Analyzing Quantitative Data

  1. Independent variable: the manipulated "causes" in an experiment (what you change); controlled by the researcher

  2. Dependent variable: the measured "effects" (what you observe); responds to those changes, acting as the outcome variable

  • Ex. A scientist wants to see if the temperature of a room affects how fast an ice cube melts.


ratio: Numeric scale with equal intervals and true zero (e.g., weight, age, BP)

interval: Numeric scale with equal intervals but no true zero (e.g., Fahrenheit temperature)

nominal: no order (gender, religion)

ordinal: order/rank (Likert scale)

high levels of measurement are more desirable due to more option instead of restriction to one or more categories


Testing Validity: if its measuring the right thing ( Parametric test are hypothesis testing)

Null hypothesis: States NO relationship or difference exists

  • can be rejected using a statistical test based on the probability that it is false; if rejected then the researcher can say the hypothesis is true aka alt.

Alt. Hypothesis: States relationship or difference exists

  • healthcare research favors the alternative hypothesis aka rejecting the null

descriptive stats: SUMMARIZES DATA ; summarizing the data you already have — no guessing, no predicting. “What does my dataset look like?”

  • measures of central tendency: has to be I/R

    • MEAN: MOST affected by OUTLIER

      • I/R

      • chosen for LOM, obj

    • MEDIAN: MOST affected by SKEWNESS (chosen for shape)

      • I/R/O

    • MODE

      • N/I/R/O

  • measures of dispersion:

    • RANGE: simplest, unstable and sensitive to outliers; not reliable

      • N/I/R/O

    • VARIANCE & STD: based on mean - deviated score (STD-square root of average)

      • takes EVERY score into account

      • R/I

      • outlier sensitive

        1 STD=68% ; 2 STD=95% ; low STD= data point are very close to the mean

  • graphical representation: Charts/graphs

inferential stats: TESTS HYPOTHESIS; Generalized predictions and drawing conclusions about a population based on sample data

  • t-test, NOVA test, chi-square tests

  • Hypothesis testing

    • Parametric test: ex. hypo testing; confidence interval, t-test, NOVA test

      • DV must be normally distributed

      • require I/R data

      • test for normality

        • is there a correlation relation between 2 groups mean

        • is there significant difference (t-test)between 2 groups mean DV

        • do x, y, and z predict regression mean scores of a

    • non-Parametric test: chi-square tests, logistical regression

      • ordinal or nominal level

      • uses medians and rank

  • p-values: Probability that results occurred by chance

    • p < 0.05 → statistically significant (Your results are unlikely to be due to chance alone) → reject null hypothesis

    • alpha level: the cutoff you choose before the study “How much type 1 error am I willing to accept?”

      • p ≤ α → statistically significant

  • Confidence intervals

Normal curve:

skewed left (bc tail is to the left): if mean is lower than median (mode is alway at tip so highest pt)

Skewed right: Numerical mean is higher

Bimodal distribution: possible to have NO modes

type 1 error: false positive you say there is a relationship but there is none

type 2 error: false negative no relationship but there is one

Parametric Tests of Relationship

Correlation: must be I/R and normally distributed

  • positive or negative

  • coefficient correlation: (r ) closer to 1 is stronger

Multiple Regression: examines how two or more independent variables predict a single dependent variable.

  • (Multiple IVs impact on 1 DV)—How well do these variables together predict an outcome?” “Which variable is the strongest predictor?”

