that the population mean lies within this range (1.96 SD either side of mean)

Negatively skewed

mode is higher than mean and median, more high scores

positively skewed

mean is higher than median and mode, more low scores

mean not likely to be. anaccurate representation of data, therefore … would better represent the average

median

parametric tests

first choice of statistical test, usually works on the means