HAN 251: Research Methodology Final Material 2016

Features of Qualitative Research

- data can be difficult to quantify or measure -power of data

Mixed Method Research

- gain deeper understanding of problem by using both QUAN and QUAL methods -determine rationale: priority for each data type - data types collected and different times? - informs research methods as a blueprint -sampling -data collection methodologies -analysis methodologies -type of conclusion

Qualitative research strategies

- interactive interviewing , verbal description of experiences -focus groups and individual interviews, collect artifacts, write descriptions -observations nonreactive- observer removed from participants participant- observer as member of the group

Qualitative Research

-attempt to verify, generate a theory from the data -broad study --begin with vague research question - words, text, video, photo, recordings - data collection involves observation, interviews, small groups - insiders pov

Quantitative Design

-begin with theory and specific hypotheses - shows statistical evidence, independent and dependent -tangible and easy to measure-involve analysis/numerical data -large group study outsider pov

Disavdvantages of a cohort study

-cohorts can be difficult to identify due to variables -imbalance in patients can exist -blinding/masking is difficult -could take time ex Framigham Heart Study MA

population type of samples

-contain definite strata equal size (only grade 3 of the school) -subgroups in different proportions (25% catholic etc) -discrete clusters (20 top colleges)

Historical Research

-deals with the meaning of the events -build a framework of interpretation -focused on fixed period of time

Qual nonexpiramental

-does not manipulate variables - no casual relationships -discover relationship -usually one group sampling methods -- numerous methods

Disadvantaged of longitude drsign

-ezpensive, require team of researchers , grant . Follow up is key -long time before result obtained can't test hypothesis quickly

content analysis

-identify research question, examine data in relation to framework identify the body of material to be studied/subgroup. Qualities. Identifies themes

Non experimental design

-no variable manipulation -discover new relationship -one group Numer o us methods

Observational

-provide QUAN as well as QUAL through grounded theory and ethnographic. - theoretical framework- -operational define- what us being observed I. Order to count or evaluate observations - data collection divided into small segments of time. -codes are predetermined -more than one independent raters, in depth training

Case Control Disadvantages

-retrospective study more problems rely on memory -finidng suitable control group

Case Control Advantages

-study rare conditons -less time needed to conduct the study -look at multiple factors at the same time -useful intitial study

Advantages of a Cohort Study

-subjects can be matched -limits influence -easier and cheaper than a randomized controlled trial (RTC)

Qualitative Research

-utilized in anthropology, history, psychology and education -increased use of phenomenological and other QUAL approaches

Qual experimental

-variables are manipulated by the investigator -seek to identify a a cause effect relationship involve -random assignment -control group sampling method -methods only most rigorous, internal validity

Experimental design

-variables manipulated -cause and effect -involve random assignment and control group -sampling method

Drug research

.001 severe

3 Types of Mixed Method designs

1. Convergent (Parallel/concurrent) 2.Explanatory 3.Exploratory

3 Features of Qualitative Research

1. Naturalistic Settings- more time in observing and interviewing subjects in their own environment. data collection and analysis vital to study 2. Local Groundedness- researchers will collect proximity data before the study is under way 3.Attempt to collect meaning of events, process and structures in subject live. getting perception and attitude

Which design is the best ?

1. simple answer- research design based on paradigm and your research purpose and your research question

grounded theory

An approach for developing theory that is "grounded in data systematically gathered and analyzed. Begin with data and let theory develop from data. code -grounded in experience and views of participants 20-50 ppl -grounded theory research involves multiple iterations of analysis - open coding, fragmented and analyzed - axial coding, patterns, interconnected emerge -selective coding, overall pic of the data -develop theory/model for understanding

I believe abortion is morally wrong

Correctly written

Purpose of study is to find strength of group

Correlational

Purpose is to see if participants change I've rtime in mid schoolers develop and change over time survey 1500 grade 6-8 NYU school

Cross sectional design.

Difference between qual and observational

Data analysis Data collection Paradigm

QUAL Data Analysis

Data analysis includes - 29 seperate methods to code data -human instruments not statistical test -field notes bracketing, diaries -details, word for word transcription read/reflective process Categorization of data - segment data into meaningful units Interpretation of data identification - saturation , generalizations

Data Analysis for correlational study

Descriptive Stats allows researcher to describe sample organize and summarize the data. -----frequencies, percentages description mean median mode Inferential Stats allow inferences and correlations to be mas=de from smaple to population. Parametric data (interval and ratio) t test ANOVA

Cross sectional design

Different age groups sampled. Data collected once and compared Ex how 70s 80s 90s do in academics

Two questions in one I believe abortion morally wrong and death penalty is great

Double Barrel questions

Explore relationship between attraction and characteristic 500 students (quan) of strangers 18-21 small (qual) group of students continued tye study by doing qual Explain key findings

Explanatory (continued the study)

Summarize and paraphrase. Theme turned to variables.

Exploratory

Do you go to the gym and then a follow up question after answering yes

Filter question

Survey question uses cumaltiive ranked scale where response statement ranked extreme to mist extreme Question scale that gets intense. Cumaltiive

Gutmann

5 types of Qualitative Designs

I Ethnography II Grounded theory III Phenomenological IV Case Study V Content Anlaysis

3 types of quantitative Designs

I Experimental II Quasi Experimental III Non Experimental/descriptive

Declarative statement agree disagree etc. Scale

Likert

Capture differences in incident of bullying. Follows group of 13 yrolds predetermined questions interview ask questions about texting behavior

Longitudinal design study

Educational intervention

Not severe so alpha .05

Grounded theory and ethnography

Observation nut no predetermined what is performed. Selective vidimg, Dara reveals codes and meaning

Cross sectional design example

Pilot study determine influenza would be in population by overview g rates Merhods: sample (sqqbs) from ppl for acute respiratory.