  • Ex. Predicting blood pressure (DV) using:

    • Age (IV)

    • Weight (IV)

    • Sodium intake (IV)

    👉 All 3 IVs together predict the DV

  1. must be normally distributed and have AT LEAST 10 subjects/participants per variable

  2. homogeneity of variance


Types of inferential test: Parametric Tests of Difference


t-test: determine if they are significantly different from each other

  1. Independent t-test: Compares means of two independent groups

  2. Paired t-test: Compares means of same group at two different times


ANOVA: analyzes the differences among means of three or more groups in a sample

  1. One-way ANOVA: Compares means of three or more groups

  2. Factorial ANOVA: Examines multiple independent variables and their interaction

  3. Repeated measures ANOVA: Compares same group over multiple time points


Non‐Parametric Tests:

chi-squared: Tests relationship between categorical variables (checks if two categories are related)

Logistic Regression: Predicts categorical outcomes (e.g., yes/no)

interpreting and reporting research findings: Results should be reported objectively

  • examining the meaning of results

  • considering significance of findings

  • drawing conclusions

  • suggest implications for practice for further study


statistical significance: findings are unlikely due to chance

clinical findings: findings that have meaning for patient i nthe presence or absences of statistical significance


generalizability: to extent to which findings can be applied beyond ginev research situations to other settings

external validity: How well the results can be applied to other populations or real-world settings

  • Example: Results from a study can be applied to all patients, not just the small sample


internal validity: How well the study shows that the independent variable actually caused the change in the dependent variable

  • Example: A study shows a drug lowers blood pressure because of the drug, not other factors


random sampling is more likely to produce representative samples

convenience sampling limits generalizability

Selecting a Qualitative Design

phenomenology: understanding the lived experience and identifying common themes (LIVED EXPERINCE)

  • ONLY BY individual who experienced it

  • GOAL

    • enhance understanding on what it means to be human

    • deeper understanding of everyday life

    • studies everyday life (what is the meaning for individual?)

    • characteristics: rich, full, insightful descriptions

    • answer: how we view ourselves, others, environment

  • Research Methods: rigorous, critical, systemic investigation

    • descriptive, retrospective, in-depth analyses

  • Steps:

  1. ID lived experience and create retrospective reflection

  2. Bracketing before data collection: must set a side personal feelings about topic

  3. purposive sampling— keep adding to experience until description become repetitive (no new/different/themes emerging)

  4. synthesis of large amt of narrative data

Ethical considerations:

  • informed consent

  • IRB approval

confirmability: Findings are shaped by the participants, not researcher bias

  1. auditability: Another researcher can follow the decision-making process of the study

  2. credibility: participants confirm findings are accurate and faithful

  3. fittingness: How well findings apply or “fit” in other contexts


Ethnography: cultural context of values, beliefs, practices of cultural groups in natural settings (CULTURE)

  • gaining meaning for social encounters

  • focused on patient observation:

    • Emic: How participants themselves understand their culture/experience; Focuses on their beliefs, meanings, language

      • “From THEIR perspective”

        • EX. A nurse describing what working in the ICU feels like in their own words

    • Etic: The outsider’s view (researcher’s perspective); Researcher interprets behaviors using their own framework; More objective/analytical

      • “From the RESEARCHER’S perspective”

        • EX. A researcher analyzing ICU nurse stress using stress theory

  • Data collection methods:

    • pre field work, field work, post field work

    • census taking: basic demographic information

    • mapping: identifying location of people

    • document analysis: examining records

    • life histories

    • event analysis reviewing photo documentation


Grounded theory: based in sociology and physiology—symbolic interaction (THEORY)

  • processes and core concept questions

  • experiences that change over time, in stages/phases

  • GOAL:

    • generate new theory

    • no attempt to test theory

    • AVD pre-concived notions

  • data collection:

    • participant observation, formal and informal interviewing (field notes)

    • theoretical sampling: process of data collection for generating therories




Samplings: in QL data we disregard generalizability

Purposive Sampling: Researchers intentionally select specific participants who meet certain criteria

  • You want a very specific group

  • Common in qualitative research

    • EX. Selecting only ICU nurses with 5+ years experience

Convenience Sampling: Selecting participants who are easiest to access

  • High bias

  • Not very representative

    • Ex. Surveying students in your own class

Probability Sampling: Every member of the population has an equal chance of being selected

  • Reduces bias

  • Increases generalizability

    • Ex. Randomly selecting patients from a hospital database