PS1103 - Psychological Research Skills 1

why do psychologists use statistics

• psychology is an experimental science that relies on empirical data - information gained through observation of experimentation

• researchers use statistics to systemically collect, summarise and interpret this data to answer specific research questions

descriptive statistics

• methods used to summarise display and provide a broad overview of data such as identifying typical values or patterns

Inferential statistics

• used to interpret that data in a standardised way to see of it supports a hypothesis

categorical data

• refers to discrete groups or categories such as sex, species or birth order

continuous data

• consists of numbers that can take any value on a measurement scale such as height, weight or reaction time

nominal data

• consists of named discrete categories or bins

• there is no meaningful order to the categories and each observation must only fall into one group

• includes eye colour, ethnicity or car type

ordinal data

• varies along a continuums where the rank order matters but the mathematical difference between the numbers is not meaningful

• finishing positions

• in a race or an academic career path

interval data

• equally spaced intervals where the difference between the numbers is meaningful

• however it has no true zero point meaning you cannot meaningfully calculate ratios

• e.g temperature in celsius and IQ scores

ratio data

• highest level of measurement - has equally spaced intervals and a true zero point

• because of the true zero ratios between numbers are equal (e.g 10kg is twice as heavy as 5kg)

• e.g mass, length, time and number of siblings

can you change data from one level of measurement to another

• yes but you can only move downwards in complexity e.g from ratio to ordinal

• e.g exam marks (ratio) can be converted to grades like A B or C (ordinal) but you cannot turn a grade back into an exact mark

steps to answering a research question

• start with an initial research question

• form a hypothesis (prediction to be tested)

• collect data to test the hypothesis

• analyse the data using statistics

• reach a conclusion on whether the data supports the hypothesis

why does the type of data you collect matter

• determines which statistical test you can perform

• higher levels of measurement provide more detailed information than lower levels

when should you use a bar chart instead of a histogram

• use a bar chart for nominal data, showing spaces between bars to indicate categories are not continuous

• use a histogram for ordinal interval, or ratio data where the order of scores are meaningful

Uni-modal distributions

• has one clear peak in the center

Bi-modal distribution

• has two well defined peaks which do not necessarily have to be the same height

Key characteristics of a normal distribution

• symmetrical, has one central peak and the mean, median and mode all share the same value

How can you tell the difference between positive and negative skew

• a positive skew has a long tail pointing to the right

• A negative skew has a long tail pointing to the left

Mode

• most frequent score

Median

• the mid point where 50% of scores are above and 50% are below

Mean

• mathematical average - calculated by summing all scores and dividing by the total number of scores

Which measure of central tendency is best for skewed data or nominal data

• the median is best for skewed data because it is unaffected by extreme scores

• Mode is the only measure that can be used for nominal data

Range

• maximum score minus the minimum score but it can be distorted by outliers

Inter-quartile range (IQR)

• calculates the range of the middle 50% of data which ignores extreme scores

What does standard deviation tell us about a data set

• the most common measure of dispersion and tells us how much scores cluster around the mean

• Large SD results in a broader flatter distribution while a small SD indicates scores are close the average

Z-score

• a standardised score that tells you how many standard deviations a value is from the mean

• Score that are more than 3 standard deviations away from the mean are typically considered outliers

Independent variable

• the variable that the researcher controls and manipulates

• The cause that the experimenter changes to see what happens

Dependent variable

• the variable that the researcher measures

• The effect that might change because of the IV

What makes a good research question

• finds a balance - must be general enough to be relevant but specific enough to be a feasible study

• Should also be realistic given the time and resources

Confounding variable

• the variable that is related to both the IV and DV but cannot be controlled

• Makes the link between the variables uncertain because you can’t be sure what caused the result

Extraneous variables

• extra variables that can be split into two types

• Participant variables - individual traits like a persons motivation or stress level

• situational variables - environmental factors like noise levels or time of the day

Non-experimental research

• research where the experimenter does not manipulate any conditions

• Includes observations, case studies and surveys

• Can describe or predict behaviour but cannot determine cause and effect

Two main features of a true experiment

• the researcher has complete control over the conditions

• Participants are randomly assigned to different groups or conditions

What is a within-subjects design

• a design where the same participants take part in every condition

• Each person acts as their own control

Between-subjects design

• A design where different groups of people are used for each condition

• Researchers often use random assignment or matching to make sure the groups are equal before the study starts

Quasi- experimental research

• research that looks like an experiment but lacks complete control

• Participants are usually selected based on existing traits like gender rather than being randomly assigned

Difference between an alternative and null hypothesis

• H1 - a testable prediction that there will be a relationship or difference between the variables

• H0- a statement that there is no difference or relationship between the variables

Directional hypothesis

• a hypothesis that specifies the exact direction of the result using words like ether, more or higher

When do you use parametric vs non-parametric tests

• p - used for interval or ratio data that follows are normal distribution

• N-p - used for categorical data - nominal or ordinal and makes no assumptions about the populations distribution

main purpose of the Chi-Square Goodness of Fit test

• used to see if the proportions in your sample match the proportions you would expect in a specific population

• essentially measures the difference between your actual data and the null hypothesis

what type of data is required for a chi-square test

• nominal or categorical data

• this means data that fits into categories - like gender or hair colour where you can count frequencies rather than taking an average

rules of chi-square test

• independence - each participant can only belong to one category

• expected frequencies - the expected count for each category should be greater than 5

• sample size - for results to be reliable the total sample size should generally be greater than 20

what is the difference between observed and expected frequencies

• observed - the actual number of people you measured in each category