-based on or depends on the sample size -formula is different for each statistical test -important for determining statistical significance of tests

N-2

n1-1 + n2-1 OR total # of subjects - 2

n-1

between groups (# of groups - 1) divided by within groups (total sample - # of groups)

of rows - 1

(# of rows - 1) x (# of colomns - 1)

N-2

0.5-1.0

0.4

0.3

0.2

0-0.1

34.13

13.59

2.15

0.13

type 1 error

correct

correct

type 2 error

reject the null hypothesis because it is statistically significant (this is the goal of researchers)

accept the null hypothesis because it is not statistically significant

non-directional hypothesis

directional hypothesis

different subjects are compared

same subjects compared more than once

-control: introduction of one or more constants -randomization/random assignment: equal chance of people being assigned to either control or experimental group -manipulation

advantages: -tests cause and effect -the highest level of evidence disadvantages: -participants drop out -testing effects

1. true experimental (pre-post test control group): measuring effects before and after treatment of control and experimental group

2. after only (post-test only): measuring only after treatment of control and experimental group

random assignment is not used and control is not fully possible due to the nature of the IV or participants. still tests cause and effect but the confidence in making causal relations are weak

measuring before and after treatment of both experimental and control group

measuring only after treatment of both experimental and control group

measuring before and after treatment of experimental group only

measuring data of intervention group only before treatment and after treatment at 1 week, 2 weeks, etc

advantages: more feasible disadvantages: difficult to make clear cause and effect statements, threats to internal validity

-explores people, places, events as they naturally occur so NO INTERVENTION/TX FROM RESEARCHER -not looking at cause and effect but rather RELATIONSHIPS/DIFFERENCES -IV occurs naturally and the researcher CANNOT MANIPULATE -no randomization

-nonexperimental (relationships/differences studies) -examines the relationship between 2 or more variables NOT cause and effect -tests if variables covary (as one variable changes, does a related change occur in the other variable?) -interested in the strength of the relationship between variables

-nonexperimental (developmental studies) -examine data at ONE TIME/OCCASION with SAME PARTICIPANTS

-nonexperimental (developmental studies) -collects data from the SAME group at DIFFERENT TIMES -by collecting data at yearly intervals, the researcher obtains a longitudinal perspective

-nonexperimental (developmental studies) -from after the fact -DV already affected by IV so the investigator attempts to link current events to past events -researcher hypothesizes that variable X (smoking) is linked to variable Y (lung cancer)

-unable to explain cause and effect -useful in making predictions -important when randomization, control or manipulation is not possible

-nonexperimental -classified as descriptive, exploratory, comparative -useful for gaining lots of information from a large population -data tends to be superficial

-methodological: evaluation of data-collection instruments -systematic review/meta-analysis: summation of research studies based on a focused question and a meta-analysis is a systematic summary using statistical techniques to assess -secondary analysis: form of research where previously collected and analyzed data is used for a different purpose -epidemiological studies: factors affecting the health and illness of a population are examined in relation to the environment -clinical practice guidelines: link research and practice and serve as a guide for HCPs

-homogenous sampling -consistency in data collection -manipulation of the IV -randomization

-makes it unknown if it is the IV that caused the DV -history -maturation -testing effects -instrumentation -mortality -selection bias

-concerns the generalizability of findings to additional populations -internal validity must first be established -selection effects: researcher cannot obtain ideal sample) -reactive effects: participants responses to being studied -measurement effects

-no randomization in selecting this sample -convenience: most readily accessible persons (first 50 patients admitted into a hospital) -quota: subjects who meet the inclusion criteria are recruited and consecutively enrolled until the total sample size is reached

-representative group -simple random -stratified random -multistage (cluster) -systematic

-researcher establishes sampling frame (list from which a sample is taken) -people in a course listed and random participants are chosen from that list

-population is divided into strata -participants randomly selected from each strata

-random sampling of units that meet eligibility criteria -from large to small -first stage: all hospitals in Canada -second stage: all ICUs in each of those selected hospitals -third stage: sampling nurses full time in those units

-involves the selection of an interval at which point individual participants are selected

used to see how large a sample should be to show statistical significance

1. rigour: we want high quality measurement instruments that reflect the concepts being studied (trustworthiness)

2. psychometric assessments: detects evidence on quality of measurement tools which include: -reliability (consistency) -validity (accuracy)

-test to which an instrument yields the same results repeatedly -stability, internal consistency, equivalence

-test-retest reliability: does the instrument produce the same results repeatedly? -however participants may recognize repeated items

