Key Words - FNR201

Mean mode and median

• measures of central tendency

• measuring the averave

• most commonly occuring score

• and the middle score where half of the scores are above and half of the scores are below

Range

• the difference between the highest and lowest score in a set of scores

Standard Deviation

• the amount on average that scores vary from the mean

• take the sum of scores minus the mean divided by degrees of freedom

Variance

• the total variation represents the total variation in a distribution

• the standard deviation squared

z- score

• standardized score representing the distance above or below the mean in standard deviation units of a raw value in a distribution

Regression towards the mean

• the tendency for results to move closer to the mean the second time they are measured

Null hypothesis

• this is a statement of no difference

• states that there is no difference between the results of the two treatments or groups

Test of significance

• looks to the probability that observed differences or relationships in a result of sampling fluctuations

  • determines if the difference is reflective of a true diference or random error

Statistical significance

• degree of risk that you are willing to take that you will reject a null hypothesis when it is actually true

• probability of concluding that a stat sig difference exists when in fact there is really no difference

  • risk you are willing to take of a type 1 error

• 0.05 is the risk that you are willing to take

One tailed test

• specifies the direction of a relationship in advance

  • if you predict the direction of a relationship in advance

  • you do a one tailed test

Two Tailed test

• Test of any relationship between variables regardless of the direction of the relationship

  • if you do not predict the direction you are doing a two tailed test

Chi-Squared test

• Used with cross tabular analysis

  • uses a nominal DV

  • IV is categorical (nominal or ordinal)

T-Test

• Small sample sizes

• dependent variable measured at a ratio level

Between Subjects t-tests

• Used in an experimental design

• with an experimental and a control group

• where groups have been independently established

Within Subjects t-test

• same person is subjected to different treatments and a comparison is made between two treatments

Analysis of Variance

• family of statistical tests that compares group means to assess whether differences across means are reliable

• compares the means across sevaral categories

  • within and between groups

• looks at 2+ categories and the means from these groups

• requires a post hoc comparison to see which levels actually have differences across

F distribution

• compares estimates of variability

• compares between and within group variability

• ratio of mean squares

Probability

• how likely something is to occur depending on circumstance

Sampling error

• the lack of fit between the sample population

• difference between characteristics of the population from which the sample was selected

Confidence level

• represents the level of confidence that you have that you will reject a true null hypothesis

Dependent variable

• the effect of a cause and effect relationship

Independent variable

• the cause in a cause and effect relationship

• the part that causes the dependent variable to change

Research bias

• we tend to seek corroboration of preconceptions to help make sense of the complicated world to reaffirm our pet theories

• bias can lead research in a direction to find results that the researcher wants to be true

Experimenter effect

• tendency to produce findings that are consistend with the experimenters expectations

Information bias

• a flaw in the measurement of the exposure or outcome that results in different accuracy between groups

Expectancy error

• anticipation of particular research results possibly leading to a distortion of the results in the direction of the characteristics

demand characteristics

• a distortion in data collection when respondents give responses that they think the the researcher wants to hear

Random error

• inconsistencies that enter into the coding process but have no pattern

  • this is noise not a bias

Systematic error

• errors that distort the data in one particular direction

  • potential source of bias

  • this is the reason why you need to tare the scales etc

data massaging

• practice of playing with a data until the analysis produces the strongest possible association identified

  • works towards supporting prererred outcomes

    • potential source of bias

    • there must be data analysis plan

      • must be stuck to

Selected evidence

• only looking at evidence that supports your hypothesis and ignoring all other reports that do not agree with what you hope to find

Criterion validity

• how well a test is related to criterion

• this is also known as predictive validity

  • how well the test can actually predit the criterion

construct validity

• measure of how well measurement tool items are tapping into the underlying theory, construct or model of behaviour

• tested over time to see if conditions to see if it holds true in all instances

• establishing that the test can discriminate the concept of what it is intended to measure

Internal validity

• how well the test measures what it is intended to measure

• how well the experiment is designed to demonstrate causal relationship

External validity

• how well results from research can be applied to people and situations beyond what is being studied

measurement error

• the extend that values fail to represent the true underlying values of variables

coefficient of repeatability

• how well results can be repeated

• take 2x the standard deviation ??????

