HEDU 4300 final exam

step 1: identify problem; step 2: research the area, primary and secondary sources; step 3: identify hypothesis/research question, form question based on what learned in step 2; step 4 research design; step 5 collect data; step 6 analyze data; step 7 findings and conclusions

Descriptive (current conditions), correlation (relationship), Experimental (differences, effects)

continuous, categorical

entire group of people of interest from which sample will be selected

a subgroup of the target population that will hopefully participate in our research

sample should be similar to the population on the characteristics of interest (think back to research question

The results can generalized to the population your studying

random sampling is when all participants have an equal chance to be selected

simple random sampling, stratified random sampling, cluster sampling, systematic sampling, convenience sampling

uses random selection, every person has equal chance, selection of one person doesn’t interfere with selection of another (put names back in hat after pick)

divide the population into specific subgroups, or strata, based on a critical characteristic. A set number of research participants are randomly selected from each strata. (can do this proportionally or equally)

sampling unit is a naturally occurring group or clusters of members of the population, instead of sampling individual members you randomly select the clusters

sample is drawn by choosing every kth person from a listing of the population (make sure the list is made up randomly)

include any participants who are readily available and accessible (not random)

Validity: when questionnaire is measuring what is intended to be measured; Reliability: questionnaire measures consistently each time it is used

• content-related: experts in field review questionnaire
• construct related: use stats approach to be sure I am measuring the construct of (whatever) the whole construct and nothing but the construct
• criterion-related: uses two questionnaires. take questionnaire and use stats approach to determine if it correlates with certain criterion (something that theoretically, resilience should be able to predict – like optimism) the take questionnaire and determine using a stats approach if it can distinguish between resilience and something very different (like defeatism)

• -  alpha reliability: using stats approach to examine each item on the questionnaire compared with the other items on the questionnaire (most common one)
• -  test-retest reliability: use statistical approach to examine day-to-day stability of the questionnaire by correlating day 1 scores with day 2 scores (time lapse on this one students take one day and then another day and scores should correlate)

values, quantities calculated using info obtained from a sample

involves individual variables (meaning one) descriptive statistics are this kind of stats (used to describe and summarize the characteristics of the data from our sample

data collected at one point in time

all three types descriptive, correlation, experimental

can be used with either type of data (continuous, categorical) its an ordered listing of the scores and their frequencies

indicates those points at which scores tend to be concentrated

mean (average), median (middle score), mode (score most frequently received)

• Range (difference between highest and lowest score),
• Standard deviation (indicates amount of variability among the scores in relation to the mean and how much each score deviates from the mean);
• quartile deviation (indicates the amount of deviation surrounding the median)

? if you collected continuous level data from everyone in your target population and created a graph, it would approach a bell curve

Bell curve

degree of peakedness

leptokurtic (majority of scores in the middle); platykurtic (scores spread out smaller cluster in the middle

many scores on one end of the scale and progressively fewer on the other end

negative (few low scores), positive (few high scores)

frequency distribution, visually display

frequency, measures of central tendency and variability

used to make inferences from a sample to a population

correlation and experimental

a tentative explanation, a prediction or the outcome, an educated guess. Based on research question

correlation and experimental

• 1) state null and alternative hypotheses
• 2) determine significance level – Alpha (.05) 3) choose appropriate test statistic and apply it to the data from your sample
• 4) calculate p-value (computer does)
• 5) compare p-value to alpha level (reject or accept null hypotheses

means the results you got are probably true (not due to chance); that the results are good enough to be believed

I’m willing to accept 5% probability that my results happened by chance. which means I want to be 95% confident that my results are statistically significant

The probability – based on the data I am actually analyzing – that my results have occurred by chance

p-value<.05 reject null: there is a significant relationship of significant difference

; p-value>.05 accept null: there is no significant relationship or no significant difference

? type I error= reject the null when it’s actually true (I see a difference between the gourps in my sample when that difference doesn’t really exist in the population);

accept the null when it’s actually false (I don’t see a difference between the groups in my sample, but that difference actually exists in the population

Correlation – calculates r value and p value

continuous

? null= there is no significant relationship

alternative= there is a significant relationship

scattergram

positive: scores go in same direction. as one score increases so does the other or as one score decreases so does the other

negative: scores go in different direction as one score increases the other decreases

no relationship: no pattern

straight line between your data points; created by minimizing the overall distance between the points and line. shows general direction that a group of data points seems to be heading

tells you the direction and strength

? Null: there is no significant difference

alternative: there is a significant difference

nondirectional= there is a difference; directional= there is a difference in a particular direction

Independent groups t-test, ANOVA, post hoc tests; chi square, paired samples t-test

compare 2 independent groups

IV categorical

DV continuous

groups based on existing characteristics (male female); groups based on something I control (teaching method, drug treatment)

compare 3+ independent groups

? DV continuous; IV categorical

tells you there is a statistically significant difference between the three mean scores

doesn’t tell you where the difference lies

Post – hoc test

compare 2 dependent groups

? IV categorical; DV continuous

think pretest postest

when you have categorical data. used to determine whether significant differences exist in the number of people who are classified in the categories of the research variables

Null: there will be no difference between expected frequencies and observed frequencies

Alternative: there will be a difference between expected frequencies and observed frequencies

categorical

morals, norms for conduct that distinguish between acceptable and unacceptable behavior

promotes aim of research, promote values essential to collaboration, can be held accountable to the public, promote other moral and social values

? examines the risks to benefits of proposed reearch studies with respect to protecting human subjects and abiding by federal regulations

• anonymous: inability to link any research information back to the individual respondents;
• confidential: ability of researchers to link research information back to the individual respondents but no one else can access the information

voluntary no coercion, fully explained expectations described, clear description of risks and benefits, statement indicating anonymous or level of confidentiality indicates who will see the individual results of the participants, participants right to ask any questions any time receive contact info of principal investigator and IRB, statement that participants can withdraw at any time with no consequences

falsification, fabrication, plagiarism

• p < .05 you reject the null – assume variances were not equal,
• p > .05 you accept the null assume variances were equal

• Use Tukey if equal variances assumed,
• use Dunnett if equal variances not assumed (use Levene’s test to find out the variance)

If p < .05 reject the null and report that there is a statistically significant difference b/w expected frequencies and observed frequencies ( with this test you hope that the numbers are proportional – which doesn’t’ mean equal numbers at beginning it means proportionate, this is what is meant by expected frequencies)