Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
149 Cards in this Set
- Front
- Back
internal validity |
threat to one group, not the whole experiment |
|
alternative explanations of study outcomes are called |
threats to internal validity |
|
4 parts of good quantitative research design |
statistical conclusion validity, internal validity, external validity, construct validity |
|
statistical conclusion validity |
ability to detect true relationships statistically |
|
internal validity |
extent it can be inferred the independent variable caused dependent variable |
|
external validity |
generalizability of observed relationships across samples, settings, or time |
|
construct validity |
degree that key constructs are adequately captured |
|
main thing in controlling confounding variables |
achieving constancy of conditions |
|
external factors |
conditions, environment, time, etc |
|
intrinsic factors |
subject characteristics |
|
5 methods of controlling intrinsic factors |
randomization, subject as own control, homogeneity, matching, statistical control |
|
main 2 methods for controlling intrinsic |
matching, analysis of covariance |
|
statistical control ie |
analysis of covariance |
|
randomization (2) |
recruitment, assignment |
|
researcher bias (2) |
double blind, more than one observer |
|
statistical analysis (4) |
test null HO, probability of type 1 or 2 error, effect size, statistic assumptions have been met |
|
type 1 error (3) |
think intervention worked when it did not, rejection of true null HO, caused by extraneous variable |
|
alpha or p value cannot be greater than |
0.05 |
|
type 2 error (3) |
accept false null HO, failure to find significance when there is one, from insufficient power (sample size) |
|
which error is more serious |
type 1 |
|
when we increase control of type 1 we |
decrease control of type 2 |
|
7 threats to internal validity |
temporal ambiguity, selection, history, maturation, attrition (mortality), testing, instrumentation |
|
temporal ambiguity |
unclear which variable occurred first |
|
selection |
difference in experimental and control groups |
|
biggest threat to non experimental designs |
selection |
|
history |
events during data collection (Katrina, 9/11) |
|
maturation |
subjects change (smarter, sicker) |
|
way to help with maturation |
ANCOVA |
|
attrition (mortality) |
loss of subjects from study |
|
testing |
subjects get better at test (pre/post) |
|
instrumentation |
data collection instrument changes during data collection |
|
3 ways to control instrumentation |
inter-rater reliability, training to correct, calibration of instruments |
|
3 tx effects |
placebo, hawthorne, multiple tx |
|
hawthorne |
subject want to please researcher or behave differently because being studied |
|
hawthorne most concern when |
researcher in position of authority, leadership role (subject's physician) |
|
4 other concerns not controlled through random assignment |
diffusion of tx, resentful controls, equalization of tx, compensatory rivalry "john henry" |
|
diffusion of tx |
subjects communicating |
|
resentful controls |
control gets no tx (may drop out or score low), blinded tx (give fake tx) |
|
equalization of tx |
people admin tx to compensate the on tx group, double blind |
|
compensatory rivalry "john henry" |
control tries to compensate for lack of tx, blinding, separate groups |
|
3 parts to external validity |
generalizability, population validity, ecological validity |
|
population validity |
ability to generalize to other populations (children to adolescents) |
|
ecological validity |
setting, geographical location |
|
6 threats to external validity |
selection effects, time, history, novelty, experimenter effects, hawthorne effect |
|
selection effects |
sample should represent population |
|
time |
winter vs summer, morning vs night |
|
history |
results considered in context of time period, may not be generalizable to future |
|
novelty |
subject response just because it is new or unusual |
|
experimenter effects |
are there different effects if different researcher |
|
hawthorne effect |
internal or external |
|
increase external validity by |
replication replication replication |
|
3 ways to deal with threats |
eliminate threat, control threat, account for threat |
|
eliminate threat |
data collection by assistant if researcher is threat |
|
control threat |
make certain some subjects with threat are in each group |
|
account for threat |
"limitations" |
|
must have internal validity in order to have |
external validity |
|
impossible to develop "best practice" without |
organizing and evaluating research evidence through a systematic review |
|
3 forms of systematic reviews |
narrative qualitative integration, meta-analysis, metasynthesis |
|
meta-analysis |
statistical integration of results |
|
metasynthesis |
theoretical integration and interpretation of qualitative findings |
|
meta-analysis advantages |
objectivity, increased power, increased precision |
|
increased power |
reduces risk of type 2 error |
|
increased precision |
smaller confidence intervals |
|
objectivity |
eliminates bias in drawing conclusions |
|
meta-analysis criteria |
hypothesis identical across studies |
|
metasynthesis |
bringing together and breaking down findings |
|
3 general rules for quantitative analysis |
statistical tests are selected a priori, run all identified tests, report all tests run |
|
selective reporting is a source of |
bias |
|
type of analysis driven by (3) |
goals, assumptions of data, number of variables |
|
frequency distributions |
arrangement of numeric values from lowest to highest and how many times each value is obtained |
|
frequency distributions in 3 ways |
shape, central tendency, variability |
|
frequency distributions represented in (3) |
chart, table, graph |
|
3 characteristics of normal distribution |
symmetric, unimodal, not too peaked and not too flat |