-homogeneity: do items measure the same concept? -need to have multiple items to establish internal consistency -Chronbach's alpha: closer to 1 is high internal consistency but greater than 0.7 is acceptable (ranges from 0-1)

-interrater reliability: will the tool produce the same results when equivalent procedures are used? refers to consistency of observations between 2 or more observers with the same measurement tool -tend to do with observational methods intrarater reliability: 1 observer observes the same behavior several times

-does instrument measure what it intends to measure? (accuracy) -concept validity, criterion-related validity, construct validity

-determines if the measurement tool and the items represent the content desired -content validity index: experts rate items 1-4 if they are relevant

-measures the relation between scores from the instrument with external criteria -predictive validity: degree of correlation between the measure of concept and a future measure of the same concept (do high school grades predict post-secondary grades?) concurrent validity: degree of correlation of 2 measures of the same construct administered at the same time (pulse oximeter vs. ABGs-gold standard)

-extent to which a test measures a theoretical construct or trait -complex process

-nominal: dichotomous or categorical, mutually exclusive (pick one only), numbers assigned have no meaning, mode and bar graph used as measure of central tendency -ordinal: categorical, mutually exclusive, numbers assigned have meaning, distance between categories is unknown, mode, median, and bar graphs can be used -interval: variables are continuous and rank ordered, equal intervals between numbers, 0 does not mean absence, can report mean, median, mode and histograms ratio: variables continuous and rank ordered, have a meaningful 0 and means the absence, can also report mean, median, mode, and histograms

mean, median, mode

-range: h-l -standard deviation: how much each score deviates from the mean -variance: spread of data (SD squared)

-the mean is greater than the median and mode -the peak is to the left while it tails off to the right

-the mean is lower than the median and mode -the peak is to the right while it tails off to the left

-the correlation between X and Y (correlation coefficient) -ranges from -1.00 to +1.00 -how the value of one variable changes in relation to changes in another variable

both variables go in the same direction

as one variable goes in one direction, the other variable goes in a different direction

r2 or the correlation coefficient squared

-DV is interval or ratio level -variances are homogenous -normal distribution -sample data is usually more than 30 -t-test, ANOVA, Pearson's r

-compare DIFFERENCES between group means -t-test for independent groups -t-test for dependent groups (paired t-test)

-test DIFFERENCES not relationships -compares means between TWO GROUPS ONLY -participants tested only once and are different people in each group (experimental vs. control group) -IV is at the nominal level -DV is at the interval or ratio level

-compare mean scores -for one group of people who are tested TWICE (within group design) -tested before and after an intervention -used in pre-test-post-test studies

-helps decide if the difference in group means is related to the IV or due to other reasons -breaks total variability in a dependent variable into 2 parts: variability between the groups and variability within groups

if the variability BETWEEN (numerator) groups is significantly greater than the variability WITHIN (denominator) groups, then the probability is high that the IV led to differences in group means -- reject the null

-d/t general differences between the groups -measures how much difference exists between groups`

-variability within each group

-F ratio -Post hoc test will be done if F ratio results are statistically significant

1. one-way/simple ANOVA

2. two-way/multifactor ANOVA

3. repeated measures ANOVA (RMANOVA)

4. multivariate statistical analysis of covariance (ANCOVA)

-tests differences between MORE THAN 2 GROUPs of treatment (IV) -ex: DV is tumour size and the IV is the treatment which consists of 3 groups (chemo, radiation, both) -participants tested only once -non-directional F test -participants in each group are different people

-same as one-way ANOVA but has more than 1 IV -ex: testing the impact of exercise and diet on weight loss with students

-participants tested MORE THAN ONCE or at different points in time (before/during/after intervention) -can be one-way or two-way ANOVA

-used to statistically control the effect of extraneous variables on DV -tests the significance of differences between group means after adjusting scores on the DV to eliminate the effect of covariates -used when control through randomization is lacking

-a non-parametric test -involves a comparison of what is observed and what is expected by chance -one-sample chi-square test (goodness of fit) and two-sample chi-square test (test for independence)

-tests the difference in the proportion of only 1 variable -ex: what is the proportion of RNs who smoke, do not smoke, and never smoked? (1 variable of smoking status)

-tests the difference in the proportion between 2 variables -ex: is there an association between exercise patterns and smoking status in RNs? (first variable is exercise and second variable is smoking)

-testing RELATIONSHIPS between 2 variables -tests the degree to which variables are related -by squaring correlation coefficient, you can determine the amount of variance