Research Flashcards
True Experimental Design - quantitative
Classic 2-group design which includes random selection & assign into group that receives tx or control. Cause & effect relationship of independent/dependent variable is examined. In human subject - often diff to design pure experimental design
Quasi Experimental Design- quantitative
I variable is manipulated to determine its effect on dependent variable but lesser degree of researcher control and/or no randomization. Often used in HC research where its unethical to control or w/hold tx. Used to study intact groups created by events or natural processes.
Non-experimental/Correlational Design - quantitative
No manipulation of I variable, randomization/researcher control are not possible. Used to study potential relations between 2(+) existing variables. Describes relationships, predicts relationships among variables w/o active manipulation. Limits: cannot est cause/effect, may fail to consider all variables. Degree of relationships is expressed by correlation coefficent: -1.0 and +1.0. Examples: retrospective (past data), prospective (present data), descriptive (several variables at once), predicitve. *Can be ex post facto
Qualitative Methodology
Descriptive research that studies ppl I-ly/collectively in natural social/cultural environments. systematic/subjective.
Phenomenological - Qual
study of 1(+) persons and how they make sense of their experience. Min interp from investigators & meanings are only ascribed by participants
Ethnographic - Qual
patterns/characteristics of cultural group: extensive field observation, interviews, participant observation, literature examination & cultural immersion. used in HC to understand insider prospective (interviews always face to face)
Heuristic - Qual
complete involvement of research in experience of subject to understand and interp phenomenon. Aim to understand human experience/meaning. Meanings can only be understood if personally experienced.
Case Study - Qual
Sing subject/group subjects investigated in-depth: easy to use in most practice settings
Trustworthiness critique of Qual reaserch
Credibility, trasnferability, dependability & confirmability
Data Collection
Observation, interview, written questionnaires, survey instruments
Surveys
Non-experimental instruments: can be open or close ended
Measures of central tendency
Determination of ave/typ scores: mean, median, mode
Measures of variability
Determination of spread of a group of scores: range, standard deviation (variability of scores from mean- appropriate w interval/ratio data), norm distribution (bell curve indicating distribution of scores) & percentile/ quartiles
Analysis and interp of data using inferential stats
Determines how likely results can be generalized to whole pop.
Standard Error of Measurement
est of expected errors in individuals score
Tests of Sig
Estimation of true differences, not d/t chance - rejection of null hypothesis. Alpha level: pre-selected level of stat sig- usually .05 or .01 (p-value) indicates expected diff is d/t chance; Deg of Freedom: based on # of sub/groups; Errors:
Types of Errors
Standard: expected chance variation among means, result of sampling error. Type 1: null hypo rejected when its true. Type 2: null hypo not rejected when false
Parametric Stats
Testing based on pop parameters (ratio/interval data), T-test (test of sig comparing 2 group means/ID diff at selected probability level), ANOVA (ana of variance: compares 2(+) tx groups/conditions at selected prob level) & ANCOVA (ana of co-variance: used to compare 2(+) groups/conditions while also controlling for effects of intervening variables
Nonparametric Stats
Testing not based on pop parameters (ordinal/nominal data); less powerful than parametric & more diff to reject null hypo. Chi Square: test of sig to compare data in form of frequency counts occurring in 2(+) mutually exclusive categories
Correlation Stats
Used to determine relations between 2 variables. Pearsons prod-moment coefficent (r): corr interval/ratio data. Spearmans rank (rs): nonparametric for ordinal. Intraclass cor co (ICC): reliability based on ana of variance. Common variance: rep of degree that variation in 1 variable is attributable to another
Strengths of Relations
Pos correlations: range from 0-+1.0 & indicates as X increases so does Y. High: .7-1, Mod: .35-.69 & Low: 0-.34. 0 means no relations. Neg correlations range from -1.0-0 & indicates as X increase, Y decreases - inverse relations.
Ethical Considerations for Studies
Participants must be provided w full disclosure of purpose, methodology, nature and scope; informed on any potential risk/discomfort w plan to remediate; vol participation w w/drawal at any time & refusal to answer Q’s honored; confidentiality ensured; IRB for human subs - needed for grants, proposals submitted prior to implementation
Entry Level OT Supervision
Supervision: Not req. Close S by an intermed/advanced us suggested. Supervises: aids, techs, all level OTAs, volunteers & L I FW
Intermediate Level OT Supervision
Supervision: Not req; routine/gen by advanced recommended. Supervises: aids, techs, all level OTAs, L I & II FW, entry OTs
Advanced Level OT Supervision
Supervision: Not req, min by advanced suggested. Supervises: ids, techs, all level OTAs, L I & II FW, entry/intermed OTs
Entry Level OTA Supervision
Supervision: close S by all level OT or intermed/advanced OTA. Supervises: Aids, techs, voluenteers
Intermediate Level OTA Supervision
Supervision: routine/gen by all levels of OT or advanced OTA. Supervises: aids, techs, entry level OTA, voluenteers, L I OT FW & L I&II OTA FW
Advanced Level OTA Supervision
Supervision: gen by all levels of OT or advanced OT. Aids, techs, entry/interm OTAs, voluenteers, L I OT FW & L I&II OTA FW
Nominal or Categorical Scale
Lowest level scale used to classify data which consists of observation/traits that can be placed into diff categories; can be grouped by numbers (1-M, 2-F) but NO numeric value. Common types: gender, race, blood type
Ordinal Scale
Include diff categories but they DO have numeric order/sequence but cannot be assumed that diff between each interval is equal. Ex: likert scale
Interval Scale
Can be assumed diff between each interval are equal & have no absolute 0. Ex: test scores, weight, height, but mostly temp
Ratio Scale
Highest level of scales which refers to none or complete absence of variable of interest.
Discrete vs Continuous Data
D: separate indivisible categories: whole # only w no values existing between 2 neighboring (# of ppl, income). C: Divisible into an infinite # of fractional parts (time, height, weight, GPA)