|
normal distribution referred to as |
bell shaped curve |
|
mode |
most frequently occurring score |
|
median |
exact middle |
|
mean |
average |
|
only thing applicable to nominal value |
mode |
|
median used when |
want to compare |
|
what is less affected by extreme scores |
median |
|
what is affected by outliers or extremes |
mean |
|
homogeneity variability |
little |
|
hetergeneity variability |
great |
|
range |
highest value minus lowest value |
|
standard deviation (SD) |
average deviation of scores in a distribution |
|
correlation coefficients used for |
describing relationships between two variables |
|
the greater the absolute values of the coefficient... |
the stronger the relationship (r = -.45 > r = +.40) |
|
pearsons r is both |
descriptive and inferential statistic |
|
correlation tests that the relationship between two variables is |
not zero |
|
reliability by setting |
confidence limits |
|
probability by |
hypothesis testing |
|
statistical inference conclusions concerned with |
probability of drawing an erroneous conclusion |
|
confidence interval |
2 numbers defining an interval that we believe with identified level of confidence actually includes the estimated population paramete |
|
interval estimation |
range of values within a population value probably lies |
|
interval estimation includes |
confidence interval (CI) and confidence limits |
|
3 things significance testing can go between |
sample and population, 2 samples, 2 variables in sample |
|
hypothesis testing used to test differences between (3) |
means, proportions, variances |
|
level of significance |
preset standard that is considered significant in determining differences |
|
most common level of significance |
.05 (5% chance of making type 1 error) and .01 |
|
for level of significance set a |
priori (alpha) (differen than chrombox alpha) |
|
priori alpha |
p value |
|
smaller p value show more evidence |
against null HO, significant |
|
larger p value show more evidence |
for null HO, not significant |
|
2 ways to thing about which stat is correct |
level of measurement, number of variables |
|
level of measurement |
nominal, ordinal, interval ratio, determines parametric or nonparametric |
|
number of variables |
bi-variant, multivariant |
|
parametric stats type of sample |
random from define population |
|
parametric stats dependent variable measured at |
interval ratio level |
|
parametric stats dependent variable is |
normally distributed |
|
parametric stats is |
more powerful |
|
parametric stats can make |
assumptions about population from which sample was drawn |
|
nonparametric stats do not require |
normal distribution |
|
nonparametric stats measures data |
on nominal or ordinal level |
|
t test tests the difference between |
two means (means for men vs women) |
|
t test ie |
is there a difference between effectiveness of tylenol and advil for HA? mean pain rating for each group |
|
t test for dependent (paired) groups: within-subjects test |
means for pt before and after surgery |
|
error of multiple comparisons |
each test 5% chance results due to standard error, 3 groups = 15% chance |
|
what avoids error of multiple comparisons |
ANOVA |
|
analysis of variance (ANOVA) |
tests the difference between more than 2 means |
|
5 ways of ANOVA variance analysis |
one way ANOVA (3 groups), multifactor (two way) ANOVA, repeated measures ANOVA (RM-ANOVA) within subjects, comparing, F statistic |
|
comparing (ANOVA) |
between group variability, within group variability |
|
with ANOVA results tells researcher that mean scores between at leats two groups are |
differnet but does not specify which groups |
|
with ANOVA you can reject null but further testing is needed to determine |
where differences lie |
|
post hoc tests |
turkeys HSD - which groups differ from one another |
|
chi squared test is nonpara or parametric |
nonparametric |
|
chi squared test the difference in |
proportions in categories within a contingency table |
|
chi squared compares observed |
frequencies in each cell with expected frequencies (expected if no relationship) |
|
chi squared test 2 data comparisons |
nominal or categorical |
|
chi squared tes stat |
X2 (squared) statistic |
|
multivariate statistics |
statistical procedures for analyzing relationships among 3 or more variables |
|
two common multivariate stats in nursing research |
multiple regression, analysis of covariance (ANCOVA) |
|
multiple linear regression used to |
predict |
|
multiple linear regression dependent variable is |
continuous |
|
predictor variables are (2) |
continuous or dichotomous |
|
correlation index for dependent and predictor variables is |
R |
|
regression |
strong statistical analysis |
|
regression is _____ not ______ |
predictive not causative |
|
logistic regression analyzes |
relationships between nominal level dependent variable and 2 or more independent variables |
|
logistic regression yields an |
odds ratio |
|
analysis of covariance (ANCOVA) extends ANOVA by |
removing effect of confounding variables (covariates) before testing whether mean group differences are significant |
|
multivariate analysis of covariance |
MANCOVA |
|
MANCOVA |
extension of ANOVA to more than one dependent variable |
|
factor analysis used to |
reduce large set of variable into smaller underlying dimensions |
|
factor analysis used primarily in |
developing scales and complex instruments |
|
causal modeling tests a |
hypothesized multivariable causal explanation of a phenomenon |
|
causal modeling includes (2) |
path analysis, structural equations modeling |
|
SPSS |
most often used in nursing research, stats for bio sciences |
|
SAS |
more purely mathematical |
|
most common reported stats (4) |
descriptive stats about sample and variables, analysis of group equivalency, stats about role of error, stats to evaluate magnitude of effect, stats to determine confidence level |