Observational research approaches

Rede termini observations, code book theory and hypothesis Use number raring and Stat analysis

Longitudinal design

Single group followed over time. Many data collection points. Simultaneous collect information - quick economical approach has no control/comparison . Associations affect variables bias . No casual inferences

Developmental

Study of characteristics variables as they develop or change over time

Methods to Validate

Triangulation -the purpose is to confirm info about a phenomenon/ obtain convergent validity 2 methods percentage of agreement - another coder independent then group discussion

Correlational

Use inferential stats

Descriptive research

What us going on Developmental Observational- cohort and case control Correlation al Survey

random sampling

any differences between the group are small . Equal chances for everyone in treatment and control group

probability sampling

attempts to specify a random sampling from larger population. Rep a whole -simple random -stratified -proportional stratified -systematic -cluster

Cohort Study vs case control

begin with a group of ppl -grouped into two categories -followed over tim to see differences case control begin with selected case , controls represent ppl ex. suspect that there is relationship between certian water soureces and diarrea among 5yr olds

Qualitative research questions

begin with statements like how, what , why

Quasi Experimental Designs

can be similar to experimental design with exception to random assignments -include control groups, no cause and effect relationship -nonequivilant -simple time series -control group time series -multiple baseline

Practical Significance

consider - quality of research question -relative size or effect -size of sample -importance of finding -link to previous research -strength of correlation -importance intervals

correlational vs experimental

correlational does not influence any new variable, no casualty proven experimental manipulaes variables measure effects data anlysis calculate correlations

internal validity

degree to which extraneous variables have been controlled, experimental effects can be attributed to treatment and intervention. -random assignment -by Canbell and Stanley O- observation/measurement R- random assignment X- manipulation/treatment/intervention

Case Control Studies

designed to estimate the odds of disease and conditions. determines if there is an association relationship between conditions and risk factors -observational no interventions retrospective with 2 groups example- is zinc oxide a good way to treat skin cancer by comparing workers with skin cancer . Ask them to recall what sunscreen they used

QUAN seeks statistical validity

how many ppl do i need to generalize my results from the sample of population. Power analysis to determine number of subjects and participants

snowball

identify potential subjects in studies where subjects are hard to locate

case study design

in depth look at a particular program, event or individual (rare disease), studied for prescribed period of time to understand, inform practice. data collection is extensive, include strategies, data sets

ethnography

in depth look at an entire group to describe culture/aspect of culture , overtime a detailed portrait of the group's shared culture, behaviors and languages subject pov understand behavior

Phenomenological

lived experience -multiple perceptions and perspectives of the remaining of the event -what is it like to have experience -interviews 5-25 ppl no structure -suspend preconceived notions or personal info

correlational

looks at extent and degree of the variables of diifferent groups or more exists if one or more variable increase or decrease -easy to get preliminary study

How to select a Research design?

must know the strengths and limitations of each methodology and corresponding design - select the design that best aligns with the research purpose and research question. Best approach to answer the question -how will the data types be combined or integrated and which lens will be used

QUAL seeks saturation

no exact number, how many ppl talked too, enough observations etc

non probability sampling

no guarantee that each element of population will be included in the sample -convenience -quota -purposive -snowball

pre experiments

non randomization, no equivalent control group, no cause or effect relationship -one group -poor internal validity -results influenced by multiple factors -weakest design/avoid

within subject design

one group receives 2 treatments at same time/ useful when localized and temporary strengths- only need one group and help reduce errors between subjects

Experimental design types

pretest/posttest control group- controls for re test effect, solomon 4 group- post test only group- for threats in internal validity

Sampling

process of selecting suitable samples or rep for each group. Finding the right sample, sample size aids researcher in finding outcome. wrong sample size means skewed data, altered misleading

Pearson r linear or product moment correlation

range from-1 to +1 -1 means negative correlation +1 mean positive correlation strong positive correlation is r=.90 r=-.90 is strong neg correlation

convenience

readily available participants, selection easy, class

What does the research question begin with and why

research question! - nee to reflect the methodology, data collection and data analysis

Statistical Significance

results real- was x better than y or is it chance allow to rejct the null hypothesis , chi square pearson t tets researcher sets a predetermine alpha level p<.5 5/100 occurences p<.01 result of chance p<.001 1/1000 result of chance

Nonequivilant pretest posttest design

same as experimental design without randomization

purposive

select participants for a particular purpose. knowledge of group is required before selection. Bias

quota

selection based on same proportions found in the general population given two ranges equal number taken, Stratified but not random

Multiple Baseline design

show effect of treatment by initiating at different times for different groups/individuals

Clinical Significance

similar to practical significance treated individuals as grop indistinguishable from non treated with respect to primary complaints following treatment? -practical applied effect of intervention

Time Series

simple time series- several measurements of the dependent variable over time. Baseline- measurements prior to the interventions control group time series- yields greater internal validity, parallel series of observations

power analysis

statistical analysis test to help determine number of participants needed to answer the question

limitation of within subject

taking part in one condition can impact performance or behavior on conditions -fatigue bored treatment tests -skew results -Two interventions

Standard deviation

tells gow far away teh values are from the mean norm distribution- 68% observations should fall within the SD smaller SD= close to mean

code books

think operational deinitions applied to QUAL

smaller value for alpha

type 1 error is serios or large sample size

Quantitative research questions

what is the relationship to what extent what is the difference impact