Bland-Altman Plot

• visual inspection of results

• compares the means with the difference between scores

• you want scores to be evenly clustered around the line of no difference indicating that the scores have low variability and no noticeable skews

• want it to be aroudn 0

nominal

• groups with no order and no difference between the groups

• the numbers are used just as labels for the variables

ordinal measurement

• these are groups that are paced in rank order with no value assigned between the ranks

Ration

• numerical organization with equidistance between numbers and also has a true 0 value

Population

• the entire group that you wish to study

• you take a sample from the population of interest

Sampling frame

• the group of the population that you select the sample from

• ideally this would be the entire population but that is often not possible to aquire a group of the entire population

Probability sampling

• random sampling

• taking a random group so that everyone in the population has an equal chance of being selected

• allows findings to be generalizable to the population

Simple random sampling

• assigning everyone in the sampling frame a number and using a random generator to select random individuals to be in the study

Systematic sample

• obtain the sampling frame then assign random numbers to everyone

• choose a skip interval

  • total units in the population / total units needed in the sample

• Round K to the nearest lower round number and use that skip interval starting from a random first case

Stratified sample

• divide the population (sample frame) into strata groups or categories

• within each stratum a simple or systematic sample is selected

  • you want to do this when there are important independent variables that you want to study

  • ensures that the strata in the population are adequately represented in the sample

Multi-stage area sampling

• when you want to look at a large geographical area but there is no population list and the pop is spread out

• draw several samples in stages

  • random area selection

  • continue to pick smaller random areas

  • eventually you will have randomly selected a list of households

  • obtain a list of individuals then randomly select individuals in the households

Non-probability sampling

• does not provide an equal or known chance of selection to be in the group

• sampling methods cannot ensure equal chances for anyone to be selected into the group

Convenience sampling

• selecting a sample based on convenience or ease of selection

• using a captive audience

• looking for who is willing or near to answer researchers questions

Snowball sample

• referral sampling

• selecting people based on desired characteristics and having them refer to others with the same characteristics

• often used when you cannot obtain a list of the population subset who shares some characteristic

  • if it is hard to locate the groups

Quota sample

• respondents are selected on the basis of meeting criteria

• subgroups (convenience samples) are identified and a specified number of individuals from each group are included

Purposive sampling

• sampling is done strategically

• often subjective with researcher making decisions about who to select

• used to select a group of characteristics within a population or to select a group with particular characteristics

Descriptive statistics

• procedures for describing individual variables and relationships between variables

  • eg describing characteristics of a study sample

Inferential statistics

• procedures used to analyze data after an experiment is completed

• procedures used to determine if the IV has a significant effect on the DV or not

• allows for extrapolations from a sample to the population from which it was drawn

Crosstabs procedure

• Used when there is a nominal devepdent variable

  • yes or no for example

• data is cross-classified

  • sorted into categories within the IV and DV to show the relationship between the IV and a DV

• Uses a Chi-square test to determine if the results are significant or not

Comparing means

• Used with a nominal IV and a ratio DV

• compares the mean values of the DV for each category of the IV

  • t-tests and anova can be used as tests of significance to compare the means

Reliability procedure

• when a measurement tool measures the same thing more than once and the results are the same outcomes

  • test-retest

  • inter-rater

  • parallel forms

• measured in SPSS using Chronbachs alpha

  • measures the amount that variables covary, among items that make up a measurement

  • reliable instruments should have high levels of covarience

Increasing reliability

• eliminate things in a test which can be unclear

• standardize the conditions under which the test is taken

• minimize the effects of external events so that true test performance is not affected

• maintaining consistent scoring procedures

• standardize instructions to respondants so that they all take the test under the same conditions

Test -retest reliability

• measures how stable a test is over time

• administration of the same test at 2 different times to the same group of particiants

• correlate scores at time 1 with scores at time 2

  • keep the conditions the same for both of the tests

Parallel forms reliability

• measure of how equivalent 2 different forms of a test are

  • administer the 2 forms of the same test to the group of participants

  • correlate the 2 sets of scores

• eg. two different sets of words to see if the people can recite them the same

Inter-rater reliability

• measure of consistency from rater to rater

• have more than one rater rate the same thing and correlate the scores between them

  • improved by increasing training for those who are administering and ratings tests

  • want 80% + agreement

Correlation procedure

• used to determine the relationship between two ratio level variables

  • describes how closely 2 variables co-vary together

  • R-value is what the measurement is (-1 → +1)

• this is only able to look at the direction and strength of a correlation of variables and cannot determine any causality

• y = a + bx

Frequencies procedure

• used to create frequency tables for categorical variables in a data set

Regression procedure SPSS

• used to determine the impact of IV on the DV

• ratio DV and preferably ratio IV

• looks at the linear relationship between 2 variables

  • multiple has more than 2 variables

  • develops a mathmatical equation describing the linear relationship between IV and DVs

Temporal precedence rule

• in order for a causal claim to be made the independent variable must precede the depedent variable

Covarience rule

• causal variable must covary with the variable it is assumed to cause

  • when IV changes the DV must also change

R² - coefficient of determination

• the percent variation y which is explained by all the x variables together

  • % of variation that is explained by a linear model such as a regression line

R-value

• shows the strength of a relationship between two variables

• determines how an increase of one unit in a variable is associated with a proportional change in the other variable